Soft Inflatable Robotic Systems for Space Applications: A Survey

---

Full Text

Soft Inatable Robotic Systems for Space Applications: 1

A Survey 2

3

Abstract 4

Soft inatable robotic systems and structures are emerging as transformative tech- 5

nologies for space applications, oering compelling advantages in mass eciency, com- 6

pact stowage, compliance, and adaptability over traditional rigid-body systems. This 7

survey provides a comprehensive review of the intersection of soft robotics, inatable 8

structures, and space engineering, organised around a unifying thesis: the same high- 9

strength fabric technologies (Vectran, Kevlar, Nextel) that enable inatable habitats 10

also enable compliant debris capture mechanisms and large deployable shields. We ex- 11

amine two primary application domainsactive debris removal, where soft compliant 12

systems address the fragmentation paradox inherent in rigid capture, and space explo- 13

ration, where inatable habitats oer order-of-magnitude mass eciency improvements 14

over metallic modules. Eight enabling technology areas are reviewed: materials and 15

structures, deployment mechanics, actuation, sensing and structural health monitoring, 16

power systems, thermal management, attitude and orbit control, and robotic in-orbit 17

assembly. We identify ve critical research gaps, including the absence of quantitative 18

soft-versus-rigid fragmentation comparisons, the lack of ight heritage for soft robotic 19

capture, and the unexplored rigid-to-exible assembly interface. A research roadmap 20

spanning 5-year and 15-year horizons is proposed, with the most ight-ready near-term 21

demonstrator identied as a gecko-adhesive gripper on an inatable arm with bre 22

Bragg grating structural health monitoring. This survey dierentiates itself from prior 23

reviews in Progress in Aerospace Sciences by focusing specically on soft and inatable 24

systemsa technology class not covered by existing reviews of rigid space robotics or 25

contact/contactless debris removal. 26

Contents 27

1 Introduction 4 28

2 The Case for Soft Inatables in Space 8 29

2.1 Space Debris Crisis and the Need for Active Removal . . . . . . . . . . . . . 8 30

2.2 Human Exploration Beyond LEO: The Habitat Challenge . . . . . . . . . . . 11 31

2.3 Unifying Thesis: Shared Fabric Technology Across Applications . . . . . . . 12 32

3 Use Cases: Active Debris Removal 15 33

3.1 Rigid Capture Approaches and Fragmentation Risk . . . . . . . . . . . . . . 15 34

3.1.1 The Fragmentation Paradox . . . . . . . . . . . . . . . . . . . . . . . 16 35

3.2 Soft and Compliant Capture Mechanisms . . . . . . . . . . . . . . . . . . . . 17 36

3.2.1 Gecko-Inspired Dry Adhesive Grippers . . . . . . . . . . . . . . . . . 17 37

3.2.2 Dielectric Elastomer Minimum Energy Structure (DEMES) Grippers 19 38

3.2.3 Bistable and Passive Capture Grippers . . . . . . . . . . . . . . . . . 19 39

3.2.4 Thermally Qualied Soft Grippers . . . . . . . . . . . . . . . . . . . . 20 40

3.2.5 Inatable Robotic Arms for Capture . . . . . . . . . . . . . . . . . . 20 41

3.2.6 INSIDeR: Net Capture with Inatable Deployment . . . . . . . . . . 20 42

3.3 Inatable Debris Shields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 43

4 Use Cases: Habitats and Exploration 23 44

4.1 Heritage Timeline: Echo to BEAM . . . . . . . . . . . . . . . . . . . . . . . 24 45

4.1.1 Early Inatables: Echo and Volga (19601965) . . . . . . . . . . . . . 26 46

4.1.2 TransHab: Proving the Five-Layer Architecture (19972000) . . . . . 26 47

4.1.3 Genesis and BEAM: Orbital Validation (20062016+) . . . . . . . . . 27 48

4.2 Current Commercial Programs: LIFE, Orbital Reef, and Beyond . . . . . . . 27 49

4.2.1 Sierra Space LIFE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 50

4.2.2 Historical Context: B330 and Commercial Ecosystem Fragility . . . . 28 51

4.2.3 NextSTEP Competitive Landscape . . . . . . . . . . . . . . . . . . . 28 52

4.3 Future Concepts: Lunar Surface, Mars Transit, Planetary Entry . . . . . . . 28 53

4.3.1 Lunar Surface Habitats . . . . . . . . . . . . . . . . . . . . . . . . . . 28 54

4.3.2 Mars Transit and Surface Applications . . . . . . . . . . . . . . . . . 29 55

4.3.3 European Programmes . . . . . . . . . . . . . . . . . . . . . . . . . . 29 56

4.4 Radiation Shielding: The BEAM SPE Findings and Design Implications . . 30 57

5 State of the Art: Materials and Structures 31 58

5.1 Space-Rated Fabrics: Vectran, Kevlar, Zylon, Nextel . . . . . . . . . . . . . 31 59

5.2 Multi-Layer Shell Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 33 60

5.3 Rigidization Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 61

5.4 Environmental Degradation: AO, UV, Radiation, Creep . . . . . . . . . . . . 38 62

6 State of the Art: Deployment Mechanics 39 63

6.1 Fold Patterns and Packaging Eciency . . . . . . . . . . . . . . . . . . . . . 39 64

6.2 Ination Sequencing and Control . . . . . . . . . . . . . . . . . . . . . . . . 39 65

6.3 Flight Heritage: InateSail, LOFTID, BEAM Deployment Lessons . . . . . . 41 66

6.4 Comparison with Rigid Deployable Alternatives . . . . . . . . . . . . . . . . 42 67

7 State of the Art: Actuation for Soft Space Systems 43 68

7.1 Dielectric Elastomer Actuators and DEMES . . . . . . . . . . . . . . . . . . 43 69

7.2 Vacuum-Gap Electrostatic Actuators: Vacuum as Enabler . . . . . . . . . . 44 70

7.3 Ionic Electroactive Polymers: Space Tolerance Assessment . . . . . . . . . . 45 71

7.4 Tendon-Driven Continuum Manipulators . . . . . . . . . . . . . . . . . . . . 45 72

7.5 Shape Memory Alloys for Deployment . . . . . . . . . . . . . . . . . . . . . 46 73

7.6 Jamming in Vacuum: A Novel Opportunity . . . . . . . . . . . . . . . . . . 46 74

7.7 Sealed Pneumatic Actuation in Space . . . . . . . . . . . . . . . . . . . . . . 48 75

7.8 Electroadhesion and Magnetic Actuation: Emerging Approaches . . . . . . . 48 76

8 State of the Art: Sensing and Structural Health Monitoring 49 77

8.1 Fibre Bragg Grating Sensors: From Proba-2 to Inatable Webbing . . . . . . 49 78

8.2 Multicore Fibre Optic Shape Sensing . . . . . . . . . . . . . . . . . . . . . . 52 79

8.3 Capacitive, Resistive, and Alternative Soft Sensors . . . . . . . . . . . . . . . 52 80

8.4 Distributed Fibre Optic Sensing: Rayleigh and Brillouin Scattering . . . . . 53 81

8.5 Distributed Impact Detection . . . . . . . . . . . . . . . . . . . . . . . . . . 53 82

9 State of the Art: Power Systems for Large Inatables 54 83

9.1 Flexible Solar Array Landscape: ROSA to Perovskite . . . . . . . . . . . . . 54 84

9.2 The Inatable-Power Integration Gap: PowerSphere and Beyond . . . . . . . 56 85

9.3 Energy Storage: Li-ion, RFC, and Mission-Dependent Selection . . . . . . . 57 86

10 State of the Art: Thermal Management 58 87

10.1 Multi-Layer Insulation for Inatable Shells . . . . . . . . . . . . . . . . . . . 58 88

10.2 The JWST Sunshield as Deployable Thermal Barrier Precedent . . . . . . . 59 89

10.3 Variable Emissivity Coatings and Smart Radiators . . . . . . . . . . . . . . . 60 90

10.4 Loop Heat Pipes for Deployed Structures . . . . . . . . . . . . . . . . . . . . 61 91

10.5 Phase Change Materials in Fabric Layers: The TRL 23 Gap . . . . . . . . . 61 92

11 State of the Art: Attitude and Orbit Control 62 93

11.1 Control-Structure Interaction for Flexible Spacecraft . . . . . . . . . . . . . 63 94

11.2 Gyroelastic Body Theory and Distributed Momentum Management . . . . . 63 95

11.3 Drag Budget for 100 m-Class LEO Structures . . . . . . . . . . . . . . . . . 64 96

11.4 The Missing Theory: AOCS for Pressure-Stabilised Membranes . . . . . . . 66 97

12 State of the Art: Robotic In-Orbit Assembly 67 98

12.1 Assembly Robot Heritage and Current Programmes . . . . . . . . . . . . . . 68 99

12.2 Walking Robots for Large Structure Assembly: E-Walker . . . . . . . . . . . 68 100

12.3 The Rigid-to-Flexible Interface Gap . . . . . . . . . . . . . . . . . . . . . . . 69 101

12.4 Assembly-Enabled Inatable Platforms: Design Requirements . . . . . . . . 70 102

13 Challenges, Open Questions, and Research Roadmap 71 103

13.1 Critical Research Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 104

13.2 Integration Challenges at System Level . . . . . . . . . . . . . . . . . . . . . 74 105

13.3 Proposed Research Roadmap: 5-Year and 15-Year Horizons . . . . . . . . . . 75 106

13.4 The Path to Flight Demonstration . . . . . . . . . . . . . . . . . . . . . . . 79 107

14 Conclusions 80 108

1 Introduction 109

Two converging pressures threaten humanity's long-term access to and presence in space. 110

The rst is the accelerating degradation of the orbital environment: the low Earth orbit 111

(LEO) regime is increasingly populated with debris that endangers operational satellites, 112

whose services  from climate monitoring to navigation  underpin the global economy. 113

The second is the ambition for sustained human exploration beyond LEO, which demands 114

habitable volumes an order of magnitude larger than current metallic modules allow within 115

existing launch vehicle constraints. This survey argues that a single technology class  116

soft inatable robotic systems based on high-strength technical fabrics  oers a coherent 117

engineering response to both challenges through a shared material and structural foundation. 118

The orbital debris environment has reached a critical threshold. The European Space 119

Agency's 2025 Space Environment Report records approximately 44,870 tracked objects, 120

with an estimated 54,000 objects larger than 10 cm, some 1.2 million objects between 1 and 121

10 cm, and an estimated 140 million fragments between 1 mm and 1 cm, totalling roughly 122

15,800 tonnes of mass in orbit ESA Space Debris Oce [2025]. The consequence is oper- 123

ational: SpaceX's Starlink constellation executed 144,404 collision avoidance manoeuvres 124

in the rst half of 2025 alone, a 65-fold increase relative to 2021 ESA Space Debris Oce 125

[2025]. Kessler and Cour-Palais identied in 1978 that mutual collision among catalogued 126

objects could generate a self-sustaining fragment cascade Kessler and Cour-Palais [1978], 127

and Liou and Johnson subsequently demonstrated with the LEGEND simulation suite that 128

the current LEO population is already gravitationally unstable: even with a complete halt to 129

new launches, the debris environment continues to grow through inter-object collisions Liou 130

and Johnson [2006, 2008]. Stabilising LEO requires the active removal of at least ve large, 131

rocket-body-class objects per year from the most critical orbital shells Liou et al. [2010]. 132

Active Debris Removal (ADR) therefore transitions from a conceptual aspiration to an 133

operational necessity. Yet the dominant design paradigm  rigid robotic arms similar to 134

ClearSpace-1's four-arm capturing system  carries an ironic risk: forceful contact with a 135

tumbling, uncooperative object can fracture it, generating new fragments faster than they 136

are removed. Simulation studies and ground tests indicate that peak joint torques of order 137

195 Nm can arise during ENVISAT-class capture operations Ledkov and Aslanov [2022], 138

and the RemoveDebris harpoon demonstration saw a carbon-bre boom snap on contact at 139

20 m/s Aglietti et al. [2020]. The fragmentation paradox  rigid capture risks accelerating 140

the very cascade it aims to halt  provides the primary motivation for compliant, soft 141

capture architectures. 142

Simultaneously, the ambition to sustain human presence beyond LEO confronts a fun- 143

damental mass budget constraint. Metallic pressurised modules  Columbus (137 kg/m3) 144

and Tranquility (205 kg/m3)  are delivered at densities an order of magnitude higher than 145

fabric-based alternatives such as the TransHab concept (39 kg/m3) Valle et al. [2019a]. Vec- 146

tran high-tenacity yarn achieves a specic strength of 2,330 kN-m/kg, versus 220 kN-m/kg 147

for Ti-6Al-4V Valle et al. [2019a]  a 10× advantage that directly translates to launch mass 148

savings. The Bigelow Expandable Activity Module (BEAM), attached to the International 149

Space Station (ISS) since 2016, has accumulated more than eight years of continuous pres- 150

surised operation on the ISS, with periodic crew access for inspection and cargo storage, at 151

Technology Readiness Level (TRL) 9 NASA Johnson Space Center [2017]. 152

The organising thesis of this survey is that the same high-strength fabric technology 153

 Vectran restraint layers, Kevlar/Nextel debris shielding, Kapton thermal insulation  154

that enables BEAM's pressure vessel integrity also enables compliant robotic capture arms, 155

large deployable debris shields, and the next generation of deep-space habitats. Material 156

qualication campaigns, manufacturing processes, and design heritage are shared across 157

these application domains, providing an unusually coherent pathway from current ight- 158

proven technology to future operational systems. 159

Scope and Organisation 160

This survey reviews the intersection of three mature elds: soft robotics, inatable space 161

structures, and the enabling subsystem technologies (materials, power, thermal manage- 162

ment, attitude and orbit control, and robotic assembly) that together determine whether 163

soft inatable systems can be realised at mission-operational scale. The scope spans two 164

primary application domains: 165

1. Active Debris Removal  soft and compliant capture mechanisms (TRL 25) and 166

large inatable debris shields (design stage), examined against the rigid-capture base- 167

line. 168

2. Human Space Exploration  the heritage from Echo 1 (1960) through BEAM 169

(2016+) to current commercial programmes (Sierra Space LIFE, Orbital Reef), and 170

future concepts for lunar surface, Mars transit, and planetary entry decelerators. 171

Eight enabling technology areas are reviewed in depth: (1) materials and structures, 172

(2) deployment mechanics, (3) actuation, (4) sensing and structural health monitoring, 173

(5) power systems, (6) thermal management, (7) attitude and orbit control, and (8) robotic 174

in-orbit assembly. The survey concludes with a consolidated gap analysis and a research 175

roadmap spanning 5-year and 15-year horizons. 176

Relationship to Existing Reviews 177

Three prior surveys in Progress in Aerospace Sciences address adjacent territory, and this 178

survey is positioned explicitly as their complement (Table 1). Flores-Abad et al. reviewed the 179

state of space robotics for on-orbit servicing in 2014 Flores-Abad et al. [2014], establishing 180

the four-phase capture framework (approach, tracking, capture, post-capture stabilisation) 181

that remains the standard reference; however, that work predates the current wave of soft 182

robotics innovation and does not address inatable structures. Ledkov and Aslanov surveyed 183

contact and contactless ADR approaches in 2022 Ledkov and Aslanov [2022], providing com- 184

prehensive coverage of nets, harpoons, ion beam shepherds, and electrodynamic tethers, but 185

soft and compliant capture mechanisms receive minimal treatment and inatable structures 186

for ADR are absent. Zhao et al. reviewed rigid robotic manipulators for in-orbit servicing and 187

ADR in 2024 Zhao et al. [2024], covering Denavit-Hartenberg kinematics, impedance control, 188

and comparative arm performance; soft and inatable manipulators are outside scope. 189

The most relevant prior survey is Zhang et al. (2023), who examined soft robotics for 190

space across actuation, sensing, and manipulation Zhang et al. [2023a]. That work identies 191

vacuum as a challenge for pneumatic actuation and catalogues the soft gripper landscape; 192

however, it does not cover the inatable structure platform on which soft robots operate, nor 193

the enabling subsystems (power, thermal, AOCS, assembly) necessary for mission viability, 194

nor the dual ADR-and-exploration organising principle developed here. 195

The unique contribution of this survey is threefold. First, it covers eight enabling tech- 196

nology areas through a single integrative lens, rather than the one or two areas addressed 197

by prior reviews. Second, it presents the rst unied treatment of both ADR and explo- 198

ration applications as manifestations of the same fabric-based technology class. Third, it 199

maps cross-domain connections  between, for example, thermal management and actuator 200

design, or fold patterns and debris protection  that can only be identied from a broad 201

survey perspective. 202

Table 1: Comparison of this survey with prior reviews in Progress in Aerospace Sciences covering adjacent domains. ✓= covered;  = not covered; ∼= partial coverage.

Topic This survey Zhao 2024 Ledkov 2022 Flores-Abad 2014

Soft/compliant capture ✓  ∼  Inatable robotic arms ✓    Inatable debris shields ✓    Inatable habitats ✓    Rigid ADR approaches ∼ ✓ ✓ ✓ Rigid manipulators ∼ ✓ ∼ ✓ Materials & fabrics ✓    Power systems ✓    Thermal management ✓    AOCS for large structures ✓    Robotic in-orbit assembly ✓ ∼  ∼ Sensing & SHM ✓    Deployment mechanics ✓   

Year 2026 2024 2022 2014 Soft/inatable focus Primary None Minimal None

The Paradigm Shift: Vacuum as Design Resource 203

A recurring theme throughout this survey is the inversion of the conventional assumption 204

that space vacuum is hostile to soft robotic systems. Three independent developments chal- 205

lenge this assumption. First, Sirbu et al. demonstrated vacuum-gap electrostatic multilayer 206

actuators in 2025 that require vacuum to function: thin-lm polymer multilayers with inter- 207

nal vacuum gaps zip closed on electrical activation, producing forces exceeding 4 N from a 208

0.7 g actuator at bandwidths above 100 Hz Sirbu et al. [2025]. On Earth, a vacuum pump 209

would be required to create this operating condition; in space, the environment provides it 210

at no mass or power cost. Second, the conning pressure for granular and layer jamming  211

which terrestrially requires evacuating a sealed membrane with a pump  is provided for 212

free by the ambient vacuum dierential against a pressurised inatable interior Fitzgerald 213

et al. [2020]. Third, DEMES gripper geometry provides a passive negative feedback loop 214

in microgravity: grip force increases as a oating target drifts away from the actuator tip, 215

oering passive capture stability without active control  a property that is useful only in 216

the microgravity environment Araromi et al. [2015]. 217

These developments suggest that soft inatable robotic systems are not merely terrestrial 218

technology adapted for space, but a distinct engineering discipline with unique environment- 219

enabled advantages. 220

Review Methodology 221

The literature for this survey was assembled through a structured search strategy span- 222

ning multiple databases and source types. Primary databases searched include Scopus, 223

Web of Science, NASA Technical Reports Server (NTRS), ESA's publication repository, and 224

Google Scholar, using the following search term families: (i) inatable space structure 225

OR expandable habitat OR deployable membrane; (ii) soft robot* AND space OR 226

orbital; (iii) active debris removal AND (compliant OR soft OR inatable); and 227

(iv) technology-specic terms for each of the eight enabling areas (e.g., dielectric elastomer 228

actuator space, bre Bragg grating spacecraft, perovskite solar cell radiation). The tem- 229

poral scope spans 1960 (Project Echo) to early 2026, with no lower date restriction applied. 230

Inclusion criteria required that sources address at least one of the two application domains 231

(ADR or exploration) or one of the eight enabling technology areas in a space-relevant con- 232

text. Conference proceedings were included when they represented the primary publication 233

venue for mission results (e.g., AIAA, IAC, IEEE Aerospace). NASA technical memoranda, 234

ESA reports, and agency mission documentation were included for heritage programme data 235

not available in peer-reviewed form. Corporate press releases and datasheets were included 236

only when no peer-reviewed alternative existed for specic mission or material property 237

data. The eight technology areas were selected based on a preliminary scoping review that 238

identied all subsystem-level capabilities required for an operational soft inatable robotic 239

system at mission scale, following the principle that reviews in Progress in Aerospace Sci- 240

ences should enable the reader to assess system-level feasibility rather than component-level 241

performance alone. TRL assessments throughout the paper follow the NASA NPR 7123.1B 242

standard denitions NASA [2020]. 243

Survey Statistics 244

This survey reviews approximately 120 primary sources spanning the period from 1960 to 245

2026. Of these, approximately 74% are peer-reviewed journal papers or conference pro- 246

ceedings from indexed venues; the remainder comprises NASA technical memoranda, ESA 247

reports, and agency mission documentation. Coverage extends across eight technology areas 248

and two application domains, with the deepest literature pools in actuation (Zhang 2023 and 249

its references), inatable habitats (Litteken 2019 and the TransHab programme), and space 250

debris (Kessler 1978 through ESA 2025). The survey is organised with application use cases 251

preceding the technology state-of-the-art review, following the principle that applications 252

should motivate the technology landscape rather than the reverse. 253

2 The Case for Soft Inatables in Space 254

2.1 Space Debris Crisis and the Need for Active Removal 255

The accumulation of orbital debris is the dening environmental challenge of the space 256

age. Since Sputnik-1's launch in 1957, every mission has contributed to a growing cloud of 257

defunct satellites, spent rocket stages, and collision fragments. The debris environment is 258

now characterised not merely by nuisance but by irreversible instability. 259

Current Debris Environment 260

The ESA Space Environment Report for 2025 provides the most current comprehensive 261

characterisation ESA Space Debris Oce [2025]. As of early 2026, approximately 44,870 262

objects are tracked by ground-based surveillance networks, of which roughly one third are 263

operational satellites and two thirds are debris. The total catalogued population has grown 264

by more than 3,000 objects from fragmentation events in 2024 alone. At altitudes between 265

500 and 700 km  where ADR missions are most urgently needed  debris density is 266

comparable to or exceeds the density of active satellites. 267

Table 2: Current LEO debris population by size category (data from ESA Space Environment Report 2025 ESA Space Debris Oce [2025]).

Size category Estimated count Trackable? Primary threat

> 10 cm ∼54,000 Yes (radar) Catastrophic collision 110 cm ∼1,200,000 No Mission-ending damage 1 mm  1 cm ∼140,000,000 No Surface/solar panel damage < 1 mm > 1012 No Erosion/coating damage

Total mass ∼15,800 tonnes  

More than 650 fragmentation events have occurred in orbit since 1961, with signicant 268

contributors including the 2007 Chinese ASAT test (Fengyun-1C), the 2009 Cosmos-Iridium 269

collision, and the 2021 Russian ASAT test (Cosmos-1408). These events collectively added 270

thousands of trackable fragments and orders of magnitude more sub-centimetre particles. 271

The Kessler Syndrome: From Prediction to Conrmation 272

Kessler and Cour-Palais (1978) predicted that beyond a critical debris density, mutual col- 273

lisions among catalogued objects would generate fragments faster than atmospheric drag 274

could remove them, leading to an exponential growth cascade now called the Kessler syn- 275

drome Kessler and Cour-Palais [1978]. For nearly three decades this remained a theoretical 276

concern. Liou and Johnson (2006) demonstrated with the LEGEND orbital debris evolution 277

model that the predicted threshold has already been crossed in the 8001000 km altitude 278

band: even if all future launches were halted immediately, the debris population in these 279

shells would continue to grow due to existing collision rates among currently catalogued 280

objects Liou and Johnson [2006]. Extended 200-year projections (Liou and Johnson 2008) 281

80,000

Projected (no ADR)

Projected (5 ADR/yr)

70,000

Number of catalogued objects in orbit

60,000

50,000

Mega-constellation

era begins Kessler & Cour-Palais

(1978) prediction ESA 2025: 44,870 tracked (~54,000 est. >10 cm)

40,000

30,000

India ASAT (Mission Shakti)

China ASAT (Fengyun-1C)

Cosmos-Iridium

collision

20,000

10,000

0

1960 1970 1980 1990 2000 2010 2020 2030 2040 Year

Figure 1: Growth of the catalogued orbital debris population from 1960 to 2025, with projec- tions to 2040. Discrete fragmentation events (Chinese ASAT 2007, Cosmos-Iridium collision 2009) are visible as step increases. Red dashed line: projected growth without active de- bris removal. Green dashed line: projected stabilisation with ve large-object removals per year Liou et al. [2010]. Data from ESA Space Environment Report 2025 ESA Space Debris Oce [2025].

conrmed that the instability is neither transient nor recoverable without active interven- 282

tion Liou and Johnson [2008]. 283

The required rate of removal has been quantied. Liou et al. (2010) showed that removing 284

at least ve large objects per year (primarily rocket bodies in the 8001000 km band) is nec- 285

essary and sucient to stabilise the LEO population over a 200-year projection horizon Liou 286

et al. [2010]. This represents an annual ADR cadence comparable to the total number of sig- 287

nicant deorbit missions conducted globally over the past decade  a formidable operational 288

challenge. 289

The Fragmentation Paradox 290

The dominant design approach to ADR  rigid robotic arms, exemplied by ESA's ClearSpace- 291

1 mission targeting the PROBA-1 satellite  faces a fundamental tension. Rigid contact 292

with a non-cooperative, tumbling debris object generates impulsive forces at the contact 293

interface. For an 8-tonne ENVISAT-class object rotating at 5 deg/s, e.deorbit trajectory 294

analyses reveal peak joint torques of 195 Nm at structural limits Ledkov and Aslanov [2022], 295

while experimental harpoon tests in the RemoveDebris mission saw a carbon-bre deploy- 296

able boom snap on contact with the capture target at 20 m/s Aglietti et al. [2020]. Wang 297

et al. (2023) note explicitly that rigid manipulation has the potential to generate fragments 298

during the capturing phase Wang et al. [2023], and Chen et al. (2024) characterise single 299

contact-based caging approaches as excessively risky for fast-tumbling targets Chen et al. 300

[2024a]. 301

This fragmentation paradox is quantiable in energetic terms. The NASA/ESA IMPACT 302

model identies a catastrophic fragmentation threshold of 10 J/g of specic energy at the con- 303

tact interface Liou and Johnson [2006]. A 100-kg debris object rotating at ω = 5 deg/s and 304

2Iω2 ≈sev- 305

grasped rigidly at a moment arm of 0.5 m experiences a contact energy of order 1

eral hundred joules at the grasp point. If this energy is absorbed by the contact structure 306

rather than dissipated, the resulting specic energy may approach or exceed the fragmen- 307

tation threshold for lightweight aluminium honeycomb solar panel structures. No published 308

paper has conducted a systematic quantitative comparison of fragment generation probabil- 309

ity between rigid and compliant capture mechanisms  this gap is identied as a priority 310

experimental question in Section 13. 311

Compliant and soft capture systems address the paradox by absorbing and redistributing 312

contact energy rather than transmitting impulsive forces. Eight distinct soft and compliant 313

capture approaches are reviewed in Section 3, ranging from gecko-inspired dry adhesives 314

(microgravity-validated at TRL 45 Jiang et al. [2017]) to DEMES grippers with mission 315

heritage on CleanSpace One Araromi et al. [2015] and inatable robotic arms Palmieri et al. 316

[2023]. None has yet demonstrated in-ight capture, establishing a clear technology gap that 317

motivates the investment in ight demonstration infrastructure discussed in Section 13. 318

Operational Consequences 319

The operational burden of the debris environment is no longer theoretical. At 550 km altitude 320

 the operating shell of many Starlink satellites  the trackable debris density is sucient 321

to require avoidance manoeuvres at a rate that consumes propellant reserves and interrupts 322

normal operations. Starlink's 144,404 avoidance manoeuvres in H1 2025 (65-fold increase 323

from 2021 ESA Space Debris Oce [2025]) represent a structural operational cost that scales 324

with constellation size. ESA's own operational satellites execute hundreds of manoeuvres 325

annually, with collision avoidance emerging as a primary mission-operations driver. The 326

economic externality  uncontrolled debris imposes avoidance costs on all operators  327

provides a market-failure argument for policy-mandated ADR that is increasingly reected 328

in international guidelines Liou et al. [2010]. 329

2.2 Human Exploration Beyond LEO: The Habitat Challenge 330

The second driver for soft inatable systems is the ambition for sustained human presence 331

beyond the ISS. NASA's Artemis programme, ESA's Moon Village concept, and private 332

ventures such as Orbital Reef collectively assume that humans will occupy permanent or 333

semi-permanent outposts in cislunar space, on the lunar surface, in Mars transit, and even- 334

tually on the Martian surface. All of these scenarios require pressurised habitable volumes 335

substantially larger than any single rigid module that can be launched within existing fairing 336

constraints. 337

The Mass and Volume Eciency Argument 338

Valle et al. (2019) provide the denitive comparative analysis of inatable versus metallic 339

pressurised structures Valle et al. [2019a]. The key metric is areal density (mass per unit 340

oor area, or equivalently, mass per unit pressurised volume): 341

Table 3: Mass eciency comparison of representative pressurised space modules (adapted from Valle et al. 2019 Valle et al. [2019a]).

Module Type Press. Vol. (m3) Mass (kg) Density (kg/m3)

TransHab concept Inatable 339 13,200 39 BEAM (as-built) Inatable 16 1,415 88 Columbus (ESA) Metallic 75 10,300 137 Tranquility (Node 3) Metallic 74 15,200 205

The mass eciency advantage derives directly from material specic strength. Vectran 342

HT, the primary restraint-layer fabric in BEAM and TransHab, has a tensile strength of 343

3.0 GPa at a density of 1.40 g/cm3, yielding a specic strength of 2,330 kN-m/kg Valle 344

et al. [2019a]. Kevlar 49, similarly used for restraint and micrometeoroid and orbital debris 345

(MMOD) protection, achieves approximately 2,080 kN-m/kg at the fabric level (3.0 GPa 346

UTS, 1.44 g/cm3 density) or 2,500 kN-m/kg at the lament level (3.6 GPa UTS) DuPont 347

[2019]. These compare to Ti-6Al-4V at 220 kN-m/kg and aluminium 7075-T6 at 204 kN- 348

m/kg: the fabric advantage is approximately one order of magnitude. This dierence directly 349

determines what pressurised volume can be delivered per kilogram of launch mass, and 350

therefore what human presence scenarios are economically feasible. 351

The volumetric launch eciency is equally compelling. A 300 m3 pressurised module at 352

metallic density would mass approximately 40,000 kg  exceeding the cargo capacity of any 353

current or planned launch vehicle for a single module. The Sierra Space LIFE 285 habitat, 354

targeting approximately 300 m3 of pressurised volume, folds into a fairing-compatible package 355

and deploys on orbit, representing a volume achievable in a single launch that has no metallic- 356

module equivalent Sierra Space Corporation [2024]. 357

BEAM as Technology Proof 358

The BEAM module, delivered to the ISS by SpaceX CRS-8 in April 2016 and expanded 359

in May 2016, constitutes the highest-TRL demonstration of crewed inatable space struc- 360

tures NASA Johnson Space Center [2017]. BEAM provides 16 m3 of pressurised volume at 361

a deployed mass of 1,415 kg and has maintained pressure integrity for more than eight years 362

without rigidisation. Operational experience includes periodic crew access for inspection and 363

equipment storage, structural health monitoring via embedded accelerometers and impact 364

detection systems, and characterisation of the thermal, radiation, and MMOD environment. 365

BEAM's deployment was not without diculty: initial expansion attempts on 28 May 366

2016 required 25 pressurisation bursts over approximately seven hours to overcome friction 367

between compressed softgoods layers, compared to the planned single-burst expansion. This 368

experience provided critical engineering data on fold-compression set and deployment relia- 369

bility that directly informs the design of future autonomous deployment systems. Kennedy 370

(2002) documents the TransHab programme's prior exploration of this challenge, including 371

burst pressure tests to 4× operating pressure and the critical importance of restraint-layer 372

preloading for deployment force prediction Kennedy [2002]. 373

Radiation: The Honest Assessment 374

BEAM data from the September 2017 solar particle event (SPE) revealed a critical nding 375

that must be stated clearly NASA Johnson Space Center [2017]. Absorbed dose measure- 376

ments in BEAM during the SPE were approximately 22.5 mGy, compared to approximately 377

0.25 mGy measured simultaneously in adjacent metallic ISS habitable volumes  an 810× 378

ratio. This nding demonstrates that fabric walls alone provide substantially less radiation 379

shielding than the aluminium walls of conventional modules. 380

This is not a disqualifying result, but it is a design constraint. The TransHab architecture 381

addressed this through a water-wall concept: a ∼10 cm thick water reservoir integrated into 382

the inner wall layers that provides both radiation shielding (hydrogen-rich material) and 383

useful crew water storage. Norbury et al. (2025) review passive shielding materials for 384

space and conrm that polyethylene achieves a 27.8% mass saving relative to aluminium 385

for equivalent proton shielding at the same areal density Norbury et al. [2025]. The design 386

solution is established; its implementation requires deliberate integration rather than passive 387

reliance on wall thickness. 388

2.3 Unifying Thesis: Shared Fabric Technology Across Applications 389

The central organising principle of this survey is that the high-strength fabric technology 390

enabling inatable habitats is the same technology enabling compliant ADR capture arms, 391

large deployable debris shields, and the soft robotic systems operating within and around 392

both. This material unity has engineering consequences that extend beyond mere analogy. 393

Material Traceability Across Applications 394

Table 4 maps the four primary fabric families across their roles in dierent application do- 395

mains. The key observation is that the same material qualication data  creep behaviour, 396

AO erosion yield, UV degradation rate, thermal cycling tolerance  is relevant across all 397

applications. A Vectran creep characterisation campaign conducted for habitat restraint- 398

layer lifetime prediction Weadon [2013] is directly applicable to Vectran inatable robotic 399

arm links Palmieri et al. [2023]. A Nextel/Kevlar debris shield hypervelocity test cam- 400

paign Destefanis et al. [2003] produces data applicable to both habitat MMOD protection 401

and inatable debris shield design Cha et al. [2024]. 402

Table 4: Shared fabric technology across application domains. The same material families serve multiple functions, sharing qualication heritage and manufacturing processes.

Material Habitat role ADR role Robotic arm role

Vectran HT Restraint layer (primary load) Inatable arm links Inatable ma- nipulator links Kevlar 49 MMOD rear wall; restraint co-layer

Net tether; shield backing Arm outer jacket

Nextel 440 MMOD bumper (ceramic) Debris shield bumper layer 

Kapton/Mylar MLI outer lay- ers; bladder liner Shield thermal layer Bladder inner liner Beta cloth AO-resistant outer cover  AO-resistant cover

The Mars Airbag Precedent 403

Vectran's role in the Mars Pathnder (1997), Mars Exploration Rover (2004), and subse- 404

quent airbag systems provides heritage that extends beyond Earth orbit. These missions 405

demonstrated that Vectran-based inatable structures can survive the combined stresses of 406

launch vibration, interplanetary cruise, hypervelocity atmospheric entry, and impact landing 407

on an extraterrestrial surface Litteken [2019]. The qualication data base thus spans not 408

merely LEO but the full range of conditions relevant to deep space exploration  a heritage 409

directly relevant to future Mars transit habitat designs. 410

Origami Geometry Unies Packaging and Protection 411

A particularly striking example of cross-domain material unication is the Inatable Modular 412

Space Shield (IMSS) proposed by Cha et al. (2024) Cha et al. [2024]. The IMSS uses a wa- 413

terbomb origami tessellation to fold a multi-layer ultra-high-molecular-weight polyethylene 414

(UHMWPE)/Kevlar/Nextel shield into a package achieving 90% volume reduction relative 415

to a rigid Whipple shield of equivalent protection. The same Miura-ori and waterbomb 416

fold patterns Miura [1985] used in IMSS for debris shield deployment are the canonical fold 417

patterns for large membrane space structures generally Schenk et al. [2014]  packaging 418

eciency and multi-shock protection are simultaneously optimised by the same tessellation 419

geometry. 420

Scale-Dependent Challenges 421

While the material foundation is shared, the engineering challenges depend strongly on scale. 422

The scale-dependent challenge landscape can be summarised as follows: at centimetre scale 423

(soft gripper ngers), actuation force and contact compliance dominate the design; at metre 424

scale (inatable arms, BEAM-class habitats), deployment mechanics and pressure-retention 425

integrity dominate; at 10-metre scale (large solar concentrators, small debris shields), control- 426

structure interaction begins to matter; at 100-metre scale (large debris shields, solar power 427

collectors), attitude and orbit control, aerodynamic drag compensation, power generation, 428

and thermal management become the primary engineering challenges, for which no ight 429

heritage exists. 430

This survey is organised to trace the technology from its best-proven applications (TRL 9 431

materials, TRL 9 BEAM habitat, TRL 89 rigid solar arrays) through to the most speculative 432

future capabilities (TRL 23 pressure-stabilised membrane AOCS, TRL 34 vacuum-gap 433

actuation), making explicit at each stage what is demonstrated, what is extrapolated, and 434

what requires new research. 435

Why Soft? Why Inatable? Why Now? 436

Three converging developments make this survey timely. 437

Material advances. Vectran and Kevlar have matured to TRL 9 in space environments. 438

Perovskite/CIGS tandem solar cells, demonstrated at 2,100 W/kg with 85% proton radia- 439

tion retention after equivalent 50-year LEO exposure Lang et al. [2020], promise to integrate 440

power generation into inatable membrane layers at specic powers unachievable with con- 441

ventional rigid panels. Cryogenic metallic cable-based soft robots (Foster-Hall et al. 2025) 442

maintain full range of motion at −196 ◦C, solving the elastomer embrittlement problem for 443

deep-space applications Foster-Hall et al. [2025]. 444

Mission context. The commercial station era (Orbital Reef, Axiom, LIFE, Starlab) cre- 445

ates the rst sustained market demand for habitable volume beyond ISS. ESA's ClearSpace-1 446

mission, targeting PROBA-1 for retrieval in the late 2020s, establishes ADR as an opera- 447

tional rather than experimental activity. The convergence of launch cost reduction (SpaceX 448

Falcon 9, Starship) with mission demand means that the technology development cost of 449

inatable systems is now justiable against a credible mission pull. 450

Paradigm shift. As outlined in Section 1, the space environment is increasingly un- 451

derstood as a resource for soft robotic systems rather than an obstacle. Vacuum-gap ac- 452

tuation Sirbu et al. [2025], jamming without pumps Fitzgerald et al. [2020], and passive 453

microgravity compliance Araromi et al. [2015] represent a qualitative shift in what the space 454

environment enables. This survey maps these opportunities systematically across the full 455

technology stack. 456

The following sections develop the application use cases (Sections 3 and 4) before re- 457

viewing the enabling technology state-of-the-art (Sections 512), and concluding with a 458

consolidated gap analysis and research roadmap (Section 13). 459

3 Use Cases: Active Debris Removal 460

The orbital debris environmentcharacterised in Section 2.1represents the most urgent 461

operational motivation for soft inatable robotic systems in space. With over 54,000 esti- 462

mated objects larger than 10 cm, 15,800 tonnes of total orbital mass, and a 65-fold increase 463

in Starlink collision avoidance manoeuvres since 2021 ESA Space Debris Oce [2025], the 464

operational urgency is undeniable. 465

The scientic foundation for active debris removal (ADR) was established by Kessler and 466

Cour-Palais Kessler and Cour-Palais [1978], who developed the rst mathematical model pre- 467

dicting cascading collisional fragmentation in low Earth orbit (LEO). Their analysis identied 468

three debris population regimesstable, critical, and cascadingand predicted the forma- 469

tion of a debris belt within a century. Subsequent Monte Carlo simulations by Liou and 470

Johnson Liou and Johnson [2006, 2008] using the NASA LEGEND model with 200-year pro- 471

jections across 50 runs demonstrated that the LEO debris population had already crossed 472

the instability threshold: the number of objects would continue to grow even with zero future 473

launches. Their work quantied the minimum intervention rate, establishing that at least 474

ve large objects per year must be removed from the 8001000 km altitude bands to stabilise 475

the environment Liou et al. [2010]. At approximately 550 km altitude, debris spatial density 476

now equals active satellite densityan unprecedented situation that fundamentally changes 477

the risk calculus for orbital operations ESA Space Debris Oce [2025]. 478

This section examines the role of soft and inatable systems in addressing the debris 479

challenge. We rst review conventional rigid capture approaches and their inherent fragmen- 480

tation risk (Section 3.1), then survey eight distinct soft and compliant capture mechanisms 481

(Section 3.2), and nally discuss inatable debris shields as passive protection infrastructure 482

(Section 3.3). 483

3.1 Rigid Capture Approaches and Fragmentation Risk 484

Active debris removal using rigid robotic manipulators has been the dominant paradigm in 485

mission planning for the past two decades. Zhao et al. Zhao et al. [2024] provide the most 486

recent comprehensive review in Progress in Aerospace Sciences of rigid manipulators for on- 487

orbit servicing and ADR, covering ight-heritage systems such as the Canadarm and the 488

European Robotic Arm (ERA), cancelled missions including ESA's e.deorbit, and planned 489

missions such as ClearSpace-1. The review documents the extensive engineering heritage of 490

rigid robotic arms but also explicitly acknowledges the potential for fragmentation generation 491

during debris capture Zhao et al. [2024]. 492

Ledkov and Aslanov Ledkov and Aslanov [2022] survey the full spectrum of ADR meth- 493

ods in Progress in Aerospace Sciences, including nets, harpoons, robotic arms, tentacles, ion 494

beam shepherding, laser ablation, electrostatic tractors, and electrodynamic tethers. Their 495

analysis notes that contactless methods such as ion beam shepherdingcapable of deorbit- 496

ing a 2-tonne debris object in 34 monthscarry zero mechanical impact risk, but require 497

extended proximity operations and signicant power budgets. Contact-based methods, while 498

operationally faster, necessarily introduce mechanical loads to the target. 499

The only in-orbit ADR technology demonstration to date is the RemoveDebris mission, 500

documented by Aglietti et al. Aglietti et al. [2020]. This mission successfully demonstrated 501

net capture of a CubeSat at 5 cm/s relative velocity and 7 m separation distance, as well 502

as harpoon ring at 20 m/s into a target panel at 1.5 m range. Two results are particu- 503

larly instructive. First, the net capture succeeded but was conducted against a cooperative 504

2U CubeSat (expanded to approximately 1 m pyramidal target), which is not representative 505

of real debris targets of 500 kg8 tonnes tumbling at 15 deg/s. Second, and more critically, 506

the harpoon test resulted in the snapping of the carbon bre boom from impact forces, de- 507

spite the harpoon itself being retained by its tether Aglietti et al. [2020]. This structural 508

failure during a controlled test illustrates the magnitude of impulse loads that contact-based 509

capture imposes. 510

3.1.1 The Fragmentation Paradox 511

The central paradox of rigid-body ADR is that the very act of removing debris may generate 512

new fragments, potentially worsening the environment it aims to protect. This concern is 513

supported by multiple lines of evidence: 514

ˆ Wang et al. Zhang et al. [2022] state explicitly that rigid behaviour has the potential 515

to generate fragments during [the] capturing phase, hence increase [the] risk of further 516

space debris. 517

ˆ Chen et al. Chen et al. [2024a] assess that single contact-based caging [is] excessively 518

risky for fast-tumbling targets with unknown massmomentum transfer could create 519

new debris. 520

ˆ Dynamic simulations of the cancelled e.deorbit mission show peak torques of 195 Nm 521

at the manipulator joints when attempting to capture a target tumbling at only 5 deg/s 522

(the ENVISAT upper stage) Stol et al. [2017], reaching the operational limits of the 523

robotic joints. 524

ˆ The Aerospace Corporation's IMPACT model establishes 10 J/g specic energy as the 525

threshold for catastrophic fragmentation of a satellite Aerospace Corporation [2020]. 526

ClearSpace-1, the rst contracted commercial debris removal mission (ESA, ¿86M con- 527

tract), plans to use four rigid robotic arms to capture the Proba-1 satellite (95 kg, 0.6×0.6× 528

0.8 m) ClearSpace SA and European Space Agency [2020]. The mission's planning was itself 529

disrupted by the debris problem: the original target, the VESPA upper stage, was struck by 530

a tracked debris object during mission preparation, illustrating the cascading urgency of the 531

debris environment ClearSpace SA and European Space Agency [2020]. Launch is currently 532

planned for approximately 2029. 533

To place the fragmentation risk in perspective, we note that a rigid robotic arm exerting 534

195 Nm of torque on a 100 kg target at a 0.5 m lever arm produces a contact force of 390 N. 535

If this force acts over a contact area of 10 cm2 on a honeycomb panel with typical crush 536

strength of 13 MPa, the resulting stress of 0.39 MPa falls below the crush threshold of 537

the primary structure. However, the fragmentation risk is not primarily to the strongest 538

structural components, but to the most vulnerable: degraded solar panel hinge joints, aged 539

thermal blanket fasteners, corroded aluminium alloy brackets, and antenna feed structures 540

that have experienced decades of thermal cycling, UV degradation, and atomic oxygen ero- 541

sion. These appendage materials may have lost 3060% of their original strength through 542

environmental degradation, reducing eective crush thresholds well below nominal values. 543

For a tumbling 1000 kg upper stage at 5 deg/s, the angular momentum is approximately 544

50 N·m·s, and the impulsive loads during despin are proportionally larger. Applying the 545

catastrophic fragmentation threshold of 10 J/g from the IMPACT model Aerospace Corpo- 546

ration [2020], Johnson et al. [2001]: if a rigid grasp concentrates 50 J of despin energy into a 547

100 g solar panel hinge assembly, the resulting specic energy of 0.5 J/g remains below the 548

10 J/g threshold, but contact with a 10 g degraded thermal blanket fastener at equivalent 549

energy would yield 5 J/gapproaching the threshold. A compliant grasp distributing the 550

same energy over a larger contact area and longer time period reduces peak specic energy 551

by one to two orders of magnitude. 552

The fragmentation risk is therefore physically plausible and supported by qualitative as- 553

sessments, though not yet experimentally quantied. This survey adopts the precautionary 554

principle: compliant capture is preferred until quantitative data become available, on the 555

basis that the consequences of inadvertent fragmentation during ADRpotentially generat- 556

ing hundreds of new tracked objectsare severe enough to warrant risk-averse technology 557

selection even in the absence of denitive comparative data. A comprehensive, quantita- 558

tive comparison of fragmentation probability as a function of contact compliance remains 559

the single highest-priority open experimental question the community must address (see 560

Section 13). 561

Table 5 summarises the principal ADR technology classes, their technology readiness 562

levels (TRL), contact characteristics, and assessed fragmentation risk. 563

3.2 Soft and Compliant Capture Mechanisms 564

The fragmentation risk inherent in rigid capture has motivated the development of soft and 565

compliant alternatives that absorb, rather than transmit, kinetic energy during the capture 566

interaction. Eight distinct soft and compliant capture approaches have been documented in 567

the literature, all currently at TRL 25. We review each in turn, organised by their operating 568

principle: adhesion-based, bistable/passive, inatable-arm, and net-plus-inatable systems. 569

3.2.1 Gecko-Inspired Dry Adhesive Grippers 570

The most mature soft capture technology is the gecko-inspired dry adhesive gripper demon- 571

strated by Jiang et al. Jiang et al. [2017]. Published in Science Robotics, this system uses 572

shear-activated van der Waals adhesion pads with a load-sharing tendon-pulley mechanism 573

that scales adhesion from small patches to large contact areas. Critically, a nonlinear pas- 574

Table 5: Comparison of active debris removal technology classes. Fragmentation risk is assessed qualitatively based on published evidence; a quantitative comparison remains an open research gap.

Method TRL Contact Frag. Risk Key Limitation

Rigid robotic arm 56 Direct, rigid High Peak torques at joint limits; brittle appendage damage Harpoon 6 Penetrative Very high Boom failure in RemoveDebris; target perforation Thrown net 7 Enveloping Moderate Impulse at net closure; entanglement dynamics Ion beam shepherd 4 Contactless None 34 month timeline; high power Laser ablation 3 Contactless None Pointing accuracy; space weapon concerns Gecko adhesive 45 Shear adhesion Very low Clean surfaces assumed; no tumbling test Soft/inatable arm 23 Compliant Low Precision; pneumatic in vacuum Bistable gripper 23 Passive snap Low Energy barrier tuning; untested in vacuum Net + inatable (INSIDeR) ∼4 Controlled net Low System integration unproven in orbit

sive wrist provides high stiness during normal manipulation but becomes compliant under 575

overload, oering inherent protection against excessive contact forces. 576

The gecko gripper was validated in actual microgravity during NASA parabolic ight 577

campaigns, achieving capture success rates of 100% for spherical targets, 75% for cubic tar- 578

gets, and 81% for cylindrical targets, with objects up to approximately 400 kg and diameters 579

exceeding 1 m Jiang et al. [2017]. Failures were attributed to human operator misalignment 580

rather than adhesive performance. The system achieves essentially zero mechanical impact 581

forcea fundamental advantage for fragmentation avoidance. We note, following the taxon- 582

omy of Shintake et al. Shintake et al. [2018], that the gecko gripper is more precisely classied 583

as a compliant end-eector mechanism on a rigid platform rather than a fully soft robotic 584

system; nevertheless, its compliant capture interface directly addresses the fragmentation 585

concern. At TRL 45, it represents the highest-readiness soft capture technology, though 586

signicant gaps remain: all testing used cooperative (stationary) targets, and performance 587

under space vacuum, UV radiation, atomic oxygen exposure, and thermal cycling has not 588

been demonstrated. 589

3.2.2 Dielectric Elastomer Minimum Energy Structure (DEMES) Grippers 590

Araromi et al. Araromi et al. [2015] developed a DEMES-based deployable gripper explic- 591

itly for the CleanSpace One ADR mission. The device uses dielectric elastomer actuators 592

(DEAs) bonded to a exible frame, achieving rollable compact storage and deployment to 593

a multi-segment gripper with bending angles exceeding 60°. Each arm produces forces in 594

the mN range, sucient only for microgravity manipulation of small, lightweight targets. 595

The system demonstrated over 860,000 actuation cycles with individual arm mass below 596

0.65 g Araromi et al. [2015]. At TRL 34, the DEMES gripper is notable as the only soft 597

capture device explicitly designed for an actual ADR mission, although the CleanSpace One 598

mission architecture subsequently evolved without the gripper ying. Key limitations in- 599

clude the high operating voltage (∼kV) required for DEAs in vacuum (arcing risk) and the 600

absence of cryogenic or thermal cycling testing. 601

3.2.3 Bistable and Passive Capture Grippers 602

Two distinct bistable gripper concepts have been proposed for ADR. Liu et al. Liu et al. 603

[2022] developed a bistable snap-through gripper that captures targets using the kinetic 604

energy of the collision itself, requiring no external power for the grasping action. The gripper 605

deforms on contact, absorbs kinetic energy, triggers a bistable snap, and locks into the closed 606

conguration. The energy barrier is adjustable through pre-deformation of the bistable 607

elements, allowing tuning for dierent target masses and approach velocities Liu et al. [2022]. 608

This passive capture concept eliminates the need for precise actuation timinga signicant 609

advantage for tumbling, non-cooperative targets. 610

Zhang et al. Zhang et al. [2023b] propose a Venus ytrap-inspired bistable origami gripper 611

actuated by a shape memory alloy spring actuator (SMASA) that provides slow energy 612

storage followed by rapid release, with a DEA bristle-locking structure that prevents target 613

escape after capture. Capture is achieved within approximately 300 ms, and the device has 614

been demonstrated on complex geometries including asteroid models and spacecraft mock- 615

ups Zhang et al. [2023b]. Both bistable concepts remain at TRL 23, with no vacuum, 616

thermal, or microgravity testing. 617

3.2.4 Thermally Qualied Soft Grippers 618

Addressing the thermal environment is critical for any space capture mechanism. Ruiz 619

Vincueria et al. Ruiz Vincueria et al. [2023] developed a multi-layered soft gripper combining 620

TPU, silicone, PTFE, and aerogel layers, tested across the full orbital thermal range from 621

−180°C to +220°C. A counter-intuitive but operationally signicant nding is that grasping 622

forces increase by 220% at cryogenic temperatures due to cold stiening of the elastomeric 623

layers, while decreasing by at most 50% at the hot extreme Ruiz Vincueria et al. [2023]. The 624

gripper uses MoS2 solid lubricant for vacuum compatibility and is available in dual and quad 625

arm congurations. This work provides the most quantitative thermal performance data 626

for any soft capture device and explicitly compares its approach against the ClearSpace-1 627

and Astroscale rigid arm architectures. However, all testing was conducted in laboratory 628

conditions without vacuum, radiation, or microgravity validation (TRL 2). 629

Foster-Hall et al. Foster-Hall et al. [2025] introduce a fundamentally dierent approach 630

to the cryogenic challenge: metallic cable-driven soft robotic structures tested at −196°C in 631

liquid nitrogen. Unlike elastomeric soft robots that embrittle at cryogenic temperatures, the 632

modular metallic cable structures exhibited only 5% stiness increase over 100 actuation cy- 633

cles, maintained full range of motion, and showed no microfractures under scanning electron 634

microscopyconsistent with cold-working behaviour in stainless steel rather than brittle 635

failure Foster-Hall et al. [2025]. Two-dimensional grasping was demonstrated at −196°C. At 636

TRL 23, this work opens a new design paradigm for soft space robotics beyond elastomers, 637

though three-dimensional manipulation and vacuum testing remain to be demonstrated. 638

3.2.5 Inatable Robotic Arms for Capture 639

Palmieri et al. Palmieri et al. [2023] developed the POPUP robot: a 7-DOF manipulator 640

with inatable links and rigid electric motor joints, incorporating visual servoing via dual 641

cameras and high-stiness bre reinforcement. The inatable links provide signicant mass 642

and volume reduction compared to equivalent rigid arms, and simulation demonstrates debris 643

capture feasibility despite the inherent compliance of the links Palmieri et al. [2023]. A 3- 644

DOF ground prototype has been statically characterised (TRL 3), but key challenges remain: 645

the compliance of inatable links reduces end-eector positioning precision, the pneumatic 646

ination system must operate in vacuum, and no thermal or radiation testing has been 647

performed. 648

3.2.6 INSIDeR: Net Capture with Inatable Deployment 649

The Innovative Net and Space Inatable structure for active Debris Removal (INSIDeR) 650

is a patented CNES/ESA-funded concept that combines the proven in-orbit heritage of 651

net capture (demonstrated by RemoveDebris) with inatable deployment structures CT 652

Ingénierie et al. [2017, 2021]. The system architecture comprises an inatable ring and 653

two inatable masts that deploy and guide a capture net, followed by a deorbit tether for 654

removal. The complete capture sequence proceeds through six phases: ination of the ring 655

and masts, net deployment, approach boost, mast detachment and deation, net capture, 656

and tether-assisted deorbit CT Ingénierie et al. [2017]. 657

A key innovation is that the inatable masts provide controlled, slow net dynamics, 658

eliminating the large impulse peaks associated with conventional spring-ejected nets and 659

thereby reducing momentum transfer to the target CT Ingénierie et al. [2021]. The system 660

packages into a cube of approximately 50 cm per side, forming a plug-and-play ADR kit 661

adaptable to any target mass, morphology, or tumbling rate. Developed over 15 years by 662

CT Ingénierie and AirCaptif (Michelin group) with CNES and ESA co-funding, INSIDeR has 663

reached TRL ∼4 at the system level (individual subsystem technologies at TRL 5+), with 664

a ground demonstrator under construction as of 2021 CT Ingénierie et al. [2021]. ABAQUS 665

nite element simulations have conrmed net capture feasibility. 666

Table 6 provides a comprehensive comparison of all documented soft and compliant cap- 667

ture approaches. 668

Flight qualified

Concept Validation

102

Tendon-driven

Gecko adhesive

SMA (one-shot)

101

Bistable gripper Inflatable arm

Vacuum-gap electrostatic

Force output (N)

100

10 1

Category / Est. mass

10 2

Adhesive

Pneumatic

Electroactive

0.1 kg

DEMES/DEA

Mechanical

2 kg

Shape memory

5 kg

10 3

Passive

INSIDeR (net capture,

TRL 4)

1 2 3 4 5 6 7 8 9 10 Technology Readiness Level (TRL)

Figure 2: Force output versus technology readiness level (TRL) for soft and compliant cap- ture approaches. Marker size indicates system mass. The gecko adhesive gripper occupies the highest-TRL, highest-force quadrant, representing the most ight-ready soft capture technology.

The most signicant observation from this landscape is the absence of orbital ight 669

heritage for any soft capture system. The gecko adhesive gripper, at TRL 4 with microgravity 670

validation, and INSIDeR, at TRL 4 with system-level ground demonstration, represent the 671

nearest-term candidates for ight demonstration. We identify the combination of a gecko 672

adhesive gripper mounted on an inatable arm with bre Bragg grating structural health 673

monitoring (see Section 8.1) as the most ight-ready near-term soft ADR demonstratora 674

system that leverages the highest-TRL end-eector, the mass eciency of inatable links, 675

and embedded sensing for operational awareness. 676

Table 6: Technology readiness and performance comparison of soft and compliant capture mechanisms for active debris removal. No soft capture system has own an orbital capture mission to date.

Approach Key Reference TRL Force Output µg Test Key Limitation

4a ≤400 kg objects Yes Clean surfaces; no tumbling

Gecko adhesive Jiang 2017 Jiang et al. [2017]

3b mN range No Very low force; HV arcing

DEMES/DEA Araromi 2015 Araromi et al. [2015]

Inatable arm Palmieri 2023 Palmieri et al. [2023]

3 Not quantied No Low precision; pneumatic in vacuum Flytrap origami Zhang 2023 Zhang et al. [2023b]

23 Bistable snap No SMA slow reset; HV in vacuum

Bistable gripper Liu 2023 Liu et al. [2022] 2 Passive (KE input) No Energy barrier tuning Cryo metallic Foster-Hall 2025 Foster- Hall et al. [2025]

23 Not quantied No 2D only; no vacuum

Thermal multi-layer Ruiz 2024 Ruiz Vin- cueria et al. [2023]

2 +220% at cryo No Lab only; no vacuum

INSIDeR (net+in.) ESA SDC 2017/21 CT Ingénierie et al. [2017, 2021]

4 N/A (net) Sim. only System integration

aTRL 4 per NASA NPR 7123.1B: parabolic ight (∼20 s µg per parabola) constitutes component validation in a simulated relevant environment rather than a fully relevant orbital environment (TRL 5). bTRL 3: 860,000 cycles demonstrated in ambient conditions, but no space environment testing (vacuum, thermal cycling, radiation) performed.

3.3 Inatable Debris Shields 677

Beyond active capture, inatable structures oer a complementary approach to the debris 678

problem through passive shielding. Conventional rigid Whipple shields Arnold et al. [2009], 679

which use spaced aluminium bumper plates to disrupt and disperse hypervelocity projectiles 680

before they reach the pressure wall, are eective but carry signicant mass and volume 681

penalties. The substitution of rigid bumper plates with exible fabric layersusing the 682

same high-strength materials (Nextel ceramic fabric, Kevlar, and ultra-high molecular weight 683

polyethylene, UHMWPE) that form the basis of inatable habitat wallsenables deployable 684

shields with dramatically improved packaging eciency. 685

Destefanis et al. Destefanis et al. [2006] demonstrated that stued Whipple shields using 686

Nextel and Kevlar layers protect against projectiles twice the diameter of those stopped by 687

standard aluminium Whipple shields at equal areal density. This nding established the 688

performance advantage of fabric-based shielding architectures that underlies both habitat 689

micrometeoroid and orbital debris (MMOD) protection and standalone shield concepts. 690

Cha et al. Cha et al. [2024] present the Inatable Multi-Shock Shield (IMSS), which ap- 691

plies waterbomb tessellation origami to create a deployable multi-bumper debris shield that 692

expands approximately 80% beyond its initial radius while achieving 90% volume savings 693

compared to an equivalent rigid Whipple shield. The IMSS uses UHMWPE bre for ballistic 694

protection within a ve-bumper conguration, with 50 mm bumper spacing accommodated 695

in a 400 mm stowed stack Cha et al. [2024]. A critical design feature is that all material 696

in the deployed conguration contributes to debris protectionthere is no structural dead 697

weight. The origami fold geometry that enables compact packaging simultaneously creates 698

the inter-bumper spacing required for eective hypervelocity projectile disruption, embody- 699

ing a dual-functionality design principle applicable to large deployable structures generally 700

(see Section 4.3 for related deployment mechanics). 701

At TRL 23, the IMSS concept requires further development in hypervelocity impact 702

validation, large-scale (>10 m) deployment demonstration, and ination system design. 703

Nevertheless, the material commonality between inatable debris shields, inatable habi- 704

tat MMOD layers, and inatable robotic arm structural fabrics reinforces the survey's 705

central thesis: the same high-strength fabric technology baseVectran, Kevlar, Nextel, 706

UHMWPEenables debris capture, debris protection, and habitable volume creation. 707

For very large-scale applications, inatable debris shields of 100 m class have been pro- 708

posed as orbital infrastructure to protect high-value assets or clear debris corridors. Such 709

structures would require the attitude and orbit control technologies discussed in Section 11 710

and the robotic in-orbit assembly capabilities reviewed in Section 12, linking the passive 711

protection concept back to the active robotic systems that are the primary focus of this 712

survey. 713

4 Use Cases: Habitats and Exploration 714

Inatable space structures for human habitation represent the second major application 715

domain where soft and exible technologies oer transformative advantages over conventional 716

rigid systems. The fundamental value proposition is mass eciency: high-strength fabrics 717

such as Vectran and Kevlar possess specic tensile strengths of 2,330 and 2,080 kN·m/kg 718

respectively at the fabric level (or 2,500 kN·m/kg for Kevlar 49 lament)more than an 719

order of magnitude greater than titanium alloy Ti-6Al-4V at 220 kN·m/kg or aluminium 720

7075 at 204 kN·m/kg Valle et al. [2019a]. This advantage translates directly into the ability 721

to launch habitable volumes that would be physically impossible with metallic construction 722

within current launch vehicle fairing constraints. A fabric-walled habitat is not merely a 723

lighter alternative to a metallic module; it enables architectural possibilitiesvolumes of 724

3001,400 m3that have no rigid equivalent. 725

This section traces the heritage of inatable space habitation from its origins in 1960 to 726

the present day (Section 4.1), reviews current commercial programs (Section 4.2), surveys 727

future concepts for lunar, Martian, and planetary applications (Section 4.3), and addresses 728

the critical issue of radiation shielding with an honest assessment of the BEAM solar particle 729

event ndings (Section 4.4). 730

4.1 Heritage Timeline: Echo to BEAM 731

The heritage of inatable space structures spans over six decades, progressing through a 732

non-linear TRL trajectory marked by both remarkable successes and programmatic setbacks. 733

Table 7 summarises the key milestones. 734

LIFE in-space

Volga airlock

IRVE-II

test (~2026, planned)

(1965)

(2009)

Mars Pathfinder

BEAM (2016)

TRL 9

TRL 7

TRL 6

(1997)

Echo 1 (1960)

Genesis I

LOFTID

TRL 9

TRL 8

(2006)

(2022)

TRL 9

TRL 8

TRL 7

IAE / Spartan 207

IRVE-3

ClearSpace-1 (~2029, planned)

(1996)

(2012)

Echo 2 (1964)

Genesis II

Sierra LIFE

TRL 7

TRL 7

TRL 5

(2007)

(UBP tests)

TransHab

InflateSail

TRL 9

TRL 8

(2024)

NASA

ESA / International

(1999)

(2017)

TRL 5

TRL 6

TRL 7

Commercial

Planned

1960 1970 1980 1990 2000 2010 2020 2030 Year

Figure 3: Heritage timeline of inatable space structures from Echo 1 (1960) to LOFTID (2022), illustrating the progression from passive communication reectors through human- rated habitats to active aerodynamic decelerators. Colour coding indicates programme ori- gin; marker size reects achieved TRL.

Table 7: Heritage timeline of inatable space structures, from passive communication reec- tors to human-rated orbital habitats. TRL ratings reect achieved (not planned) readiness at programme conclusion or present status.

Year Programme TRL Key Achievement

1960 Echo 1 (NASA) 9 30.5 m (100 ft) Mylar sphere; 8+ years on-orbit; global communications relay 1965 Volga airlock (USSR) 9 First human-rated inatable; Voskhod-2 EVA (Leonov); 40 airbags, 3 independent groups, 7 min ination 1996 IAE/Spartan 207 (NASA) 7 14 m antenna; 28 m Kevlar/Neoprene booms; Shuttle deployment demonstration 1997 Mars Pathnder airbags 9 Vectran fabric; operational landing on 3 missions (Pathnder, Spirit, Opportunity) 19972000 TransHab (NASA JSC) 56 8.2 m × 11 m; 5-layer shell; tested to 4× operating pressure; cancelled by Congress (HR 1654) 200607 Genesis I/II (Bigelow) 78 Orbital validation; 2.5+ years on-orbit; pressure retention conrmed 2009 IRVE-II (NASA LaRC) 7 3 m inatable reentry vehicle experiment; suborbital demonstration 2016+ BEAM (Bigelow/NASA) 9 16 m3; 1,415 kg; 8+ years on ISS; converted to cargo storage; operational 2022 LOFTID (NASA) 78 6 m inatable aerodecelerator; orbital reentry at Mach 30

4.1.1 Early Inatables: Echo and Volga (19601965) 735

Project Echo, initiated by NASA in 1960, deployed Echo 1 as a 30.5 m diameter Mylar 736

balloon serving as a passive communications reector Litteken [2019]. The satellite operated 737

for over eight years and enabled global communications experiments and geodetic measure- 738

ments. Echo 2 (1964) advanced the concept with a rigidisable aluminium foil/Mylar laminate 739

structure. While neither was habitable, the Echo programme demonstrated that large, thin- 740

walled inatable structures could survive the LEO environment for extended periods. 741

The Volga airlock, deployed for the Voskhod-2 mission in 1965, represents the rst human- 742

rated inatable space structure Litteken [2019]. Designed for Alexei Leonov's historic rst 743

spacewalk, the Volga used 40 airbags arranged in three independent groups to inate a 2.4 m 744

long, 1.2 m diameter cylindrical airlock in seven minutes. The successful EVA validated 745

the fundamental concept that pressurised inatable structures could safely support human 746

operations in space, albeit for a single use. 747

4.1.2 TransHab: Proving the Five-Layer Architecture (19972000) 748

The Transit Habitat (TransHab) programme at NASA Johnson Space Center represented the 749

most ambitious inatable habitat development prior to BEAM. Under Principal Architect 750

Kriss Kennedy Kennedy [2002] and shell lead Gerard Valle, the team developed an 8.2 m 751

diameter, 11 m long module with a ve-layer shell architecture that has become the standard 752

for all subsequent inatable habitat designs Valle et al. [2019a]: 753

1. Inner liner: Nomex scu protection layer. 754

2. Bladder: Multiple redundant layers, oversized relative to the restraint layer and car- 755

rying zero structural load. 756

3. Restraint layer: Tight basket-weave Kevlar/Vectran biaxial membrane, designed to 757

a safety factor of 4.0× per NASA-STD-5001. 758

4. MMOD shield: Ceramic (Nextel) bumper, open-cell foam spacer, and Kevlar rear 759

wallvacuum-packed for launch, with foam self-expanding in orbit. 760

5. Multi-layer insulation (MLI): 19 layers of double-aluminised Mylar/Kapton, with 761

perforated inner layers for venting during depressurisation. 762

TransHab was tested to 4× ambient pressure (>54 psig) in a September 1998 hydrostatic 763

burst test, and full-scale vacuum deployment was demonstrated Kennedy [2002]. Hyperveloc- 764

ity impact testing conrmed that the MMOD shield outperformed the aluminium structure 765

of ISS modules. The programme also pioneered the water wall radiation shelter concept, 766

positioning crew quarters within a rigid central core surrounded by water-lled containers 767

for radiation protection Kennedy [2002]. 768

Despite reaching TRL 56, TransHab was cancelled by Congressional action (HR 1654, 769

2000). The technology investment was preserved through patent licensing to Bigelow Aerospace, 770

which continued development commercially Kennedy [2002]. 771

4.1.3 Genesis and BEAM: Orbital Validation (20062016+) 772

Bigelow Aerospace launched Genesis I (2006) and Genesis II (2007) as uncrewed orbital test 773

modules, demonstrating pressure retention (69.672.4 kPa for Genesis II) and thermal per- 774

formance (average 26°C, range 4.532°C for Genesis I) over 2.5+ years Litteken [2019]. These 775

missions validated the TransHab-derived shell architecture in the actual orbital environment 776

for the rst time. 777

The Bigelow Expandable Activity Module (BEAM), launched to the International Space 778

Station in April 2016, represents the culmination of this heritage. BEAM provides 16 m3 of 779

habitable volume at a mass of 1,415 kg (88 kg/m3), compared to 137 kg/m3 for the Columbus 780

module and 205 kg/m3 for the Tranquility node Valle et al. [2019a]. While BEAM's mass- 781

per-volume ratio is higher than TransHab's projected 39 kg/m3reecting BEAM's small 782

size and relatively heavy end-ttingsthe comparison to metallic modules demonstrates the 783

eciency advantage of fabric-walled construction Valle et al. [2019a]. 784

BEAM's deployment provided a critical engineering lesson. Initial expansion attempts 785

failed, and the module required 25 short pressure bursts over approximately 7 hours to 786

achieve full deploymentin contrast to the planned rapid ination sequence NASA Johnson 787

Space Center [2017]. The root cause was attributed to softgoods layers adhering after years 788

of compression in the launch conguration. For future free-ying deep-space modules where 789

ISS crew intervention would not be available, this deployment failure mode must be resolved 790

through autonomous ination protocols. 791

After its planned two-year demonstration, BEAM's mission was extended to at least 2028. 792

The module has been converted to active cargo storage (approximately 130 cargo transfer 793

bags), demonstrating practical volumetric value beyond its test objectives NASA Johnson 794

Space Center [2017]. No pressure loss, structural degradation, or signicant MMOD impacts 795

have been recorded in over eight years of operation. The Distributed Impact Detection 796

System (DIDS) has continuously monitored for debris impacts throughout the mission. 797

4.2 Current Commercial Programs: LIFE, Orbital Reef, and Be- 798

yond 799

4.2.1 Sierra Space LIFE 800

The Large Integrated Flexible Environment (LIFE) programme by Sierra Space represents 801

the most advanced current inatable habitat development. The programme has conducted 802

a systematic Ultimate Burst Pressure (UBP) test campaign at NASA Marshall Space Flight 803

Center, producing two landmark results Sierra Space Corporation [2024]: 804

ˆ January 2024 (full-scale): A full-scale LIFE 285 expandable structure (approx- 805

imately 300 m3, over 6 m tall) burst at 77 psi (531 kPa), exceeding NASA's rec- 806

ommended threshold of 60.8 psi (4× the 15.2 psi maximum operating pressure per 807

NASA-STD-5001) by 27% Sierra Space Corporation [2024]. 808

ˆ OctoberNovember 2024 (1/3 scale): The LIFE 10 module burst at 255 psi 809

(1,758 kPa), achieving a factor of safety of 16× for LEO operations (at 15.2 psi) and 810

23× for lunar surface operations (at 10.8 psi) Sierra Space Corporation [2024]. 811

The LIFE product line spans three variants: LIFE 10 (∼100 m3 equivalent, 1/3 scale, 812

for lunar surface applications), LIFE 285 (∼300 m3, full-scale, for ISS-attached or free- 813

ying stations), and LIFE 500 (6001,440 m3, exceeding the total pressurised volume of the 814

ISS) Sierra Space Corporation [2024]. The restraint layer uses Vectran straps manufactured 815

by ILC Dover, the same organisation responsible for TransHab, Mars Exploration Rover, and 816

BEAM softgoods. Sierra Space is partnered with Blue Origin for the Orbital Reef commercial 817

space station, which received a $130M NASA Commercial LEO Destinations (CLD) award 818

in December 2021. An in-space test is targeted for no earlier than 2026. 819

4.2.2 Historical Context: B330 and Commercial Ecosystem Fragility 820

The history of Bigelow Aerospace provides a cautionary counterpoint. The B330 (330 m3, 821

18,50023,000 kg, 2436 layers totalling approximately 0.46 m wall thickness Valle et al. 822

[2019a]) was the most advanced commercial inatable habitat design as of 2019, with a full- 823

scale ground prototype (XBASE) tested under NASA's NextSTEP programme. The B330's 824

restraint design used a distinctive hoop webbing approach (US Patent 7,100,874) diering 825

from NASA's basket-weave architecture Valle et al. [2019a]. 826

Bigelow Aerospace ceased operations in March 2020 following COVID-19 layos, and 827

BEAM's ownership was transferred to NASA JSC in December 2021. The collapse of the 828

most mature commercial inatable habitat programme illustrates that high TRL does not 829

guarantee commercial viability. Future programmes cannot rely on government safety nets 830

to preserve technology investments, and the commercial ecosystem supporting inatable 831

habitat development remains fragile. 832

4.2.3 NextSTEP Competitive Landscape 833

NASA's NextSTEP-2 programme (20162019) selected six companiesBigelow, Boeing, 834

Lockheed Martin, Orbital ATK, Sierra Nevada Corporation, and NanoRacksto develop 835

habitat prototypes for evaluation NASA [2016]. Lockheed Martin's inatable prototype 836

achieved a burst pressure of 285 psi with hundreds of sensors and high-speed cameras mon- 837

itoring the failure Lockheed Martin [2022]. However, this programme subsequently pivoted: 838

the Starlab commercial station (originally Lockheed Martin/NanoRacks) adopted a rigid 839

architecture with Airbus as partner, abandoning the inatable approach. Of the six original 840

NextSTEP-2 companies, only Sierra Space (evolved from Sierra Nevada Corporation) has 841

continued to develop inatable habitats. This consolidation, combined with Bigelow's exit, 842

suggests that the inatable habitat technology faces unresolved commercialisation challenges 843

that complement the technical risks discussed elsewhere. 844

4.3 Future Concepts: Lunar Surface, Mars Transit, Planetary En- 845

try 846

4.3.1 Lunar Surface Habitats 847

Multiple concepts have been proposed for inatable habitats on the lunar surface, where 848

the reduced gravity (1/6 g) and absence of orbital debris shift the design requirements 849

from MMOD protection toward radiation shielding and dust management. The ESA-Hassell 850

collaboration has designed a scalable inatable pod system at the Shackleton Crater (lunar 851

south pole), partially constructed from lunar regolith via 3D printing and expandable to 852

house up to 144 people Hassell Studio and European Space Agency [2024]. The ESA-SOM 853

Moon Village concept proposes a semi-inatable shell that doubles its internal volume upon 854

deployment, supporting a four-person crew for up to 300 days Skidmore, Owings & Merrill 855

and European Space Agency [2019]. The ESA Pneumocell concept is specically designed 856

for burial under 45 m of regolith, using the lunar soil itself as radiation shielding European 857

Space Agency [2018]an elegant solution that leverages the inatable structure's compliance 858

to conform to the excavated cavity. 859

For lunar operations, the MMOD layer that constitutes approximately 68% of the shell 860

mass in LEO Valle et al. [2019a] can be substantially reduced or eliminated, oering signi- 861

cant mass savings. However, lunar dust intrusion and abrasion present a new challenge for 862

exible fabric surfaces that has not been addressed in any inatable habitat design to date. 863

4.3.2 Mars Transit and Surface Applications 864

TransHab was originally conceived as a Mars transit vehicle, and the deep-space habitat 865

architecture inherits directly from this heritage. Valle et al. Valle et al. [2019a] present a 866

launch-to-activation deployment owchart for a deep-space inatable habitat, identifying key 867

operational challenges: autonomous deployment without crew intervention, up to 4 kW of 868

heater power required post-ination to bring the bladder above minimum operating tem- 869

perature, and up to 24 hours before crew entry is permitted. For a three-year Mars transit 870

mission at solar minimum with three solar particle events (SPEs), radiation shielding re- 871

quirements range from 25 cm to 400 cm of water equivalent depending on the allowable bone 872

marrow dose Valle et al. [2019a]a signicant design driver discussed further in Section 4.4. 873

Mars surface applications extend to entry systems. The Low-Earth Orbit Flight Test of 874

an Inatable Decelerator (LOFTID, 2022) demonstrated a 6 m diameter inatable aerodecel- 875

erator at Mach 30 during orbital reentry NASA [2022], achieving TRL 78 and establishing 876

the viability of inatable heat shields for planetary entry. The Inatable Reentry Vehicle 877

Experiment (IRVE-II, 2009) had previously validated a 3 m prototype in suborbital ight Lit- 878

teken [2019]. For Mars, where the thin atmosphere limits the eectiveness of parachutes for 879

large payloads, inatable aerodecelerators oer the only viable path to landing human-scale 880

masses (>20 tonnes) on the surface. More exotic concepts include the HAVOC Venus air- 881

ship and the Titan Aerover blimp, both leveraging inatable structures for buoyancy-based 882

exploration Litteken [2019]. 883

4.3.3 European Programmes 884

European contributions to inatable habitat development include the ASI-funded FLECS 885

(Flexible Commercial Structure), the ESA-funded IHAB (Inatable Habitation) and IMOD 886

(Inatable Module) programmes, and the 2002 ESA/ESTEC First European Workshop 887

on Inatable Space Structures (ESA-WPP-200) ESA/ESTEC [2002]. These programmes 888

have contributed materials characterisation, hypervelocity impact testing of exible MMOD 889

shields (notably Destefanis et al. Destefanis et al. [2006]), and architectural concepts. How- 890

ever, it must be noted that no European inatable has own in a habitation role. After 891

more than two decades of investment, all European inatable habitat programmes remain at 892

TRL 24. The Volga airlock (1965) remains the only European-adjacent (Soviet-era) ight 893

precedent for a human-rated inatable in space. 894

4.4 Radiation Shielding: The BEAM SPE Findings and Design Im- 895

plications 896

Radiation shielding represents the single most serious unresolved technical challenge for 897

inatable habitats in deep space. The BEAM module has provided the only in-ight radiation 898

data for an inatable habitat, and the ndings demand honest assessment. 899

During the September 2017 solar particle event (SPE), radiation dosimeters inside BEAM 900

recorded approximately 22.5 mGy, compared to approximately 0.25 mGy measured in typ- 901

ical ISS metallic habitable modules during the same eventa ratio of 810× higher dose 902

inside the inatable module NASA Johnson Space Center [2017]. For galactic cosmic ra- 903

diation (GCR), which is continuous rather than episodic, BEAM dose rates were similar 904

to other ISS modules at baseline, indicating that the fabric shell provides adequate GCR 905

shielding in LEO where the Earth's magnetic eld supplies primary protection. 906

The SPE nding has signicant implications: 907

ˆ Fabric walls alone are insucient for SPE protection. The multi-layer shell 908

(60+ individual layers, 3050 cm total thickness) provides substantially less shielding 909

than the aluminium structure of ISS modules during particle events. 910

ˆ The mitigation is designed-in, not absent. Both TransHab and the LIFE archi- 911

tecture incorporate a rigid central core functioning as a storm shelter during SPEs. 912

Crew quarters are positioned within this core, surrounded by water wall containers 913

(a concept originating with Kennedy's TransHab design Kennedy [2002]) that provide 914

eective hydrogen-rich shielding. The inatable volume provides habitable space for 915

non-storm operations, while the rigid core provides radiation protection. 916

ˆ Material selection matters. Polyethylene provides 27.8% mass savings compared to 917

aluminium for equivalent radiation shielding eectiveness, and three-layer composite 918

shields (combining high-Z, medium-Z, and low-Z materials) achieve up to 70% total 919

ionising dose improvement for electrons and 50% for protons Norbury et al. [2025]. 920

For deep-space missions beyond Earth's magnetosphere, the GCR environment is more 921

severe and continuous. Valle et al. Valle et al. [2019a] model that a three-year deep-space 922

mission at solar minimum with three SPEs requires between 25 cm and 400 cm of water- 923

equivalent shielding depending on the allowable bone marrow dosetranslating to substan- 924

tial mass within the rigid core. Active magnetic shielding and pharmaceutical countermea- 925

sures remain at low TRL and are not viable near-term solutions. 926

The honest framing is that inatable habitats are not radiation protection structures, 927

and were never designed to be. They are mass-ecient volume structures with integrated 928

MMOD protection. Radiation protection is the responsibility of the rigid core and water wall 929

architecture. The BEAM SPE data conrms this design philosophy rather than undermining 930

it, but the data must be presented without minimisation to maintain credibility with the 931

radiation protection community. The absence of post-2017 follow-up publications detailing 932

BEAM's continued radiation environment data over its now eight-year mission represents a 933

gap in the available evidence base that future studies should address. 934

5 State of the Art: Materials and Structures 935

The material systems underpinning inatable space structures occupy a unique design space: 936

they must combine the tensile strength of structural metals, the exibility to package into 937

compact launch volumes, and the environmental durability to survive atomic oxygen, ultra- 938

violet radiation, and micrometeoroid impacts for mission lifetimes spanning years to decades. 939

This section reviews the four dominant fabric families, the canonical multi-layer shell archi- 940

tecture derived from TransHab, established rigidisation technologies, and the environmental 941

degradation mechanisms that govern long-term performance. 942

5.1 Space-Rated Fabrics: Vectran, Kevlar, Zylon, Nextel 943

Four high-performance fabric families dominate inatable space structure design, each oc- 944

cupying a distinct functional niche determined by the intersection of mechanical properties, 945

environmental tolerance, and ight heritage. 946

Vectran HT (liquid crystal polymer, Kuraray Co.) has emerged as the preferred ma- 947

terial for restraint layers in inatable habitats. With a tensile strength of approximately 948

3.0 GPa at a density of 1.40 g/cm3, Vectran achieves a specic strength of 2,330 kN·m/kg 949

an order of magnitude above Ti-6Al-4V (220 kN·m/kg) and Al 7075 (204 kN·m/kg) Valle 950

et al. [2019b]. Vectran's principal advantage over the earlier-generation Kevlar is its superior 951

creep resistance: under sustained load at the NASA-mandated factor of safety of 4.0 (corre- 952

sponding to 25% of ultimate tensile strength), Vectran fabric exhibits no failure over extended 953

test periods of months Weadon [2013]. This characteristic is critical because creep is the life- 954

limiting mechanism for restraint layers in pressure-stabilised structures. However, Weadon's 955

systematic characterisation revealed that time-to-failure is exponentially sensitive to load 956

level, and manufacturing variability in ultimate tensile strength (±10% for 12K webbing, 957

±6% for 6K webbing) introduces signicant uncertainty in lifetime predictionat 7585% 958

UTS, time-to-failure ranges from 4 minutes to 5.5 months for identical test congurations 959

Weadon [2013]. This nding underscores the importance of quality control in inatable 960

habitat fabrication. Two important qualications must be noted. First, Weadon's creep 961

characterisation was conducted at room temperature; no published Vectran creep dataset 962

exists for space-representative thermal cycling conditions (approximately −100◦C to +120◦C 963

for LEO), and the eective creep rate under such cycling may dier signicantly from room- 964

temperature datathis represents a critical materials gap for habitat lifetime prediction. 965

Second, the no failure over extended test periods result at 25% UTS, while encouraging, is 966

based on a limited number of specimens at the design operating point; given the wide man- 967

ufacturing variability, condence intervals on lifetime prediction remain large, and the creep 968

behaviour exhibits bimodal characteristics where some specimens show substantially earlier 969

failure than others at identical load levels. Vectran's ight heritage includes Mars Pathnder 970

airbags (1997), BEAM restraint layers (2016present), and the Sierra Space LIFE program 971

Litteken [2019]. 972

Kevlar 49 (poly-paraphenylene terephthalamide, DuPont) was the original restraint 973

layer material for TransHab, with a tensile strength of approximately 3.0 GPa at the fabric 974

level and 3.6 GPa at the individual lament level, at a density of 1.44 g/cm3 Valle et al. 975

[2019b], DuPont [2019]. The corresponding specic strength is 2,080 kN·m/kg (fabric) or 976

2,500 kN·m/kg (lament); throughout this survey, fabric-level properties are reported un- 977

less otherwise noted, as these are the engineering-relevant values for woven restraint layers. 978

While Kevlar's fabric-level specic strength is comparable to Vectran's, its higher creep rate 979

under sustained biaxial loading led to its replacement by Vectran in subsequent habitat de- 980

signs Kennedy [2002]. Kevlar retains an important role as a rear-wall material in multi-layer 981

micrometeoroid and orbital debris (MMOD) shields, where its combination of high energy 982

absorption and relatively low cost makes it the material of choice for fragment capture layers 983

Destefanis et al. [2003]. Space environment characterisation by Destefanis et al. conrmed 984

that Kevlar suers UV-induced discoloration and embrittlement but shows acceptable perfor- 985

mance when shielded from direct solar exposure within the MMOD sub-assembly Destefanis 986

et al. [2009]. 987

Zylon (poly-p-phenylene-2,6-benzobisoxazole, PBO; Toyobo Co.) oers the highest ten- 988

sile strength of any commercially available high-performance bre at 5.8 GPa, yielding a 989

specic strength of 3,840 kN·m/kg Toyobo Co., Ltd. [2005]. However, Zylon exhibits catas- 990

trophic UV degradation: strength loss of approximately 35% within 6 months of unshielded 991

exposure, rendering it unsuitable for any application without comprehensive UV protec- 992

tion Toyobo Co., Ltd. [2005], Said et al. [2006]. Despite this limitation, Zylon has found 993

niche space applications where UV shielding is inherently provided: SpaceX Crew Dragon 994

parachute risers and NASA high-altitude balloon tendons Litteken [2019]. For inatable 995

structures, Zylon could serve in interior tensile elements (e.g., oor suspension webbings 996

within pressurised habitats) where the multi-layer shell provides UV shielding, but its UV 997

sensitivity eectively precludes use in any externally exposed role. 998

Nextel 440 (3M alumina-boria-silica ceramic fabric) occupies a unique position as the 999

only ceramic bre used in inatable space structures. With a density of 3.05 g/cm3 and con- 1000

tinuous use temperature of 1370◦C, Nextel is employed exclusively as the outer bumper layer 1001

in MMOD shielding Christiansen and Davis [2019], Destefanis et al. [2003]. Upon hyperveloc- 1002

ity impact, Nextel fragments incoming particles into smaller, more widely dispersed debris, 1003

reducing the energy density impinging on subsequent shield layers. The stued Whipple 1004

conguration (Nextel bumper + open-cell foam + Kevlar rear wall) protects against projec- 1005

tiles approximately twice the diameter of those defeated by a standard aluminium Whipple 1006

shield at equal areal density Destefanis et al. [2003]. Nextel is inherently immune to UV and 1007

atomic oxygen degradation due to its ceramic composition, but its high density limits its use 1008

to the thin bumper layer. 1009

Two additional materials complete the palette for inatable structures. Beta cloth 1010

(PTFE-coated breglass) serves as the outermost atomic oxygen protection cover layer, with 1011

LDEF ight data demonstrating excellent durability over 68 months of LEO exposure Linton 1012

et al. [1993], Banks et al. [2004]. Kapton H (polyimide, DuPont) is the workhorse lm for 1013

multi-layer insulation, operating from −269◦C to +400◦C, though it is susceptible to atomic 1014

oxygen erosion at a rate of 3.0×10−24 cm3/atom Banks et al. [2004], Finckenor and Dooling 1015

[1999]. 1016

Table 8 presents a comprehensive comparison of these material systems across eight 1017

performance parameters relevant to inatable space structures. 1018

Table 8: Comparison of space-rated materials for inatable structures.

Material Type σUTS ρ Tmax UV AO Primary TRL (GPa) (g/cm3) (◦C) Sens. Resist. Role

Vectran HT LCP bre 3.0 1.40 330 Mod. Low Restraint 9 Kevlar 49 Aramid 3.0 1.44 427 High Low MMOD rear 9 Zylon AS PBO bre 5.8 1.54 650 V. High Low Interior only 7 Nextel 440 Ceramic  3.05 1370 None N/A MMOD bumper 9 Kapton H Polyimide 0.23 1.42 400 Low Low MLI layers 9 Beta cloth PTFE/glass 0.34  650 Low High AO cover 9

Table 9: Specic strength comparison: high-performance fabrics versus structural metals (data from Valle et al. Valle et al. [2019b]).

Material σUTS (GPa) ρ (g/cm3) Specic Strength (kN·m/kg) Ratio to Ti-6Al-4V

Zylon AS 5.8 1.54 3,840 17.5× Kevlar 49 (fabric) 3.0 1.44 2,080 9.5× Vectran HT 3.0 1.40 2,330 10.6× Ti-6Al-4V 0.95 4.43 220 1.0× Al 7075-T6 0.57 2.81 204 0.9×

5.2 Multi-Layer Shell Architecture 1019

The TransHab program (19972000) established the canonical ve-layer shell architecture 1020

that remains the reference design for all subsequent inatable habitats Kennedy [2002, 2016]. 1021

From innermost to outermost, the layers are: 1022

1. Liner: Nomex fabric backed by Kevlar felt provides the crew-contact interior surface, 1023

oering acoustic attenuation and a substrate for equipment mounting. 1024

2. Bladder: Three redundant layers of polymeric gas barrier (Combitherm or urethane- 1025

coated Nylon), each sandwiched between Kevlar felt separators. The bladder is deliber- 1026

ately oversized relative to the restraint layer so that it carries no structural loadthe 1027

positive pressure dierential is transmitted entirely to the restraint layer Kennedy 1028

[2016]. The triple redundancy ensures continued pressure containment after a single- 1029

layer puncture. 1030

3. Restraint layer: The primary load-carrying element, comprising Kevlar (TransHab) 1031

or Vectran (BEAM and subsequent designs) in a biaxial basket-weave conguration. 1032

TransHab's restraint layer was designed to sustain 12,500 lb/in hoop loading and 1033

High-performance

High-perf. fabric Structural metal Ceramic Polymer film

fabrics

Zylon AS

Kevlar 49

Specific strength (kN·m/kg)

Vectran HT

103

Ceramics

Nextel 440

Ti-6Al-4V

Al 7075-T6 Kapton H

Structural metals

102

200 400 600 800 1000 1200 1400 1600 Maximum service temperature (°C)

Figure 4: Materials Ashby chart comparing specic strength versus maximum service tem- perature for space-rated fabrics and structural metals. High-performance fabrics (Vec- tran, Kevlar, Zylon) occupy a design space inaccessible to metals, combining an order- of-magnitude advantage in specic strength with adequate thermal performance for LEO applications.

6,000 lb/in axial loading at a factor of safety of 4.0 per NASA-STD-5001 Kennedy 1034

[2016]. The restraint layer attaches to rigid bulkheads via clevis ttings that transfer 1035

membrane loads to the metallic core structure. Ground testing demonstrated sustained 1036

pressure at 4× operating pressure (60 psid) without failure, and burst at 196 psid in 1037

sub-scale articles Kennedy [2002]. 1038

4. MMOD shield: A stued Whipple conguration comprising Nextel 440 ceramic fab- 1039

ric bumper layers, open-cell polyurethane foam spacers, and Kevlar rear walls Deste- 1040

fanis et al. [2003]. The MMOD assembly is vacuum-packed during launch to maintain 1041

the folded conguration and expands passively on orbit when exposed to vacuum. 1042

TransHab's MMOD design was tested against projectiles up to 1.7 cm diameter at 1043

hypervelocity, meeting the no-penetration probability requirement of PNP ≥0.9820 1044

Kennedy [2002]. Damage tolerance testing by Trevino et al. demonstrated that a 2 in 1045

× 3.5 in hole in the restraint layer at 25% of burst pressure resulted in load redistri- 1046

bution without catastrophic failurean inherent advantage of woven textile structures 1047

over metallic shells Edgecombe et al. [2009]. 1048

5. Thermal protection system (TPS): Multi-layer insulation comprising nylon-reinforced 1049

double-aluminized Mylar and double-aluminized Kapton layers, with inner layers perfo- 1050

rated for gas venting during deployment Finckenor and Dooling [1999]. The outermost 1051

element is an atomic oxygen cover of Beta glass fabric for LEO operations Kennedy 1052

[2016]. Eective emittance values for properly installed MLI range from 0.015 to 0.05, 1053

though practical performance with seams, penetrations, and attachment hardware typ- 1054

ically falls at the upper end of this range Finckenor and Dooling [1999], Gilmore [2002]. 1055

The total shell assembly comprises 60+ individual layers deployed to a thickness of 30 1056

50 cm Valle et al. [2019b]. For TransHab, the overall packaged dimensions were 10.5 m length 1057

with a deployed width of 8.3 m, yielding an internal habitable volume of approximately 1058

161 m3 and a total packaged shell volume of 329 m3 Kennedy [2016]. BEAM, the ight- 1059

demonstrated derivative, achieves a habitable volume of 16 m3 in a 1,415 kg module Valle 1060

et al. [2019b]. 1061

Table 10: Layer-by-layer specication of the TransHab/BEAM shell architecture. The her- itage convention identies ve functional sub-assemblies; the AO cover (Beta cloth) is the outermost element of the TPS sub-assembly but is listed separately here for clarity, yielding six table rows for ve sub-assemblies.

Sub-assy Layer Material(s) Function Key Specica

1 Liner Nomex + Kevlar felt Crew contact, acoustic Non-structural 2 Bladder (×3) Combitherm / Urethane-Nylon Gas barrier 3× redundant, 3 Restraint Vectran basket-weave Primary structure FOS = 4.0, 12 4 MMOD Nextel + foam + Kevlar Debris protection PNP ≥0.9820

5 TPS/MLI Aluminized Mylar/Kapton Thermal control εe = 0.0150.0 AO cover Beta glass fabric AO protection (outer TPS) LDEF-validate

TransHab / BEAM Shell Architecture

(five functional sub-assemblies, not to scale)

Exterior (space environment)

AO protection cover

(Beta cloth) Atomic oxygen barrier

4. Thermal Protection

MLI blankets (Kapton + Dacron) Thermal insulation

Sub-assembly

MMOD: Nextel + Kevlar

(Stuffed Whipple) Projectile fragmentation

3. MMOD Shield

Sub-assembly

MMOD: Open-cell foam

(Solimide) Energy absorption

Restraint layer (Vectran webbing) Primary structural element

2. Restraint Sub-assembly

Bladder (Combitherm film) Gas barrier

Liner (Nomex/Kevlar felt) Crew-contact surface

1. Liner

Interior (pressurised)

Figure 5: TransHab/BEAM multi-layer shell architecture, showing the ve functional sub- assemblies from the crew-contact liner (innermost) to the atomic oxygen protection cover (outermost). The restraint layer (Vectran basket-weave) carries all pressure loads; the blad- der, MMOD shield, and thermal protection system are non-structural. Total deployed thick- ness: 3050 cm; total number of individual layers: 60+.

5.3 Rigidization Technologies 1062

While habitats remain pressure-stabilised throughout their operational life (at a factor 1063

of safety of 4.0), many inatable componentsparticularly booms, masts, and structural 1064

supportsrequire rigidisation after deployment to eliminate dependence on continued gas 1065

containment. Cadogan and Scarborough established the canonical classication of rigidisa- 1066

tion technologies into three families Cadogan and Scarborough [2001]: 1067

Mechanical (strain hardening): Aluminum-polymer laminates (e.g., 14.5 µm Al / 1068

16 µm Mylar / 14.5 µm Al) undergo plastic deformation during ination, work-hardening 1069

the aluminium layers and locking the deployed shape Schenk et al. [2014]. This approach 1070

has the longest ight heritage, from Echo 2 (1964) through InateSail (2017), where a 1 m 1071

strain-rigidized mast achieved deployment in approximately 2 seconds via CO2 pressurization 1072

Viquerat et al. [2019], Underwood et al. [2017]. Lenticular boom cross-sections achieve 1073

packaging ratios of approximately 10:1, while circular cross-sections achieve approximately 1074

5:1 under z-fold Schenk et al. [2014]. Current TRL: 89. 1075

Physical (sub-Tg and shape memory): Resin-impregnated composites heated above 1076

their glass transition temperature (Tg) become pliable for packaging; upon deployment and 1077

cooling below Tg in the space thermal environment, the resin solidies and rigidizes the struc- 1078

ture Cadogan and Scarborough [2001], Defoort et al. [2005]. This approach is reversible in 1079

principle, enabling re-stowage. Shape memory polymers extend this concept with engineered 1080

Tg transitions. Current TRL: 45. 1081

Chemical (UV-curable): Cationic epoxy resins cure upon exposure to solar UV radi- 1082

ation, achieving the highest post-rigidisation stiness of the three approaches Allred et al. 1083

[2002]. The Rigidization on Command (ROC) technology demonstrated by Adherent Tech- 1084

nologies achieves mechanical properties equivalent to thermally cured composites using sun- 1085

light alone Adherent Technologies Inc. [2001]. However, UV curing requires unobstructed 1086

solar access and is sensitive to shadowing by other spacecraft elements. Current TRL: 45. 1087

An emerging fourth approach uses shape memory alloy (SMA) elements integrated into 1088

inatable toroidal structures. Patel et al. developed an analytical framework for SMA-based 1089

rigidisation where NiTi alloy wires, embedded in the inatable wall and heated above their 1090

austenite nish temperature, contract and lock the deployed geometry Rastogi et al. [2024]. 1091

This approach remains at the analytical stage (TRL 23) but oers the potential for active 1092

shape control during rigidisation. 1093

Table 11: Rigidization technology comparison for inatable space structures.

Method Mechanism TRL Heritage Best Application

Strain hardening Al-polymer plastic deformation 89 Echo 2, InateSail Thin booms, sails Sub-Tg resin Glass transition solidication 45 Ground demos Structural booms UV curing Solar-initiated polymerization 45 Ground demos Max. stiness booms SMA rigidisation Thermoelastic contraction 23 Analytical only Toroidal structures

A critical distinction: large inatable habitats (BEAM, TransHab, LIFE) do not em- 1094

ploy rigidisation. They remain pressure-stabilised structures throughout their operational 1095

life, relying on the continuous pressure dierential across the multi-layer shell to maintain 1096

structural integrity at a factor of safety of 4.0 Valle et al. [2019b]. Rigidization is primar- 1097

ily relevant for booms, masts, and structural supports where prolonged gas containment is 1098

impractical or where a loss-of-pressure failure mode is unacceptable. 1099

5.4 Environmental Degradation: AO, UV, Radiation, Creep 1100

Four environmental mechanisms govern the long-term performance of inatable structures 1101

in the space environment, each aecting dierent layers of the shell assembly. 1102

Atomic oxygen (AO) is the dominant surface degradation threat in LEO. At ISS 1103

altitude (∼400 km), AO ux is approximately 1015 atoms/cm2/s, and Kapton H exhibits 1104

an erosion yield (Ey) of 3.0 × 10−24 cm3/atomthe practical erosion rate (thickness loss 1105

per unit time) is Ey × Φ, where Φ is the AO ux, which varies with altitude, solar activity, 1106

and ram direction; at ISS altitude this corresponds to approximately 1 µm/year Banks 1107

et al. [2004]. Unprotected Mylar, Kevlar, and Vectran all exhibit comparable erosion rates. 1108

SiO2 coatings reduce Kapton erosion by 23 orders of magnitude, and novel AO-resistant 1109

polymers (TOR, COR) developed at NASA Glenn demonstrate near-zero erosion Banks 1110

et al. [2004]. In practice, inatable habitats are protected by the outermost Beta cloth 1111

layer, which is inherently AO-resistant due to its PTFE coating. In-situ measurements from 1112

JAXA's SLATS satellite (160560 km altitude range) have recently provided direct on-orbit 1113

validation of erosion models Verker et al. [2023]. 1114

UV degradation primarily aects Kevlar (discoloration and embrittlement) and Zylon 1115

(catastrophic strength loss of ∼35% in 6 months) Destefanis et al. [2009], Toyobo Co., Ltd. 1116

[2005]. Vectran shows moderate UV sensitivity. The multi-layer shell architecture naturally 1117

provides UV shielding for interior layers, but any externally exposed fabric elements require 1118

dedicated UV protection. 1119

Radiation eects on high-performance fabrics are comparatively modest for LEO mis- 1120

sions. The primary radiation concern for inatable habitats is crew dose rather than material 1121

degradationBEAM measurements during a September 2017 solar particle event recorded 1122

22.5 mGy inside BEAM versus approximately 0.25 mGy in adjacent ISS metallic modules, 1123

an 810× ratio attributable to the lower areal density of the fabric shell NASA Johnson Space 1124

Center [2017]. Polyethylene supplemental shielding oers 27.8% mass savings over equivalent 1125

aluminium shielding, and multi-layer congurations achieve up to 70% total ionizing dose 1126

improvement for electrons and 50% for protons Norbury et al. [2025]. 1127

Creep is the life-limiting mechanism for Vectran and Kevlar restraint layers under sus- 1128

tained biaxial pressure loading. Weadon's characterisation demonstrated three-stage vis- 1129

coelastic creep with exponential sensitivity to the ratio of applied load to ultimate tensile 1130

strength Weadon [2013]. At the design operating point of 25% UTS (FOS = 4.0), specimens 1131

showed no failure over test periods of months. However, the wide manufacturing variability 1132

in UTS (±10%) dominates lifetime uncertaintynot the average material properties them- 1133

selves. Combined synergistic eects (AO + UV + thermal cycling + sustained load) remain 1134

inadequately characterised, representing a research gap that limits condence in multi-decade 1135

lifetime predictions for deep-space habitats Zhai et al. [2023]. 1136

6 State of the Art: Deployment Mechanics 1137

The deployment of inatable structures in the space environment presents a unique engineer- 1138

ing challenge: a large, compliant membrane must transition from a compactly folded launch 1139

conguration to a precise deployed geometry under vacuum conditions where gas dynamics, 1140

thermal gradients, and material memory eects all inuence the nal state. This section 1141

reviews fold pattern selection, ination control strategies, and lessons from ight heritage. 1142

6.1 Fold Patterns and Packaging Eciency 1143

The choice of fold pattern determines deployment reliability, packaging eciency, and the 1144

number of actuators required for controlled deployment. Three primary pattern families are 1145

employed, each optimised for a dierent structural geometry. 1146

Miura-ori Miura [1985] is the foundational pattern for at membrane deployment. The 1147

tessellation of parallelogram facets creates a one-degree-of-freedom rigid-foldable mecha- 1148

nism: the entire membrane deploys via a single actuator force without requiring elastic 1149

deformation of the panels. This property is critical for fragile thin lms (metallized My- 1150

lar, ceramic-coated Kapton) that cannot sustain repeated fold stress. The negative Pois- 1151

son's ratio characteristiccontraction in one direction when extended in the perpendicular 1152

directionassists controlled deployment by preventing bunching Miura [1985]. Compaction 1153

is theoretically unlimited: an N × M panel array compacts to a stack of 2 panels thick, 1154

achieving compaction ratios of N/2 in each direction. Miura-ori is optimal for solar sails, 1155

antenna reectors, and drag sails where at-membrane deployment is required. 1156

For cylindrical structures (booms, masts), Schenk and Guest adapted the Miura- 1157

ori pattern to cylindrical geometry, enabling origami-based compaction of inatable booms 1158

with geometrically determined deployment kinematics Schenk et al. [2013, 2014]. The z-fold 1159

variant oers the simplest implementation and highest packaging ratio but lower deployment 1160

reliability, as individual folds must sequentially release without jamming. Wrapping (coiling) 1161

provides more controlled deployment at lower packaging ratios. The lenticular boom cross- 1162

section achieves ∼10:1 packaging ratios versus ∼5:1 for circular cross-sections Schenk et al. 1163

[2014]. 1164

For habitats, a 7-gore S-fold approach is employed: the bladder and restraint layers 1165

are folded in an S-pattern around the rigid central core, with individual MMOD and MLI 1166

gore panels attached separately Kennedy [2016], Valle et al. [2019b]. The habitat packaging 1167

ratio is substantially lower than for membranes or booms because the rigid core occupies a 1168

signicant fraction of the stowed volume. TransHab achieved a stowed-to-deployed volume 1169

ratio of approximately 2.1:1 (habitable volume), while BEAM achieves approximately 4.4:1 1170

(16 m3 deployed / ∼3.6 m3 stowed) Valle et al. [2019b]. 1171

6.2 Ination Sequencing and Control 1172

Ination rate control is critical for successful deployment: ination that is too rapid generates 1173

shock waves in the gas column that can damage thin lms and cause asymmetric expansion, 1174

while ination that is too slow allows thermal gradients to develop that aect the nal 1175

geometry Jenkins [2001]. Minimum tension requirements must be maintained throughout 1176

BEAM (habitat)

400:1

TransHab

340:1

(habitat)

LOFTID (aeroshell)

Inflatable

180:1

InflateSail (sail boom)

150:1

PowerSphere

120:1

(power)

ROSA (solar array)

20:1

TRAC boom

50:1

(structural)

Rigid deployable

Mesh reflector

30:1

(antenna)

Coilable mast

15:1

(structural)

Hinged panel

8:1

(solar array)

0 100 200 300 400 Deployed-to-stowed volume ratio

Figure 6: Deployed-to-stowed volume ratio comparison between inatable and rigid deploy- able structures. Inatable systems achieve packaging ratios an order of magnitude higher than rigid deployable alternatives, with BEAM demonstrating a 400:1 ratio. Data compiled from mission documentation and manufacturer specications.

Table 12: Packaging eciency by structure type for inatable space systems.

Structure Type Fold Pattern Packaging Ratio Heritage Example

Flat membrane (sail) Miura-ori / z-fold ∼500:1 (membrane) InateSail (10 m2) Boom (lenticular) Origami / coil ∼10:1 InateSail (1 m boom) Boom (circular) z-fold ∼5:1 Various CubeSat booms Habitat (with rigid core) 7-gore S-fold 25:1 BEAM (∼4.4:1), TransHab Origami shield Waterbomb tessellation ∼5:1 (80% expansion) IMSS concept Cha et al. [20

ination to prevent wrinkling, which can create permanent creases in metallized lms and 1177

compromise thermal or RF performance. 1178

The BEAM deployment sequence provides the most instructive ight data on ination 1179

control challenges. Initial deployment in May 2016 failed to expand BEAM beyond a small 1180

fraction of its intended volume. Over the following 7 hours, mission controllers executed 1181

25 sequential pressure bursts, each providing a small increment of expansion, before BEAM 1182

reached its full deployed geometry NASA Johnson Space Center [2017]. This arduous recov- 1183

ery illustrates a fundamental tension: the folded softgoods develop stronger memory eects 1184

during extended stowed periods than ground testing predicted, requiring more expansion 1185

energy than designed. For autonomous missions (lunar surface habitats, Mars transit mod- 1186

ules), such manual intervention is not viable, and deployment reliability must be established 1187

at substantially higher condence levels Valle et al. [2019b]. 1188

Several ination methodologies have been demonstrated or proposed. Stored gas (typi- 1189

cally CO2 or N2) provides the most controllable ination but requires tanks, regulators, and 1190

plumbing that add mass and failure modes. InateSail used a cold-gas CO2 system for boom 1191

deployment Viquerat et al. [2019]. Sublimation-based ination eliminates gas handling 1192

hardware: benzoic acid or naphthalene powder generates sucient vapour pressure at ambi- 1193

ent space temperatures to inate simple structures, though residual air in the packed struc- 1194

ture can cause premature partial ination Horn [2017]. The PowerSphere concept employed 1195

passive vapour-pressure ination from sublimation powder for a multifunctional sphere Cado- 1196

gan et al. [2006]. Active pressure control using real-time pressure-volume feedback with 1197

variable ination rates has been studied analytically by Wei et al., who demonstrated that 1198

instantaneous optimal control of ination rate can minimise deployment loads and improve 1199

nal shape accuracy Li et al. [2022a]. 1200

6.3 Flight Heritage: InateSail, LOFTID, BEAM Deployment Lessons 1201

Three ight demonstrations provide the primary deployment heritage for inatable struc- 1202

tures, each operating at a dierent scale and in a dierent deployment regime. 1203

InateSail (2017) demonstrated the most compact packaging and fastest deployment: a 1204

1 m aluminium-Mylar laminate boom (14.5 µm Al / 16 µm Mylar / 14.5 µm Al) and 10 m2 1205

aluminized Mylar drag sail packaged into a 0.5U volume (∼50 mm cube), deploying and 1206

strain-rigidizing in approximately 2 seconds via CO2 pressurization Viquerat et al. [2019]. 1207

The deployed membrane-to-stowed volume ratio of approximately 500:1 represents the high- 1208

est documented packaging eciency for a complete deployable system. InateSail de-orbited 1209

from 505 km in 72 days, compared to an estimated 4+ years without the sail, validating the 1210

drag deorbit concept at TRL 89 Viquerat et al. [2019]. 1211

IRVE-3 (Inatable Reentry Vehicle Experiment, 2012) demonstrated a 3 m diameter in- 1212

atable aeroshell surviving Mach 10 reentry with peak heating of 14.4 W/cm2 Hughes et al. 1213

[2005]. Its successor, LOFTID (Low-Earth Orbit Flight Test of an Inatable Decelerator, 1214

2022), scaled this concept to 6 m diameter and survived Mach 30 reentry, achieving TRL 89 1215

for inatable aerodynamic decelerators. These demonstrations establish the thermal protec- 1216

tion performance of exible fabric systems under extreme heating conditions, conrming that 1217

multi-layer woven ceramic and polymer fabrics can provide thermal protection comparable 1218

to rigid ablative shields at a fraction of the mass. 1219

BEAM (2016present) provides the denitive deployment lesson for large pressurised 1220

habitats. Beyond the 25-burst recovery described above, BEAM demonstrated that pack- 1221

aged softgoods develop adhesion between layers during extended stowage that signicantly in- 1222

creases deployment energy requirements NASA Johnson Space Center [2017]. Post-deployment, 1223

thermal performance was more benign than predicted because folded softgoods create addi- 1224

tional insulation beyond the designed MLI performance. BEAM has now operated on ISS for 1225

over 8 years, providing the most extensive in-service data for any inatable habitat. These 1226

deployment lessons directly inform the design of future autonomous systems: residual fold 1227

adhesion must be characterised and accounted for, deployment energy budgets must include 1228

substantial margin, and passive deployment mechanisms (sublimation, spring) may be more 1229

reliable than active pressurization for autonomous operations. 1230

6.4 Comparison with Rigid Deployable Alternatives 1231

The survey's thesisthat inatables oer advantages over rigid systemsrequires adequate 1232

characterisation of the rigid deployable baseline. Three competing technology classes merit 1233

explicit comparison. 1234

Composite booms (e.g., CFRP bi-stable tape springs, Triangular Rollable and Col- 1235

lapsible (TRAC) booms) achieve packaging ratios exceeding 50:1 and are ight-proven at 1236

TRL 9 Murphey et al. [2015], Banik and Murphey [2010]. The TRAC boom, used on 1237

LightSail-2 and the Aeroboom Innovative Mechanism (AIM), provides high deployed stiness 1238

with no ination requirement. Sickinger and Herbeck Sickinger and Herbeck [2004] charac- 1239

terised CFRP boom deployment for solar sails, demonstrating that non-inatable composite 1240

booms are the dominant competing technology for CubeSat-class deployables. 1241

Mesh reector antennas (e.g., Harris/L3Harris AstroMesh, 1222 m deployed diam- 1242

eter, TRL 9) achieve large deployed apertures through cable-net tensioned mesh without 1243

requiring ination Santiago-Prowald and Rodrigues [2018]. These are the primary competi- 1244

tor to inatable antenna concepts and represent the state of the art for deployable high-gain 1245

antennas. 1246

Mechanically hinged trusses (e.g., NASA Langley's Compact Telescoping Array, 1247

CIRAS) provide high stiness and precise geometry through articulated rigid elements, at 1248

the cost of higher mass and complexity compared to inatable deployment. 1249

Table 13 presents a comparative assessment. 1250

Table 13: Comparison of inatable and rigid deployable technologies.

Technology Pkg Ratio Deployed Sti. Mass/m TRL Key Limitation

TRAC composite boom 50100:1 High Low 9 Length <10 m AstroMesh reector 1020:1 High Medium 9 Complex cable-net Mech. hinged truss 310:1 Very high High 9 Mass, complexity Inatable boom (Al-lam.) 510:1 Med. (post-rigid.) Very low 89 Rigidisation req'd Inatable membrane 100500:1 Low (press.-stab.) Very low 79 Pressure maint.

The inatable approach oers its greatest advantage at the largest scales (>10 m), where 1251

composite boom stiness-to-length scaling becomes unfavourable and mesh reector cable- 1252

net complexity grows prohibitively. For CubeSat-class deployables (<3 m), TRAC booms 1253

are the dominant technology; for medium-scale antennas (522 m), mesh reectors compete 1254

strongly. Inatables become uniquely enabling above approximately 30 m, where no rigid 1255

deployable alternative exists at acceptable mass. 1256

7 State of the Art: Actuation for Soft Space Systems 1257

The space environment imposes four principal constraints on actuator selection for soft in- 1258

atable systems: (1) ultrahigh vacuum eliminates ambient pressure support for unsealed 1259

pneumatic systems; (2) extreme temperature cycling (−150◦C to +120◦C in LEO) chal- 1260

lenges elastomers, smart materials, and ionic actuators; (3) high-energy particle and UV 1261

radiation degrades polymers, electrodes, and electrolytes; and (4) the absence of conven- 1262

tional lubricants eliminates standard gearing options. Against this backdrop, research has 1263

converged on several non-pneumatic actuation principles. This section reviews six technol- 1264

ogy families, organised from highest space-mission specicity to most novel, and presents a 1265

comparative assessment for inatable system integration. 1266

7.1 Dielectric Elastomer Actuators and DEMES 1267

Dielectric Elastomer Actuators (DEAs) convert high-voltage electrical input into mechanical 1268

deformation of a thin elastomer membrane sandwiched between compliant electrodes. Di- 1269

electric Elastomer Minimum Energy Structures (DEMES) extend this principle by bonding a 1270

pre-stretched DEA membrane to a exible frame, creating a self-deploying bending actuator 1271

that rolls compactly for stowage Araromi et al. [2014, 2015]. 1272

The most mission-specic DEA application is the DEMES gripper developed by Araromi et al. 1273

for ESA's CleanSpace One microsatellite, targeting the 820 g SwissCube CubeSat for active 1274

debris removal Araromi et al. [2014]. The four-arm gripper achieves the following specica- 1275

tions: mass less than 0.65 g per arm, tip angle change of approximately 60◦, gripping force 1276

of 0.8 mN at 5 mm deection (up to 2.2 mN in optimised frame variants), and over 860,000 1277

actuation cycles at 1 Hz and 2000 V without degradation. The actuator stores rolled to a 1278

14 mm diameter cylinder and deploys by burning a retaining Nylon wire. A mechanically 1279

elegant property emerges from the force-displacement characteristic: grip force increases as 1280

the target drifts away from the actuator tip, creating a passive negative feedback loop that 1281

enhances capture stability without active control Araromi et al. [2014]. 1282

Li et al. subsequently extended the 2D DEMES concept to a three-dimensional congura- 1283

tion specically designed for on-orbit servicing, enabling triaxial manipulation of irregularly 1284

shaped targets Liang et al. [2023]. The 3D conguration achieves higher load capacity and 1285

more favorable specic force output than planar DEMES. 1286

The critical limitation of DEA/DEMES for space applications is force output: the sub- 1287

millinewton to millinewton range, while sucient for microgravity contact-only operations 1288

on CubeSat-class targets, is inadequate for structural loads or capture of debris exceeding 1289

a few kilograms. DEA membranes (PDMS, acrylic) are also vulnerable to outgassing in 1290

vacuum and UV degradation, though neither has been systematically quantied under space 1291

conditionsa notable gap. 1292

7.2 Vacuum-Gap Electrostatic Actuators: Vacuum as Enabler 1293

A paradigm-shifting development emerged in 2025 with Sîrbu et al.'s introduction of vacuum- 1294

gap electrostatic multilayer actuators Sirbu et al. [2025]. These devices use thin-lm polymer 1295

multilayer structures enclosing vacuum gaps that zip closed upon electrical activationa 1296

mechanism that fundamentally benets from, rather than suers from, the space vacuum. 1297

In terrestrial operation, the vacuum gaps must be maintained against atmospheric pressure; 1298

in space, the ambient ultrahigh vacuum (∼10−7 Pa in LEO) is the default state. 1299

The performance specications represent a qualitative advance over existing soft actuator 1300

technologies: actuators weighing 0.7 g deliver forces exceeding 4 N, operate at bandwidths 1301

above 100 Hz, and achieve specic power of 1.4 kW/kg Sirbu et al. [2025]. For comparison, 1302

DEMES achieves 0.82.2 mN force at comparable massvacuum-gap actuators thus exceed 1303

DEA performance by three orders of magnitude in force at the same mass scale. The gearless, 1304

lubricant-free construction eliminates two major space reliability concerns. 1305

(a) Terrestrial operation

(b) Space operation

Atmospheric pressure opposes vacuum gap

Ambient vacuum provides functional gap

Electrode (+)

Electrode (+)

Fe V

Fe V

Vacuum gap

Vacuum gap

(pumped)

(ambient)

Flexible membrane

Flexible membrane

Electrode (-)

Electrode (-)

Ambient vacuum = functional gap

Atmospheric pressure

must be overcome

No pump required

1 atm

~0 Pa (vacuum)

Sirbu et al. 2025: 0.7 g, >4 N force, >100 Hz bandwidth, specific power 614 W/kg

Figure 7: Vacuum-gap electrostatic actuator operating principle (after Sirbu et al. 2025 Sirbu et al. [2025]). (a) In terrestrial operation, vacuum gaps between electrodes must be main- tained against atmospheric pressure, requiring a vacuum pump. (b) In space, the ambient vacuum provides the functional dielectric gap directly, eliminating the pump and enabling higher bandwidth (>100 Hz) at extremely low mass (0.7 g, >4 N, specic power 614 W/kg).

The thin-lm polymer construction of vacuum-gap actuators is structurally analogous 1306

to the multilayer membrane systems already used in inatable habitat construction. The 1307

possibility of laminating vacuum-gap actuator layers to the inner liner of an inatable robotic 1308

arm, combined with bre optic shape sensors woven into the restraint webbing, suggests 1309

a pathway toward fully sensorized, actively controlled inatable manipulatorsa system 1310

architecture not yet demonstrated in the literature. The primary unresolved qualication 1311

gaps are thermal cycling (−150◦C to +120◦C), radiation tolerance, and scale-up beyond the 1312

current laboratory-scale prototypes. 1313

7.3 Ionic Electroactive Polymers: Space Tolerance Assessment 1314

Ionic electroactive polymer (IEAP) actuators operate through ion migration within a polymer 1315

membrane, producing bending deformation at low voltages (15 V). Punning et al. conducted 1316

the only systematic, large-sample space environment tolerance study for this actuator class, 1317

testing 320 samples across 7 IEAP material types under six space-relevant conditions: X-ray 1318

irradiation (167.4 Gy), gamma irradiation (2036 Gy from 60Co), UV exposure (180 hours, 1319

xenon lamp), vacuum (<1 mbar, 2 weeks), and cryogenic storage at 77 K (liquid N2, 2 weeks) 1320

and 4.22 K (liquid He) Punning et al. [2014]. 1321

The results establish three design rules for space IEAP deployment: 1322

1. Use ionic liquid electrolytes: IEAP types employing ionic liquid (IL) electrolytes 1323

(EMIBF4, EMITF, EMITFSI) showed no notable degradation under vacuum or cryo- 1324

genic conditions. Aqueous IPMC actuators (Type A) dry out in vacuum, requiring 1325

encapsulation for space use. 1326

2. Provide UV shielding for external applications: UV irradiation destroys PE- 1327

DOT and PEO-based IEAP materials via photo-oxidation. This is the primary space 1328

environment threat. Materials using carbonaceous or conducting polymer electrodes 1329

with ionic liquid electrolytes (Types B, C, G) survived UV testing with no notable 1330

eect. 1331

3. Plan for cryogenic dormancy: All tested IEAP types survived cryogenic stor- 1332

age (77 K for 2 weeks, 4.22 K for 15 minutes) and recovered full functionality upon 1333

warmingthe materials cannot operate while frozen but survive and revive Punning 1334

et al. [2014]. 1335

A counter-intuitive nding is that X-ray radiation initially increases IEAP performance 1336

through radiation-induced doping of conducting polymer electrodes, an eect that normalizes 1337

within a few actuation cycles Punning et al. [2014]. The force output of current IEAPs 1338

remains in the low-millinewton range, limiting applications to sensing-adjacent tasks and 1339

micro-manipulation. 1340

7.4 Tendon-Driven Continuum Manipulators 1341

Tendon-driven continuum manipulators represent the highest-force soft actuation approach 1342

compatible with space constraints. NASA's Tendril robot (Mehling et al., 2006) established 1343

the heritage origin: a 1:1000 aspect-ratio inspection robot designed for conned-space inspec- 1344

tion inside the Space Shuttle external tank Mehling et al. [2006]. The Tendril architecture 1345

multiple antagonistic tendons routed along a compliant backboneprovides both the force 1346

density and bandwidth necessary for structural manipulation tasks. 1347

Ouyang et al. proposed a hybrid rigid-continuum dual-arm space robot combining a rigid 1348

primary arm for strength and reach with a continuum secondary arm for dexterity and com- 1349

pliance Ouyang et al. [2021]. The Generalized Jacobian Matrix analysis demonstrated coor- 1350

dinated motion planning for free-oating operations, establishing the mathematical frame- 1351

work for hybrid architectures where inatable continuum arms complement rigid primary 1352

manipulators. 1353

For space-compatible tendon routing, MoS2 solid lubricant enables vacuum-compatible 1354

sliding contacts, addressing the lubrication challenge that would otherwise limit tendon- 1355

driven systems to short operational lifetimes Ruíz et al. [2023]. The primary limitation 1356

of tendon-driven approaches is that routing tendons over long lengths (>1 m) introduces 1357

increasing friction and hysteresis, requiring careful mechanical design. 1358

7.5 Shape Memory Alloys for Deployment 1359

Shape memory alloys (SMAs), principally NiTi (Nitinol), have the highest ight TRL (8 1360

9) among actuator technologies applicable to soft inatable systems, though primarily for 1361

one-shot deployment rather than cyclic actuation. Nitinol achieves up to 10% recoverable 1362

strain and cycle life up to 600,000 activation cycles under controlled conditions Costanza and 1363

Tata [2020]. Space heritage includes Mars Pathnder deployment hinges, numerous CubeSat 1364

solar array release mechanisms, and ESA satellite solar array root hinges Costanza and Tata 1365

[2020], Blanc et al. [2013]. 1366

For inatable structures specically, the critical limitation of SMA is its slow cooling 1367

rate in the vacuum thermal environment. Without convective cooling, SMA actuators rely 1368

on radiative heat transfer alone, limiting cyclic actuation frequency to well below 1 Hz for 1369

typical element sizes. This eectively restricts SMA to single-deployment or low-frequency 1370

repositioning applications in space. 1371

An emerging application combines SMA with inatable structures: Patel et al. devel- 1372

oped an analytical framework for SMA-based rigidisation of inatable toroidal structures, 1373

where NiTi wires embedded in the inatable wall contract upon heating to lock the deployed 1374

geometry Rastogi et al. [2024]. This represents a potential fourth rigidisation approach be- 1375

yond the three families established by Cadogan and Scarborough Cadogan and Scarborough 1376

[2001], though it remains at the analytical stage (TRL 23). 1377

7.6 Jamming in Vacuum: A Novel Opportunity 1378

Variable stiness by granular or layer jamming presents a counter-intuitive advantage in 1379

the space environment that has not been previously identied in the literature. In terres- 1380

trial soft robotics, jamming requires a dedicated vacuum pump to evacuate the jammed 1381

medium's enclosure, with external atmospheric pressure (∼101 kPa) providing the conning 1382

force Fitzgerald et al. [2020]. Zhang et al. noted that jamming structures are more likely 1383

to be used in soft space robots because of scalability, easy fabrication, and low cost Zhang 1384

et al. [2023c], but did not explore the vacuum-specic advantage. 1385

In the space environment, this constraint inverts: the ambient vacuum of LEO (∼10−7 Pa) 1386

serves as the external conning medium, while an inatable structure's internal pressuriza- 1387

tion (∼100 kPa) provides the pressure dierential across the membrane wall. A sealed jam- 1388

ming structure integrated into or attached to a pressurised inatable therefore achieves sti- 1389

ness modulation without any vacuum pumpa simplication unavailable on Earth. Layer 1390

jamming, which achieves stiness ratios exceeding 25:1 in terrestrial systems Fitzgerald et al. 1391

[2020], could be particularly well-suited for variable-stiness robotic elements embedded in 1392

inatable arms. 1393

(a) Terrestrial jamming

(b) Space jamming

Requires vacuum pump

Ambient vacuum = confining pressure

Patm = 101.3 kPa

Pvacuum 0 Pa

(external confining pressure)

(ambient space vacuum)

Granular

Granular

medium (particles)

medium (particles)

No pump

Vacuum

Inflatable structure

needed

pump

(internal pressure)

1 atm

~0 Pa

Evacuates interior

Stiffness transition: compliant (unjammed) to rigid (jammed) via pressure differential

Figure 8: Jamming-in-vacuum principle for variable stiness in space. (a) Terrestrial cong- uration: a vacuum pump evacuates the sealed granular membrane, and atmospheric pressure (∼101 kPa) provides the external conning force that locks the particles. (b) Space congu- ration: the ambient space vacuum provides external conning pressure directly; the internal pressurisation of the host inatable structure provides the pressure dierential. The vacuum pump is eliminated, and the stiness transition from compliant to rigid is achieved passively.

The primary engineering challenges are: (1) selecting space-compatible granular media 1394

that do not outgas (candidates include hollow glass microspheres and sintered ceramic gran- 1395

ules); (2) maintaining gas-tight seals over mission duration against micrometeoroid puncture; 1396

and (3) characterising friction behaviour of jammed interfaces in vacuum, where the absence 1397

of adsorbed water layers may alter surface friction coecients. This insight represents a logi- 1398

cal deduction from known physics and inatable structure operating principles, and requires 1399

experimental validationa 5-year research priority identied in Section 13.3. 1400

7.7 Sealed Pneumatic Actuation in Space 1401

The opening constraint of this sectionthat ultrahigh vacuum eliminates ambient pressure 1402

support for unsealed pneumatic systemsdoes not preclude sealed pneumatic actuators that 1403

carry their own gas supply. BEAM itself is the supreme example of a sealed pneumatic 1404

structure in space. Ataka et al. Ataka et al. [2020] demonstrated a closed-loop pneumatic 1405

eversion arm with observer-based control that is directly relevant to inatable continuum 1406

manipulators for space inspection tasks. Eversion robots, which navigate their environment 1407

through growth by turning inside-out Hawkes et al. [2017], are particularly promising for 1408

space applications because the growth mechanism inherently manages the gas supply within 1409

the extending structure. 1410

Sealed pneumatic actuation with onboard gas storage is viable for missions where the total 1411

number of actuation cycles is bounded (limiting gas consumption) or where the inatable 1412

structure's own pressurisation system can serve as the gas source. The mass penalty of gas 1413

storageapproximately 0.52 kg per litre at 200 bar, depending on tank technologymakes 1414

this approach less competitive for sustained cyclic actuation than electrical alternatives, but 1415

appropriate for deployment and one-shot or low-cycle capture operations. 1416

7.8 Electroadhesion and Magnetic Actuation: Emerging Approaches 1417

Two additional actuation families, while not yet proposed for space inatable systems, merit 1418

assessment for completeness. 1419

Electroadhesion (electrostatic adhesion to a target surface) diers from the vacuum-gap 1420

actuators of Section 7.2 in operating principle: Coulombic attraction to an external target 1421

surface rather than internal gap zipping. Guo et al. Guo et al. [2020] demonstrated elec- 1422

troadhesion pads integrated with soft robotic grippers for manipulation of non-cooperative 1423

surfaces, achieving adhesion pressures of 15 kPa on conductive substrates. For debris cap- 1424

ture on metallic spacecraft surfaces, electroadhesion oers a contactless-force alternative to 1425

mechanical grasping. The primary space qualication gaps are dielectric breakdown in par- 1426

tial vacuum (outgassing-induced), surface contamination from space debris, and radiation 1427

degradation of the dielectric layer. 1428

Magnetic soft actuators with programmed 3D magnetisation proles Kim et al. [2018] 1429

represent a fundamentally dierent approach that avoids the vacuum and temperature limi- 1430

tations of pneumatics and elastomers. While not yet proposed for space, magnetic actuation 1431

in the eld-free environment of orbit would require onboard eld sources (permanent magnets 1432

or electromagnets), adding mass but eliminating the outgassing and embrittlement concerns 1433

of polymer-based actuators. This approach remains at TRL 2 for space applications. 1434

Table 14 presents a comparative assessment of the nine actuation technologies assessed 1435

for inatable space systems. 1436

Table 14: Actuator technology comparison for soft inatable space systems.

Technology Force Speed Mass TRL Critical Space Gap (Space)

DEA/DEMES 0.82.2 mN ∼1 Hz <0.65 g 34 UV, outgas., low force Vacuum-gap electrost. >4 N >100 Hz 0.7 g 34 Radiation, thermal IL-IEAP (types B,C) Very low Medium Excellent 34 UV (shield), frozen op. Tendon-driven High High Good 56 Long-path friction SMA (one-shot) Medium Slow Low 89 Slow cooling, fatigue Jamming (layer) Stiness only Medium Good 23 Unvalidated in vacuum Sealed pneumatic High Medium Mod. (gas) 45 Gas supply mass Electroadhesion 15 kPa Fast Low 23 Surface contam., diel. brkdn Magnetic (programmed) Medium Fast Mod. (magnet) 12 Requires onboard eld

8 State of the Art: Sensing and Structural Health Mon- 1437

itoring 1438

Structural health monitoring (SHM) for inatable space structures must address three simul- 1439

taneous requirements: detection of micrometeoroid and orbital debris (MMOD) impacts that 1440

may compromise pressure integrity, continuous monitoring of creep deformation in restraint 1441

layers under sustained pressure loading, and shape sensing for actively controlled inatable 1442

manipulators. Fibre Bragg Grating (FBG) sensors have emerged as the leading technology 1443

platform for all three functions, with a coherent maturation pathway from rigid spacecraft 1444

heritage through soft actuator integration to inatable habitat application. 1445

8.1 Fibre Bragg Grating Sensors: From Proba-2 to Inatable Web- 1446

bing 1447

The FBG sensing principlewavelength-selective reection from a periodic refractive index 1448

modulation inscribed in a bre coreenables wavelength-division multiplexing (WDM) and 1449

time-division multiplexing (TDM) of large sensor arrays on a single bre strand. A single 1450

bre can carry 100+ independent FBG sensors, each at a distinct Bragg wavelength, pro- 1451

viding distributed strain and temperature measurement with no electrical connections at 1452

the measurement points Mckenzie et al. [2021]. Temperature sensitivity is approximately 1453

10 pm/◦C in the 15001600 nm wavelength range. 1454

ESA's 20+ year investment in bre optic sensing for spacecraft culminated in the Fiber 1455

Sensor Demonstrator (FSD) aboard Proba-2, launched in November 2009the rst bre 1456

optic sensor network demonstrated in the space environment Mckenzie et al. [2021]. The 1457

FSD incorporated 12 temperature sensors, a high-temperature thruster sensor, and a xenon 1458

tank pressure sensor, establishing TRL 78 for FBG technology on rigid spacecraft platforms. 1459

DEA / DEMES

Vacuum-gap electrostatic

Tendon-driven

SMA (deployment)

Sealed pneumatic

Jamming (vacuum-enabled)

Electroadhesion

Space TRL Force output Bandwidth Vacuum compatibility Mass efficiency

IEAP / IPMC

TRL 6 (flight qualified)

0 2 4 6 8 10 Rating (0 = lowest, 10 = highest)

Figure 9: Comparative assessment of actuation technologies for soft inatable space sys- tems across ve performance dimensions: space TRL, force output, bandwidth, vacuum compatibility, and mass eciency. Ratings on a 010 scale reect the combined evidence from literature reviewed in Sections 7.17.8. Vacuum-gap electrostatic actuators Sirbu et al. [2025] and jamming Fitzgerald et al. [2020] score highest on vacuum compatibility, reecting the vacuum as enabler paradigm shift.

Radiation tolerance of appropriately selected bre types (nitrogen-doped, uorine-doped) has 1460

been conrmed through ground testing, with Type II and Type III FBGs showing the highest 1461

radiation hardness Girard et al. [2022], Baba et al. [2025]. 1462

The critical transition from rigid spacecraft to inatable structures was demonstrated by 1463

Bally Ribbon Mills (BRM) and Luna Innovations under a NASA SBIR program Bally Ribbon 1464

Mills and Luna Innovations [2020]. High-Denition Fibre Optic Sensing (HD-FOS) elements 1465

were woven directly into Vectran structural restraint webbing during the manufacturing 1466

processnot bonded after fabrication. Testing on 0.61 m and 2.74 m (1/3-scale) inatable 1467

habitat test articles at NASA Johnson Space Center demonstrated detection of: 1468

ˆ Creep deformation under sustained pressure loading 1469

ˆ Internal pressure changes during ination and operational cycling 1470

ˆ Micrometeoroid impact events (conrmed via hypervelocity impact testing on inated 1471

articles) 1472

The partnership included NASA, Sierra Nevada Corporation, ILC Dover, BRM, and Luna 1473

Innovations, targeting applications for the Lunar Gateway and Mars transit habitats Bally 1474

Ribbon Mills and Luna Innovations [2020]. However, these results have been reported only 1475

in technical briefs and SBIR documentation, not in peer-reviewed publicationsa gap that 1476

limits independent assessment of sensitivity metrics, minimum detectable impact size, and 1477

long-term reliability. 1478

The TRL assessment for FBG sensing across application domains is: 1479

ˆ FBG on rigid spacecraft: TRL 78 (Proba-2 FSD ight heritage, 2009) 1480

ˆ FBG in Vectran restraint webbing: TRL 45 (NASA JSC ground testing, 0.61 m and 1481

2.74 m articles) 1482

ˆ FBG in operational inatable habitat (ight): TRL 23 (not yet demonstrated) 1483

Table 15: Sensing technology comparison for inatable structural health monitoring.

Technology Accuracy Channels Space Demo TRL /Fiber Heritage Scale

FBG (rigid s/c) ±10 µε / ±1◦C 100+ Proba-2 (2009) Satellite 78 FBG (Vectran webbing) Creep/MMOD det. Multiple JSC ground 2.74 m 45 Multicore FOSS 0.64 mm tip Multicore Lab only Actuator 34 DFOS (Rayleigh) ±1 µε Continuous Lab only m-scale 23 Capacitive (stretch.) ±5% strain Per-sensor Lab only cm-scale 23 Resistive (fabric) ±2% strain Per-sensor Lab only cm-scale 23 Piezoelectric (PVDF) Impact detection Array Lab only Panel 23

8.2 Multicore Fibre Optic Shape Sensing 1484

For soft actuator shape sensing, Galloway et al. demonstrated the rst integration of a 1485

monolithic multicore Fibre Optic Shape Sensor (FOSS) into a bre-reinforced soft pneumatic 1486

actuator Galloway et al. [2019]. The multicore bre contains multiple sensing cores within a 1487

single cladding, enabling three-dimensional shape reconstruction from dierential curvature 1488

measurements without requiring multiple separate bre installations. Key results include a 1489

mean tip position error of 0.64 mm, successful reconstruction of six distinct planar shape 1490

proles, and simultaneous detection of collision events, environmental shape changes, and 1491

material stiness variations within a single sensing modality. 1492

The eld has advanced signicantly since Galloway's initial demonstration. Paloschi et al. Paloschi 1493

et al. [2020] developed improved 3D shape reconstruction algorithms for multicore optical 1494

bres, comparing transformation matrix approaches with Frenet-Serret equations for real- 1495

time applications and demonstrating that transformation matrix methods achieve superior 1496

accuracy for large-curvature deformations characteristic of soft actuators. Sefati et al. Sefati 1497

et al. [2021] demonstrated real-time 3D shape reconstruction using multicore FBG sensors 1498

on continuum robot manipulators, achieving sub-millimetre accuracy relevant to the tendon- 1499

driven continuum arms discussed in Section 7.4. These advances collectively bring multicore 1500

FOSS from a proof-of-concept to a viable shape sensing modality for soft space manipulators, 1501

though the interrogator hardware miniaturisation and radiation tolerance gaps remain. 1502

The multicore FOSS approach oers two advantages over distributed single-core FBG 1503

arrays for soft structure applications. First, the monolithic construction eliminates the need 1504

for multiple bre routing paths through complex soft geometries. Second, the dierential 1505

curvature measurement provides inherent common-mode rejection of temperature-induced 1506

wavelength shifts, improving strain measurement accuracy in the thermally variable space en- 1507

vironment. The primary barriers to space qualication are the mass and power requirements 1508

of the multicore FOSS interrogator (readout) hardware, which has not yet been miniaturized 1509

for spacecraft integration, and the radiation tolerance of the multicore bre itself, which has 1510

not been characterised. 1511

For broader context, Rajan et al. Rajan et al. [2021] provide a comprehensive review of 1512

FBG sensors for structural health monitoring across aerospace applications, conrming that 1513

FBG-based SHM is the most mature bre optic sensing technology for spacecraft structures 1514

and identifying the key challenges for transitioning from rigid to exible substrates. 1515

8.3 Capacitive, Resistive, and Alternative Soft Sensors 1516

While FBG sensors dominate the space-qualied sensing landscape, alternative soft sensing 1517

technologies merit assessment for completeness. Zhang et al. Zhang et al. [2023a] devote 1518

signicant attention to stretchable capacitive sensors, resistive fabric sensors, and liquid 1519

metal strain sensors for soft space robots. The advantages of these technologies include: no 1520

requirement for specialised interrogator hardware (unlike FBG, which requires wavelength- 1521

swept laser sources), simpler integration into soft structures via printing or embedding, and 1522

lower per-sensor cost. However, for space applications, three signicant limitations arise: 1523

ˆ Electromagnetic interference (EMI) sensitivity: Capacitive and resistive sensors 1524

operate in the electrical domain and are vulnerable to the charged particle environ- 1525

ment of LEO, solar radio bursts, and EMI from onboard electronics. FBG sensors, 1526

operating in the optical domain, are inherently immune to EMIa decisive advantage 1527

for spacecraft. 1528

ˆ Radiation vulnerability: Liquid metal sensors (e.g., eutectic gallium-indium, EGaIn) 1529

and conductive polymer sensors have not been characterised for radiation tolerance. 1530

Ionising radiation can alter the resistivity of conductive polymers and the wetting prop- 1531

erties of liquid metals, degrading sensor calibration over mission-duration timescales. 1532

ˆ Multiplexing limitations: A single optical bre can carry 100+ independent FBG 1533

sensors via wavelength-division multiplexing; achieving comparable channel density 1534

with electrical sensors requires extensive wiring harnesses that add mass and failure 1535

modes to exible structures. 1536

For inatable habitat applications, capacitive pressure sensors could complement FBG 1537

strain sensors by providing direct pressure measurement at locations inaccessible to bre 1538

routing. For soft robotic manipulators, resistive bend sensors oer simplicity advantages for 1539

prototype development, though FBG remains the preferred technology for ight systems. 1540

8.4 Distributed Fibre Optic Sensing: Rayleigh and Brillouin Scat- 1541

tering 1542

Distributed bre optic sensing (DFOS) by Rayleigh or Brillouin scattering provides contin- 1543

uous strain and temperature proles along the entire bre length, rather than at discrete 1544

FBG grating locations. Rayleigh-based DFOS (e.g., Luna Inc. ODiSI platform) achieves 1545

spatial resolution of approximately 0.65 mm with strain resolution better than ±1 µε, while 1546

Brillouin-based systems provide sensing over distances up to 100 km at reduced spatial res- 1547

olution (typically 0.51 m). For inatable habitats with large membrane areas requiring 1548

continuous monitoring (rather than point-by-point FBG interrogation), DFOS oers the 1549

potential for comprehensive strain mapping of the entire restraint layer from a single bre 1550

installation. 1551

The principal barriers to space deployment of DFOS are: (i) interrogator size, mass, 1552

and power (current laboratory DFOS systems exceed 10 kg and 50 W, compared to <2 kg 1553

and <10 W for space-grade FBG interrogators); (ii) sensitivity to bre bending loss, which 1554

is exacerbated by the tight bend radii in folded inatable structures during stowage; and 1555

(iii) the absence of any space-environment characterisation data. DFOS is assessed at TRL 2 1556

3 for space inatable applications, but its unique capability for continuous spatial coverage 1557

makes it a high-priority development target for large-scale habitat SHM systems. 1558

8.5 Distributed Impact Detection 1559

The Distributed Impact Detection System (DIDS) installed on BEAM represents the highest- 1560

TRL implementation of impact sensing for inatable habitats. DIDS uses distributed sensors 1561

to detect and locate MMOD impacts on the inatable shell, providing real-time structural 1562

integrity monitoring. 1563

Beyond the BEAM DIDS, two emerging approaches extend impact detection capabilities. 1564

The BRM/Luna FBG-in-Vectran-webbing system described in Section 8.1 detected hyperve- 1565

locity impacts during ground testing, with the woven integration providing inherent coverage 1566

of the restraint layer structural grid Bally Ribbon Mills and Luna Innovations [2020]. Sepa- 1567

rately, Liao et al. demonstrated on-demand fabrication of PVDF-trFE piezoelectric sensors 1568

via in-space manufacturing techniques, enabling the production of impact detection arrays 1569

directly on deployed inatable structures White et al. [2024]. This approach could address 1570

the challenge of instrumenting structures that are too large or complex to pre-instrument 1571

before launch. 1572

Wei et al. proposed a complementary SHM approach based on low-frequency vibration 1573

response characterisation of pressurised inatable structures, where changes in modal fre- 1574

quencies indicate structural degradation Li et al. [2022b]. This global monitoring approach 1575

could complement the local sensing provided by FBG arrays, together forming a hierarchical 1576

SHM architecture: global vibration monitoring for overall structural health assessment, and 1577

local FBG sensing for precise damage location and magnitude quantication. 1578

The pathway from current demonstrated capabilities to a ight-qualied inatable SHM 1579

system requires: (1) formal peer-reviewed publication of the BRM/Luna FBG-in-webbing 1580

results with full sensitivity characterisation; (2) space qualication of FOSS interrogator 1581

hardware (mass, power, radiation tolerance); (3) development of data fusion algorithms 1582

combining local FBG and global vibration sensing; and (4) a ight demonstration, potentially 1583

as an ISS external payload experiment, to bridge the TRL 45 to TRL 78 gap. 1584

9 State of the Art: Power Systems for Large Inatables 1585

The integration of electrical power generation with inatable space structures is a critical 1586

enabling challenge for large deployable platforms. Unlike rigid spacecraft, where solar ar- 1587

rays are mechanically decoupled from the primary structure, inatable systems present the 1588

possibilityand the engineering challengeof co-locating photovoltaic generation on the de- 1589

ployable membrane itself. This section reviews the exible solar array landscape, the singular 1590

attempt at inatable-power integration (PowerSphere), and energy storage considerations for 1591

mission architectures ranging from 100 m-class debris shields to inatable habitats. 1592

9.1 Flexible Solar Array Landscape: ROSA to Perovskite 1593

The current state of the art in exible solar arrays for space is dened by the Roll-Out Solar 1594

Array (ROSA), which achieved TRL 9 via ISS ight demonstration in June 2017 as part 1595

of the STP-H5 experiment Spence et al. [2018]. The demonstration unit (5.40 m × 1.67 m) 1596

deployed successfully using stored strain energy in carbon-bre-reinforced polymer (CFRP) 1597

slit-tube booms, requiring no motors. The subsequent production variant, iROSA, scaled 1598

to 6 m × 13.7 m wings generating over 28 kW per wing at beginning of life with XTJ Prime 1599

triple-junction cells at 30.7% eciency. Six iROSA wings installed on the ISS between 1600

2021 and 2023 added over 120 kW of generation capacity. At system level (blanket plus 1601

booms, excluding spacecraft attachment hardware), ROSA achieves a specic power exceed- 1602

ing 100 W/kgapproximately 3.7× the legacy ISS silicon rigid-panel arrays at ∼27 W/kg 1603

Spence et al. [2018], Yan et al. [2025]. Critically, however, ROSA's exible photovoltaic 1604

blanket is deployed on rigid composite booms; the deployment mechanism is structurally 1605

distinct from inatable substrate concepts. 1606

Beyond ROSA, three deployment architectures compete for next-generation high-power 1607

arrays Yan et al. [2025]: (i) Z-fold accordion panels on a central mast, representing the 1608

ISS legacy approach at TRL 9; (ii) fan-fold blankets on deployable masts, exemplied by 1609

China's CST arrays on the Wentian laboratory module (2022), achieving approximately 4× 1610

the specic power of rigid baselines; and (iii) roll-out arrays (ROSA/iROSA class). Mega- 1611

ROSA and SOLAROSA concepts target 200500 W/kg for systems exceeding 100 kW, though 1612

these remain at TRL 45 Yan et al. [2025]. For very large arrays approaching the kilometre 1613

scale (Space Solar Power Station concepts), wireless power transmission between modules 1614

has been identied as a necessity Yan et al. [2025]. 1615

A paradigm shift in exible photovoltaic technology is emerging from perovskite-based 1616

tandem cells. Lang et al. Lang et al. [2020] provided the critical nding that perovskite/CIGS 1617

(copper indium gallium selenide) tandem cells are radiation-hard, while perovskite/silicon 1618

heterojunction (SHJ) tandems are emphatically not. Under 68 MeV proton irradiation at a 1619

uence of 2 × 1012 p+/cm2 (equivalent to over 50 years at ISS altitude), perovskite/CIGS 1620

tandems retained approximately 85% of initial power conversion eciency, whereas per- 1621

ovskite/SHJ devices degraded catastrophically to ∼1% retention due to proton-induced deep 1622

trap states in the silicon bottom cell Lang et al. [2020]. The perovskite top cell itself was 1623

essentially unaected, with quasi-Fermi level splitting changing by only 0.004 eV. With ac- 1624

tive layers only 4.38 µm thick (2.8 mg/cm2), perovskite/CIGS achieves a specic power of 1625

7,400 W/kg at the active-layer level, or 2,100 W/kg when including a 25 µm exible poly- 1626

imide substrate Lang et al. [2020]. More recently, Kim et al. Kim et al. [2024] demonstrated 1627

23.6% ecient exible perovskite/CIGS tandems surviving 100,000 bending cycles with a 1628

specic power of approximately 6,150 W/kg at the cell level. 1629

These gures represent a 1060× improvement over ROSA's system-level specic power, 1630

though the comparison requires careful attention to system boundaries: cell-only gures 1631

exclude interconnects, encapsulant, wiring harness, and structural substrate, which collec- 1632

tively reduce specic power by a factor of 36× at the system level. Table 16 summarises 1633

the specic power versus TRL landscape across exible photovoltaic technologies. 1634

Table 16: Specic power versus TRL for exible photovoltaic technologies for space applica- tions. Cell-only and system-level gures are distinguished where data are available.

Technology Specic Power (W/kg) Eciency (%) TRL Ref.

Legacy ISS SAW (rigid) ∼27 (system) 14 9 Spence et al. [20 ATK UltraFlex ∼150 (system) 2830 9  ROSA/iROSA >100 (system) 30.7 9 Spence et al. [20 Mega-ROSA (target) >200400 30.7 45 Yan et al. [202 Perovskite/CIGS (25 µm sub.) 2,100 (cell+sub.) 19.2 34 Lang et al. [202 Perovskite/CIGS (Kim 2024) ∼6,150 (cell) 23.6 34 Kim et al. [202 PowerSphere (a-Si, measured) ∼7 (system) 10 45 Cadogan et al. [2 PowerSphere (proj. w/ III-V) ∼85 (projected) 2730  Cadogan et al. [2

Category / Cell efficiency

1400

Target region: high specific power

= 12%

Heritage

Thin film

= 25%

+ flight-qualified

Perovskite (single jxn)

Emerging

= 33%

1200

Inflatable

Specific power (W/kg)

1000

800

Perovskite/Si

tandem

600

CIGS thin-film

400

a-Si thin-film

200

PowerSphere

III-V MJ (rigid) ROSA/iROSA

(integr.)

(III-V flex)

0

1 2 3 4 5 6 7 8 9 10 Technology Readiness Level (TRL)

Figure 10: Specic power versus technology readiness level for exible photovoltaic technolo- gies relevant to inatable space structures. Marker size indicates cell eciency. Perovskite- based technologies Lang et al. [2020], Kim et al. [2024] oer 1060× improvements over heritage ROSA systems Spence et al. [2018] at the cell level, but remain at TRL 34. The green shaded region indicates the target design space for next-generation inatable-power integration: high specic power (>400 W/kg) at ight-qualied TRL (>6).

9.2 The Inatable-Power Integration Gap: PowerSphere and Be- 1635

yond 1636

The most direct attempt to integrate thin-lm photovoltaics with an inatable deployable 1637

structure was NASA's PowerSphere programme (20042009), led by ILC Dover (structure), 1638

NASA Glenn Research Center (cells), and Sandia National Laboratories (interconnects) 1639

Cadogan et al. [2003], Scheiman et al. [2005]. The PowerSphere Engineering Development 1640

Unit was a 0.6 m diameter UV-rigidisable inatable geodetic sphere clad with thin-lm amor- 1641

phous silicon (a-Si) solar cells on a polyimide substrate. The complete system comprised a 1642

1 kg PowerSphere subsystem mounted on a 3 kg bus, with 15 cells per hemisphere (9 hexag- 1643

onal, 6 pentagonal) connected via copper wrap-around ex-circuit interconnects that could 1644

survive folding during stowage without cracking Cadogan et al. [2003], Scheiman et al. [2005]. 1645

The UV-rigidisation mechanism is particularly signicant for the survey's themes. Thirty 1646

hinges per sphere used S-glass bre reinforced with ATI-P600-2 UV-curing epoxy (glass tran- 1647

sition temperature Tg = 211 ◦C), encapsulated in UV-transparent 1-mil Mylar lm. Upon 1648

exposure to solar UV radiation (λ < 385 nm) for 1045 minutes post-deployment, the resin 1649

polymerised, converting the inatable into a self-supporting rigid structure and eliminating 1650

the requirement for long-term ination gas retention Cadogan et al. [2003]. Ination was 1651

achieved passively through vapour pressure from sublimation powder at approximately 1 psi 1652

(∼6.9 kPa). 1653

Thermal cycling tests (−120 ◦C to +80 ◦C, 1000 cycles per NASA specication) demon- 1654

strated cell and interconnect survival with less than 2% power degradation, although one 1655

of four interconnect coupons failed, prompting the addition of a titanium binder layer as a 1656

design modication. Cell interconnect technology was partially validated on the MISSE-5 1657

experiment aboard the ISS Cadogan et al. [2003]. At 10% a-Si cell eciency, the 0.6 m 1658

prototype generated approximately 29 W at design point, yielding a system specic power 1659

of ∼7.25 W/kg. With projected III-V triple-junction cells at 2730% eciency, the concept 1660

was estimated to reach ∼85 W/kg. 1661

The PowerSphere programme reached TRL 45 but never ew. Planned missionsthe 1662

PowerSphere Flight Experiment and PSIREX (Pico Satellite Inatable Reector Experiment) 1663

were not implemented, and the programme appears inactive since the nal publication by 1664

Jenkins in 2009 on thermal cycling results Jenkins et al. [2009]. No successor programme inte- 1665

grating thin-lm photovoltaics with inatable structure deployment has been identied. This 1666

represents a critical gap: ROSA (TRL 9) demonstrates that exible photovoltaic blankets 1667

function reliably in space, and PowerSphere (TRL 45) demonstrated that cells can survive 1668

fold/deploy on an inatable substrate, but nobody is currently pursuing inatable-integrated 1669

photovoltaics. A revival of the PowerSphere concept using modern perovskite/CIGS cells 1670

which oer 200300× higher specic power than the original a-Si cells and validated radiation 1671

hardness Lang et al. [2020]represents a logical and compelling research direction. 1672

9.3 Energy Storage: Li-ion, RFC, and Mission-Dependent Selection 1673

Energy storage for large inatable structures follows established space heritage, with tech- 1674

nology selection driven primarily by eclipse duration and mission architecture. The current 1675

standard is lithium-ion, with state-of-the-art cell-level specic energy of 200300 Wh/kg 1676

and system-level (including battery management, thermal control, and structure) of 100 1677

160 Wh/kg Sharma and Santasalo-Aarnio [2025]. The ISS lithium-ion upgrade programme 1678

(20172021), replacing nickel-hydrogen (Ni-H2) with 24 lithium-ion Orbital Replacement 1679

Units (ORUs) at 4 kWh each, provides direct heritage for large-structure lithium-ion energy 1680

storage. 1681

For a 100 m-class inatable debris shield in LEO (90-minute orbit, 36-minute eclipse), 1682

the power demand is driven by supporting subsystems rather than the passive membrane 1683

itself. Station-keeping via electric propulsion dominates at 150 kW depending on orbit and 1684

attitude strategy (Section 11.3); attitude control, telemetry, and sensors add 17 kW. A total 1685

system power demand in the range of 250 kW is appropriate, requiring 440 kWh of eclipse 1686

energy storagetranslating to 25250 kg of lithium-ion battery mass at 160 Wh/kg system 1687

level. This is a non-trivial but manageable fraction of the estimated 5,000 kg total system 1688

mass. 1689

For missions requiring extended eclipse storagenotably lunar surface operations (354- 1690

hour lunar night) or deep-space transitregenerative fuel cells (RFCs) oer 4001,000 Wh/kg 1691

at system level but remain at TRL 56 for space applications Sharma and Santasalo-Aarnio 1692

[2025]. Supercapacitors (515 Wh/kg) are poorly suited for eclipse energy storage but may 1693

serve pulsed-load applications such as electric propulsion ignition or deployment actuators. 1694

Table 17 summarises the energy storage technology comparison. 1695

Table 17: Energy storage technologies for large inatable space structures.

Technology Sp. Energy (Wh/kg) Cycle Life TRL Best Use Case

Li-ion (cell) 200300 >30,000 9 LEO eclipse storage Li-ion (system) 100160 >30,000 9 LEO eclipse storage Ni-H2 (legacy) 3060 >40,000 9 Heritage only RFC (H2/O2) 4001,000  56 Lunar night, deep space Supercapacitor 515 >500,000 7 Pulsed loads RTG N/A  9 No-sun environments

10 State of the Art: Thermal Management 1696

Thermal management for inatable space structures presents unique challenges that stem 1697

from the fundamental nature of the structural material: thin fabric membranes oer minimal 1698

thermal mass, poor through-thickness conductivity, and large surface area-to-volume ratios. 1699

These characteristics amplify the orbital thermal cycling environment and demand thermal 1700

control approaches that are compatible with the fold/deploy lifecycle, vacuum exposure, and 1701

the mechanical exibility of the host structure. This section reviews established approaches 1702

(multi-layer insulation, loop heat pipes), the JWST sunshield as a large-area deployable 1703

thermal barrier precedent, and emerging technologies (variable emissivity coatings, phase 1704

change materials) that oer particular promise for inatable applications. 1705

10.1 Multi-Layer Insulation for Inatable Shells 1706

Multi-layer insulation (MLI) is the primary passive thermal control technology for spacecraft 1707

and achieves eective emittance εe = 0.0050.05 for 1040 layer blankets Gilmore [2002], 1708

Finckenor and Dooling [1999]. For conventional rigid spacecraft, MLI is draped over external 1709

surfaces with controlled layer separation maintained by low-conductance spacers (typically 1710

Dacron netting). For inatable structures, MLI integration is more complex: the insulation 1711

must survive folding, accommodate deployment kinematics, and maintain layer separation 1712

without rigid structural support. 1713

The TransHab/BEAM heritage shell architecture represents the current standard for 1714

inatable habitat thermal design Kennedy [2002], Valle et al. [2019a]. In this architecture, 1715

MLI forms the outermost thermal protection sub-assembly of a ve-layer softgoods stack, 1716

ordered (outer to inner) as: (1) BETA cloth exterior for atomic oxygen protection; (2) nylon- 1717

reinforced double-aluminised Mylar/Kapton MLI layers with perforated inner surfaces for 1718

venting during deployment; (3) Nextel/Kevlar stued-Whipple MMOD shield; (4) Vectran 1719

restraint layer carrying hoop and axial pressure loads; and (5) multi-redundant gas-tight 1720

bladder. The MLI sub-assembly in TransHab comprised over 20 individual reector layers 1721

with eective emittance on the order of 0.0150.05 Finckenor and Dooling [1999]. 1722

BEAM's on-orbit thermal performance has been characterised as more benign than 1723

predicted NASA Johnson Space Center [2017], an observation attributed to the additional 1724

insulation provided by folded softgoods layers that act as low-conductance barriers even 1725

when not specically designed as MLI. This nding has positive implications for inatable 1726

structure design: the inherent multi-layer nature of the fabric wall stack provides a degree 1727

of passive thermal buering beyond that of the dedicated MLI layers alone. 1728

10.2 The JWST Sunshield as Deployable Thermal Barrier Prece- 1729

dent 1730

The James Webb Space Telescope (JWST) sunshield is the largest deployed thermal barrier 1731

ever own and provides the benchmark for what large-area passive thermal isolation can 1732

achieve Arenberg et al. [2016]. At 21.2 m × 14.2 m (approximately 300 m2), the kite-shaped 1733

sunshield comprises ve layers of Kapton E polyimide membrane: Layer 1 (sun-facing) at 1734

50 µm thickness, Layers 25 at 25 µm. All layers are coated with 100 nm aluminium on both 1735

sides for reectivity; Layers 1 and 2 additionally carry 50 nm doped silicon on the sun-facing 1736

surface for enhanced emissivity and electrostatic discharge grounding. 1737

The thermal performance is extraordinary: the sun-facing side of Layer 1 reaches approx- 1738

imately +110 ◦C while the telescope-facing side of Layer 5 operates at −233 ◦Ca gradient 1739

of 343 ◦C across ve layers. Incoming solar power of approximately 200250 kW is attenu- 1740

ated to ∼23 mW transmitted to the cold side, an attenuation ratio of approximately 106:1 1741

Arenberg et al. [2016]. This performance is achieved through the V-groove geometry: angled 1742

layers radiate inter-layer thermal energy sideways to deep space through the vacuum gaps 1743

between membranes. 1744

However, the JWST sunshield is not an inatable structure. Layer separation is main- 1745

tained by six rigid spreader bars, with centre gaps of ∼2550 mm expanding to ∼250 mm 1746

at the edges. The deployment system required 139 of JWST's 178 release mechanisms, 400 1747

pulleys, 90 cables (∼0.5 km total), 8 motors, and 70 hinges Arenberg et al. [2016]. Table 18 1748

compares the JWST sunshield and TransHab shell architectures. 1749

Table 18: JWST sunshield versus TransHab inatable shell comparison.

Feature JWST Sunshield TransHab Shell

Primary function Thermal isolation Structural + MMOD + thermal Layer count 5 membranes 5 sub-assemblies (60+ layers) Layer material Kapton E (all 5) Vectran, Kevlar, Nextel, Mylar Structural role None (spreader bars) Vectran restraint carries pressure Energy attenuation 106:1 ∼150 ◦C gradient Deployment 139 mechanisms, 8 motors Ination pressure Deployed area 300 m2 220 m2 (cylinder)

It should be noted that JWST operates at the Sun-Earth L2 point, not in LEOthe 1750

thermal environment is fundamentally dierent (no orbital cycling, no atmospheric drag, no 1751

atomic oxygen), and this limits the direct applicability of JWST thermal performance num- 1752

bers to LEO inatable structures. Nevertheless, for inatable debris shields or large-area 1753

thermal barriers, the JWST heritage demonstrates that multi-layer Kapton stacks achieve 1754

extreme thermal gradients at 20+ metre scales. Adapting this concept to a fully inat- 1755

able deployment mechanismreplacing rigid spreader bars with ination-maintained layer 1756

separationremains an open engineering challenge. A hybrid approach combining inatable 1757

outer layers with rigid-bar-maintained inner separation represents a plausible intermediate 1758

architecture. 1759

10.3 Variable Emissivity Coatings and Smart Radiators 1760

Variable emissivity materials (VEMs) oer electronic louver functionality for dynamic ther- 1761

mal regulation without mechanical moving partsa capability uniquely suited to large inat- 1762

able surfaces where conventional mechanical louvers are impractical due to mass, complexity, 1763

and incompatibility with membrane substrates. Two technology families have received sus- 1764

tained development: passive thermochromic coatings and active electrochromic devices. 1765

Among passive thermochromic approaches, vanadium dioxide (VO2) based coatings are 1766

technically most advanced. Kim et al. Kim et al. [2019] performed the rst direct calorimetric 1767

measurement of a VO2-based switchable radiator in a simulated space environment (vacuum 1768

10−7 Torr, cold block at 108 K). Their multilayer structureSi substrate / VO2 (40100 nm) 1769

/ BaF2 dielectric spacer (1,500 nm) / Au back reector (200 nm)operates as a Fabry-Pérot 1770

resonant absorber. In the low-temperature insulating state (T < 340 K), hemispherical 1771

emissivity is εL = 0.16; above the phase transition (T > 340 K, metallic VO2), εH = 0.51, 1772

yielding ∆ε = 0.35. The practical consequence is a net radiated power dierence of 480 W/m2 1773

between 300 K and 373 Ka factor of 7× in radiative cooling capacity Kim et al. [2019]. 1774

The silicon substrate provides an incidental benet: protection of the VO2 lm from atomic 1775

oxygen erosion, addressing a known degradation mechanism. An earlier design by Hendaoui 1776

et al. Hendaoui et al. [2013] achieved a higher normal emissivity swing of ∆ε = 0.49 but 1777

without the atomic oxygen protection. 1778

The sole ight-demonstrated variable emissivity technology is the EclipseVEDTM elec- 1779

trochromic coating (Ashwin-Ushas Corporation), own on the MidSTAR-1 satellite in 2007, 1780

achieving TRL 78. EclipseVED operates by applying a low voltage (13 V) to an elec- 1781

trochromic polymer lm, switching emissivity across the range ε ≈0.190.90 in the 812 µm 1782

thermal infrared band. It requires no mechanical actuators, making it compatible with large- 1783

area application including inatable surfaces. The principal limitation is the requirement for 1784

a thin-lm conductor and electrical interconnects across the deployed areaa tractable but 1785

non-trivial integration challenge for inatable structures. 1786

Table 19 compares variable emissivity technologies. 1787

For the survey's inatable structures context, VEMs oer a path to autonomous ther- 1788

mal self-regulation: at high temperature (sunlit, electronics active), emissivity increases to 1789

reject heat; at low temperature (eclipse), emissivity decreases to conserve heat. This self- 1790

regulating behaviour eliminates active heaters in many scenarios, reducing power demand 1791

on power-constrained large inatable platforms. The principal barrier to inatable applica- 1792

tion is substrate compatibility: VO2 coatings currently require rigid silicon substrates, while 1793

EclipseVED has been demonstrated only on rigid aluminium panels. Developing these tech- 1794

nologies on exible polymer substrates (Kapton, polyimide) is a near-term research priority. 1795

Table 19: Variable emissivity coating technologies for spacecraft thermal control.

Technology ∆ε Tswitch Power TRL Flight Heritage

VO2 (Kim 2019) 0.35 (hemi.) 67 ◦C Zero 34 None VO2 (Hendaoui 2013) 0.49 (normal) 67 ◦C Zero 3 None EclipseVED (electrochromic) ∼0.71 Voltage ctrl 13 V 78 MidSTAR-1 (2007) MEMS louvers ∼0.8 (e.) Bimetal Zero 78 Multiple

10.4 Loop Heat Pipes for Deployed Structures 1796

Loop heat pipes (LHPs) are the preferred heat transport technology for active thermal 1797

systems in space, oering passive capillary-driven two-phase uid transport with zero pump 1798

power, distances up to several tens of metres, and heat loads up to 5+ kW per evaporator 1799

Maydanik [2005]. The capillary driving force is generated by a sintered porous wick conned 1800

to a compact evaporator body; vapour and liquid travel in separate smooth-wall transport 1801

lines. A compensation chamber at the evaporator provides thermal buering and enables 1802

active setpoint control to ±0.5 ◦C via low-power heaters (15 W). Working uids for space 1803

include ammonia (−40 to +70 ◦C, the standard), propylene (−60 to +50 ◦C, when ammonia 1804

freeze risk exists), and ethane (−100 to +30 ◦C) for cryogenic applications. 1805

LHP spaceight heritage extends over 35 years, beginning with the Granat astrophysics 1806

satellite in 1989 and encompassing over 30 systems own by 2005 across Russian, American, 1807

and European programmes Maydanik [2005]. The Hughes HS-702 communications satel- 1808

lite (1999) demonstrated the rst LHP-coupled deployable radiatorthe directly relevant 1809

precedent for inatable structures, as the LHP exible transport lines accommodated the 1810

mechanical hinge between the deployed radiator panel and the spacecraft bus. NASA's EOS 1811

Terra and Aqua missions, ICESat/GLAS, and Swift/BAT all employed LHP thermal control. 1812

For inatable habitats, LHPs are the natural technology for transporting waste heat 1813

from interior systems (avionics, crew metabolic load) to external deployable radiators. The 1814

exible transport lines can be routed through deployment hinges and accommodate the 1815

geometric changes between stowed and deployed congurations. Current single-evaporator 1816

LHP systems transport 50700 W in typical spacecraft congurations, with multi-loop archi- 1817

tectures providing aggregate capacities exceeding 10 kW for large platforms. The principal 1818

engineering challenge for inatable integration is the condenser interface: bonding or me- 1819

chanically attaching the condenser panel to the exible membrane requires a solution to the 1820

rigid-to-exible interface problem discussed in Section 12.3. 1821

10.5 Phase Change Materials in Fabric Layers: The TRL 23 Gap 1822

Phase change materials (PCMs) oer passive thermal buering by absorbing and releasing 1823

latent heat during orbital day/night transitions. For LEO inatable habitats experiencing 1824

90-minute thermal cycles, the most promising PCM candidates are n-eicosane (melting point 1825

36.4 ◦C, latent heat 247253 J/g) and n-octadecane (28.2 ◦C, 244 J/g) Diaconu et al. [2023]. 1826

PCM-based thermal control for rigid electronics enclosures has extensive spaceight her- 1827

itage spanning from Apollo Lunar Roving Vehicle battery management (1971) through Mars 1828

rovers (Spirit, Opportunity, Curiosity, Perseverance; TRL 9) and ISS experiments (TRL 56) 1829

Diaconu et al. [2023]. 1830

However, integration of PCMs into exible fabric layers for inatable structuresthe 1831

conguration needed to provide distributed thermal buering across large membrane areas 1832

remains at TRL 23. Five specic technical barriers have been identied: 1833

1. Microgravity containment: Liquid-phase PCM migrates freely in zero-g. Microen- 1834

capsulation (1100 µm capsules) addresses this at small scale, but capsule integrity 1835

during the fold/deploy lifecycle has not been tested for space-grade materials. 1836

2. Fold/deploy cycling: PCM-loaded fabric must survive hundreds to thousands of 1837

fold/deploy cycles without capsule rupturea requirement with no demonstrated so- 1838

lution in the space-qualied materials literature. 1839

3. Outgassing: PCM solvents and vapour can contaminate optical surfaces (solar cells, 1840

sensors). Space-qualied encapsulation that meets ASTM E595 outgassing require- 1841

ments has not been characterised for PCM-textile systems. 1842

4. Thermal conductivity: Raw paran PCMs have thermal conductivity k ≈0.2 W/(m·K) 1843

approximately 1,000× lower than aluminiumresulting in slow thermal response. Car- 1844

bon nanotube or graphene additives can improve conductivity to 15 W/(m·K) but at 1845

the cost of reduced fabric exibility and increased mass. 1846

5. Atomic oxygen interaction: PCM capsule shells (typically PMMA or gelatin) may 1847

erode under atomic oxygen ux in LEO, releasing PCM material and contaminating 1848

the local environment. 1849

Despite these barriers, the potential benet is substantial. A 1 kg/m2 layer of microencap- 1850

sulated n-eicosane would provide ∼250 J/g × 1,000 g/m2 = 250 kJ/m2 of thermal storage 1851

sucient to buer the rst ∼10 minutes of eclipse entry for a membrane with low thermal 1852

mass, signicantly reducing peak-to-trough temperature excursions. The technology needs 1853

a structured development programme analogous to what IRVE provided for exible thermal 1854

protection systems. 1855

11 State of the Art: Attitude and Orbit Control 1856

Attitude and orbit control for large inatable space structures is dominated by a single over- 1857

arching challenge: control-structure interaction (CSI). When structural exibility approaches 1858

or overlaps the attitude control bandwidth, conventional rigid-body AOCS designs become 1859

inadequate or unstable. For 100 m-class inatable structures, where the lowest natural fre- 1860

quencies may fall well below 0.1 Hz, CSI is not merely a complicationit is the central design 1861

driver. This section reviews the CSI challenge, the theoretical framework of gyroelastic body 1862

dynamics, the drag budget for large LEO structures, and the critical gap in AOCS theory 1863

for pressure-stabilised membranes. 1864

11.1 Control-Structure Interaction for Flexible Spacecraft 1865

CSI has been studied since the 1970s in the context of large space systems including the 1866

Solar Power Satellite concept, the Space Station, and large deployable antennas. For me- 1867

chanically sti structuresrigid trusses, mesh antennas, deployable solar arraysthe lowest 1868

structural modes typically fall in the 0.11 Hz range for 1030 m scale structures, and struc- 1869

tural damping ratios ζ ≈0.0010.005 are small but predictable Nicassio et al. [2022]. The 1870

standard approach is modal truncation and notch ltering: identify the structural modes, ex- 1871

clude them from the control bandwidth, and ensure adequate frequency separation between 1872

rigid-body and exible modes. 1873

For inatable (pressure-stabilised) structures, the CSI problem is qualitatively dierent 1874

in four respects. First, structural stiness is primarily provided by membrane tension arising 1875

from internal pressure (σhoop = pR/t for a cylindrical geometry) rather than material bending 1876

stiness, and this stiness changes if pressure is lost due to microleaks or thermal cycling. 1877

Second, the lowest natural frequencies scale inversely with structure size and can be ≪ 1878

0.1 Hz for 100 m-class structures, potentially falling within the AOCS bandwidth. Third, 1879

membranes cannot carry compressive stressthey wrinkle, creating local zones of nonlinear 1880

stiness that invalidate linear modal analysis. Fourth, actuator forces transmitted through a 1881

exible membrane diuse spatially rather than transmitting cleanly through a rigid structural 1882

path, degrading actuator-to-mode coupling. No paper in the published literature explicitly 1883

addresses AOCS for pressure-stabilised inatable structures at the 100 m scale. 1884

Angeletti et al. Nicassio et al. [2022] developed a minimum complexity hybrid ODE- 1885

PDE model for large exible spacecraft that provides a useful methodological template: the 1886

rigid bus is treated as an ODE system (6 DOF) coupled to the exible structure as a PDE 1887

system (beam/plate). Even a 2-mode truncation captured over 80% of the relevant dynamics 1888

for control design. However, this framework assumes conventional bending stiness and is 1889

not directly applicable to pressure-stabilised membranes. 1890

11.2 Gyroelastic Body Theory and Distributed Momentum Man- 1891

agement 1892

The theoretical foundation for distributed attitude actuators on exible structures was estab- 1893

lished by D'Eleuterio and Hughes in a series of foundational papers D'Eleuterio and Hughes 1894

[1984, 1986, 1987]. The 1984 paper introduced the concept of gyricitythe distribution of 1895

angular momentum per unit volume embedded within an elastic continuum. The governing 1896

equations couple elastic deformation to rigid-body rotation through the gyricity distribu- 1897

tion g(x), showing that distributed angular momentum fundamentally modies elastic wave 1898

propagation and natural frequencies. The key theoretical nding is that gyroelastic systems 1899

have complex eigenvalues (gyroelastic frequency splitting), providing passive damping-like 1900

behaviour without explicit energy dissipationanalogous to Zeeman splitting in quantum 1901

mechanics D'Eleuterio and Hughes [1984]. The 1986 companion paper D'Eleuterio and 1902

Hughes [1986] derived the modal parameters (complex mode shapes, orthogonality condi- 1903

tions, participation factors) needed for practical numerical analysis, while the 1987 paper 1904

D'Eleuterio and Hughes [1987] extended the framework to complete spacecraft systems, 1905

treating a vehicle with distributed angular momentum storage as a unied gyroelastic body. 1906

Damaren and D'Eleuterio Damaren and D'Eleuterio [1989] solved the optimal gyricity 1907

distribution problem using calculus of variations: the spatial distribution g∗(x) that min- 1908

imises a quadratic performance index concentrates angular momentum where modal kinetic 1909

energy is highestat the antinodes of the dominant vibration modes. This is directly analo- 1910

gous to collocating sensors at modal antinodes and provides the theoretical basis for actuator 1911

placement optimisation on large exible structures. 1912

The most recent quantitative validation of distributed momentum management was pro- 1913

vided by Cachim et al. Cachim et al. [2024], who compared centralized (6 large reaction 1914

wheels on the bus) versus distributed (33 small reaction wheels throughout the structure) 1915

attitude control for a ∼30 m hexagonal plate-like structure (4,200 kg, Jxx = 2.2×105 kg·m2). 1916

Using LQG control with 25 retained modes below 80 Hz, the distributed conguration 1917

achieved 3.3× faster settling (30 s versus 100 s), 7× less structural deformation (0.33 µm 1918

versus 2.3 µm) during a 0.5◦slew, and improved ne pointing (RMS error 0.038 versus 1919

0.068 arcsec), at the cost of approximately 2× more total torque Cachim et al. [2024]. The 1920

structure was modelled as a Kirchho plate (bending-only, shear neglected), which is valid 1921

for thin plates with thickness-to-span ratio >1:30 but is not applicable to pressure-stabilised 1922

membranes. 1923

11.3 Drag Budget for 100 m-Class LEO Structures 1924

A 100 m-class inatable structure in LEO faces a severe drag penalty due to its extreme area- 1925

to-mass ratio. At 500 km altitude, atmospheric density varies from ρ ≈5×10−13 kg/m3 (solar 1926

minimum) to ρ ≈3×10−12 kg/m3 (solar maximum)a factor of 6× variation driven by solar 1927

EUV heating of the upper atmosphere Jiang et al. [2022], Andreussi et al. [2022]. For a 100 m 1928

diameter circular membrane presented broadside to the velocity vector (Ae ≈7,850 m2), the 1929

drag force FD = 1

2ρv2CDA yields the estimates in Table 20. 1930

Table 20: Drag force estimates for a 100 m inatable structure at 500 km altitude. CD ≈2.4 3.2 for at membrane in free molecular ow with atomic oxygen accommodation.

Scenario ρ (kg/m3) Ae (m2) CD FD (N)

Solar min, edge-on 5 × 10−13 100 2.4 0.007 Solar min, broadside 5 × 10−13 5,000 2.4 0.35 Solar min, broadside (max) 5 × 10−13 7,850 3.2 0.72 Solar max, broadside 3 × 10−12 5,000 3.2 14 Solar max, broadside (max) 3 × 10−12 7,850 3.2 21

The drag coecient range of CD = 2.43.2 for a at membrane in free molecular ow is 1931

based on the standard models of Sentman Sentman [1961] and Moe and Moe Moe and Moe 1932

[2005], where the upper bound corresponds to complete diuse reection with full thermal 1933

accommodation on atomic oxygen surfaces. 1934

The area-to-mass ratio is the fundamental problem: if the 100 m structure totals 5,000 kg, 1935

A/m ≈1.6 m2/kg (broadside), compared to ∼0.02 m2/kg for the ISSapproximately 80× 1936

higher. Using the ballistic coecient β = m/(CDA), the orbital decay time at 500 km during 1937

100 m diameter structure, CD = 2.2

Broadside, solar max

103

Broadside, solar min

500 km ref.

Edge-on, solar min

SRP reference (0.035 N)

102

101

3.64 N

100

Drag force (N)

0.28 N

10 1

0.02 N

10 2

10 3

Conventional

10 4

spacecraft drag range

10 5

10 6

300 400 500 600 700 800 Altitude (km)

Figure 11: Drag force versus altitude for a 100 m diameter inatable structure in LEO, showing solar minimum and solar maximum atmospheric conditions. The shaded region illustrates the factor-of-six variation in atmospheric density driven by the solar cycle, which dominates the orbit maintenance propellant budget.

solar maximum can be estimated at approximately 36 months for broadside orientation, 1938

conrming that orbital lifetime without propulsion would be months, not years. 1939

Second-Order Eects 1940

Three additional forces merit consideration for a complete 100 m-class force budget: 1941

Solar radiation pressure (SRP): For a 100 m diameter membrane at 500 km, the SRP 1942

force is FSRP = (P⊙/c) · A · (1 + r) ≈(4.56 × 10−6 N/m2) × 7,850 m2 × 1.5 ≈0.054 N, where 1943

P⊙= 1,361 W/m2 is the solar ux, c is the speed of light, and r ≈0.5 is the reectivity. 1944

This SRP force is comparable to the drag at solar minimum edge-on (0.007 N) and non- 1945

negligibleat solar minimum with edge-on orientation, SRP may actually dominate over 1946

atmospheric drag. 1947

Attitude-dependent cross-section: The table presents edge-on (100 m2) and broad- 1948

side (7,850 m2) as discrete cases, but a real membrane oscillates between attitudes unless 1949

actively controlled. The time-averaged eective area depends on AOCS capabilitycoupling 1950

the drag analysis to the AOCS gap (C4). Passive spin stabilisation about the minimum- 1951

inertia axis would yield a time-averaged Aeff intermediate between edge-on and broadside, 1952

approximately 0.5×Abroadside ≈3,900 m2, roughly halving the broadside drag but still orders 1953

of magnitude above edge-on. 1954

Propellant mass rate derivation: The xenon propellant consumption for Hall thruster 1955

drag compensation can be derived as ˙m = FD/(g0Isp), where g0 = 9.81 m/s2 and Isp = 3,000 s 1956

for a representative Hall thruster. For the solar-minimum broadside case (FD = 0.35 N): 1957

˙m = 0.35/(9.81 × 3,000) = 1.19 × 10−5 kg/s = 1.03 kg/day = 376 kg/year. For the solar- 1958

maximum broadside case (FD = 21 N): ˙m = 21/(9.81 × 3,000) = 7.14 × 10−4 kg/s = 1959

61.7 kg/dayclearly unsustainable without in-orbit refuelling. The corresponding thrust 1960

power is P = FDve/(2η), where ve = g0Isp = 29,430 m/s and η = 0.6 (thruster eciency): 1961

yielding 8.6 kW for the solar-minimum broadside case and 515 kW for the solar-maximum 1962

broadside case. The 150 kW range stated in Section 13.2 corresponds to solar-minimum 1963

conditions with partial edge-on attitude control. 1964

Air-Breathing Electric Propulsion (ABEP), which collects atmospheric gas for use as 1965

propellant, has been proposed for drag compensation in Very Low Earth Orbit (VLEO, 1966

150450 km) Andreussi et al. [2022]. However, at 500 km the atmospheric density is approx- 1967

imately 100× lower than at the 250350 km altitudes where ABEP is designed to operate, 1968

reducing achievable thrust to 0.0010.1 mNorders of magnitude insucient for the 0.35 1969

21 N drag forces computed above. Conventional electric propulsion (Hall eect or gridded 1970

ion thrusters) with onboard xenon propellant is the only viable station-keeping option. This 1971

propulsion requirement fundamentally constrains mission architecture and represents a sig- 1972

nicant fraction of the overall mass budget. 1973

11.4 The Missing Theory: AOCS for Pressure-Stabilised Mem- 1974

branes 1975

The gyroelastic body framework of D'Eleuterio and Hughes assumes elastic continua with 1976

Cauchy stress tensor constitutive relationsvalid for beams, plates, and shells with inher- 1977

ent bending stiness. Extending this framework to pressure-stabilised inatable membranes 1978

requires four theoretical modications that represent a signicant gap in the published lit- 1979

erature: 1980

1. Pressure-stiness coupling: For an inatable structure, the eective stiness Ke = 1981

Kmembrane + Kpressure, where the pressure contribution depends on ination state and 1982

couples to deformation through the ideal gas law. When pressure changes due to mi- 1983

croleaks or thermal cycling, natural frequencies shift and gyroelastic modes recongure 1984

a time-varying system for which xed-gain controllers may become unstable. 1985

2. Wrinkling constraint: Membranes cannot carry compressive stress; they wrinkle, 1986

creating zones where σn = max(0, Tmembrane · εn). This state-dependent nonlinearity 1987

causes mode shapes to change with the deformation state, invalidating the linear modal 1988

analysis assumption that underpins both the D'Eleuterio framework and the Cachim 1989

optimisation. 1990

3. Orthotropic fabric constitutive model: Space fabrics (Vectran, Kevlar) are woven 1991

materials with highly anisotropic stinesswarp versus weft direction stiness can 1992

dier by 25×. The isotropic elastic continuum in the D'Eleuterio formulation requires 1993

replacement with an orthotropic constitutive model. 1994

4. Gas-structure interaction coupling: For large inatable volumes, internal gas has 1995

its own dynamics (acoustic modes, pressure wave propagation). This is analogous to 1996

liquid sloshing in fuel tanksa well-studied problembut the gas-structure coupling 1997

for inatable membranes has received no published treatment. 1998

Each of these extensions builds upon established prior work, and the timeline can be 1999

estimated with some granularity: 2000

ˆ Pressure-stiness coupling (estimated 34 years): The coupling of ination pressure 2001

to membrane stiness is well-understood for simple geometries through the gossamer 2002

structure dynamics literature Jenkins [2001]. The novel challenge is coupling this to the 2003

gyroelastic formulation, requiring a pressure-dependent constitutive model within the 2004

D'Eleuterio framework. This is the most tractable extension and could be addressed 2005

within a focused doctoral programme. 2006

ˆ Wrinkling constraint (estimated 34 years): Tension-eld theory Stein and Hedgepeth 2007

[1961] provides a well-established framework for membranes that cannot sustain com- 2008

pression. Roddeman et al. Roddeman et al. [1987] developed the modern computa- 2009

tional treatment. Integrating wrinkling-induced state-dependent stiness into gyroe- 2010

lastic eigenvalue analysis is non-trivial but has analogues in rotor dynamics (cracked 2011

shaft models with breathing cracks). 2012

ˆ Orthotropic fabric constitutive model (estimated 12 years): Replacing isotropic 2013

with orthotropic constitutive relations requires substituting the appropriate fourth- 2014

order stiness tensor into the D'Eleuterio equations. The D'Eleuterio formulation uses 2015

the general Cauchy stress tensor, making the extension algebraically systematic. This 2016

is the most tractable extension and could constitute the early phase of a doctoral 2017

programme or a Master's thesis. 2018

ˆ Gas-structure interaction coupling (estimated 45 years): This is the most novel 2019

and uncertain extension. The fuel-sloshing analogy Abramson [1966] is useful but 2020

incompletegas is compressible while classical sloshing models assume incompressibil- 2021

ity. Coupled gas-membrane problems have been studied in the aeroelasticity literature 2022

(utter of inated membrane wings Leclercq and de Langre [2018]), providing a starting 2023

point, but the three-dimensional coupling for large inatable volumes in the gyroelastic 2024

context has no precedent. This is the genuine multi-year research challenge. 2025

The total estimated timeline is 1215 years if pursued sequentially by individual doctoral 2026

candidates, or 57 years if pursued in parallel by a coordinated research group with 23 2027

concurrent doctoral projects. The sequential estimate of 1015 years stated in Section 13 2028

is therefore conservative but reasonable. This is among the most signicant fundamental 2029

research gaps identied in this survey. 2030

12 State of the Art: Robotic In-Orbit Assembly 2031

The vision of large inatable space structures100 m-class debris shields, large-aperture 2032

antenna reectors, or orbital habitats exceeding ISS volumewill likely require in-orbit 2033

assembly of subsystems that exceed the launch vehicle fairing envelope or are too complex 2034

for single-deployment architectures. This section reviews the state of in-space servicing, 2035

assembly, and manufacturing (ISAM) robotics, the E-Walker concept for walking robots on 2036

large structures, and the critical gap in rigid-to-exible interface technology that currently 2037

prevents assembly on inatable substrates. 2038

12.1 Assembly Robot Heritage and Current Programmes 2039

In-orbit robotic assembly heritage begins with the ISS, whose construction (19982011) relied 2040

on the Canadarm2 Space Station Remote Manipulator System (SSRMS): a 17.6 m, 7-DOF 2041

arm operating from xed Power Data Grapple Fixtures (PDGFs) on the truss structure. 2042

Canadarm2 demonstrated that large-scale orbital assembly is achievable with telerobotic 2043

systems, but at the cost of extensive EVA support and ground-in-the-loop operations. 2044

The ISAM landscape has expanded substantially since ISS assembly. NASA's 2025 State 2045

of Play report catalogues 524 capability entries across 145 developers in 21 countries, with 2046

over $2 billion in government investment NASA [2025]. Current programmes span mul- 2047

tiple technology readiness levels: GITAI's S2 experiment demonstrated autonomous ISS 2048

solar array assembly (2021); Project GHOST validated tool manipulation in orbit (2024); 2049

DARPA's NOM4D programme targets LEO truss assembly demonstration by Caltech in 2050

2026; and NASA Langley's CIRAS/TALISMAN/SAMURAI/NINJAR ground demonstra- 2051

tions have validated multi-robot truss assembly at 15 m scale Li et al. [2022c], Doggett et al. 2052

[2018]. The European PULSAR project targets autonomous assembly of a 12 m telescope 2053

mirror Rognant et al. [2019]. Northrop Grumman's MEV-1 (2020) and MEV-2 (2021) rep- 2054

resent the rst commercial ISAM operations, though these are servicing (docking with client 2055

spacecraft) rather than structural assembly. 2056

A critical observation for the present survey is that all 524 entries in the NASA ISAM 2057

catalogue address assembly of rigid structurestrusses, beams, modular satellites, and mir- 2058

ror segments NASA [2025]. Not a single entry addresses assembly on or of inatable/exible 2059

substrates. This is not a mere omission; it reects a fundamental gap in the technology base: 2060

the rigid-to-exible interface problem remains unsolved (Section 12.3). 2061

12.2 Walking Robots for Large Structure Assembly: E-Walker 2062

The End-over-End Walking Robot (E-Walker) represents the current state of the art in 2063

walking manipulators designed for ISAM missions Nair et al. [2022, 2024]. Inheriting the 2064

Canadarm2 design philosophy of end-over-end locomotion via grapple xtures, the E-Walker 2065

is a 7-DOF dexterous manipulator at full scale of approximately 475 kg with 350 kg pay- 2066

load capacitysucient to handle one primary mirror segment for a 25 m Large Aperture 2067

Space Telescope (LAST). Maximum joint torque reaches ∼70 Nm at Joint 2, and nite ele- 2068

ment analysis conrms maximum link deection of only 0.04 mm under full payload, with a 2069

buckling safety factor exceeding 129 Nair et al. [2022]. 2070

A scaled prototype (1.3 m, 12 kg, 2 kg payload at 1:6 scale) has been demonstrated in 2071

ground testing. Nair et al. Nair et al. [2024] evaluated 11 concepts of operations for 25 m 2072

telescope assembly, concluding that a dual E-Walker conguration is optimal. The 8 m E- 2073

Walker requires 4.5 m less workspace than an equivalent xed-base arm, making walking 2074

locomotion particularly advantageous for assembly tasks distributed over large structures. 2075

However, all E-Walker analysis assumes a rigid assembly substrate. The grapple x- 2076

tures are ISS-standard PDGFs requiring rigid interfaces with ±10 mm capture tolerance and 2077

multi-kN load capacity. When the E-Walker applies 70 Nm joint torques during assembly 2078

operations, Newton's third law transmits equal and opposite reactions into the mounting 2079

substrate. On the ISS rigid truss, these are absorbed globally; on an inatable membrane, 2080

they would cause local deformation, potential wrinkling, and excitation of global vibration 2081

modes. The 475 kg robot's every movement in microgravity creates reaction forces that, on 2082

a exible membrane, propagate as structural disturbances. 2083

12.3 The Rigid-to-Flexible Interface Gap 2084

All existing docking and assembly interfaces assume rigid-to-rigid connections. Chen et al. 2085

Liu et al. [2024] designed an androgynous docking port with ±23.5 mm translation tolerance 2086

for on-orbit assemblya practical engineering specication for robotically-assisted mating 2087

of rigid modules. ISS Power Data Grapple Fixtures, common berthing mechanisms, and all 2088

ISAM interface concepts in the literature share this rigid-to-rigid assumption. 2089

No published work specically addresses distributed rigid-module attachment to inat- 2090

able membranes in the space environment. However, several bodies of adjacent work provide 2091

relevant design heritage that should be acknowledged: 2092

ˆ Tensegrity structures: Tensegrity platforms Skelton and de Oliveira [2009] inher- 2093

ently address the rigid-to-exible interface through bar-cable connections. NASA 2094

Ames' Super Ball Bot Sabelhaus et al. [2015] demonstrates rigid node attachment to 2095

tensioned cables in a recongurable structure; the load-spreading problem at hardpoint- 2096

membrane interfaces is structurally analogous to the bar-cable joint in tensegrity. 2097

ˆ Deployable antenna feed support: Large deployable mesh antennas (Harris/L3 2098

AstroMesh, Northrop Grumman CRAF) attach a rigid feed assembly to a tensioned 2099

cable-net/mesh reector surface Santiago-Prowald and Rodrigues [2018]. The feed 2100

support struts connect rigid hardware to a exible, tension-stabilised structurea 2101

direct analogue to the rigid-module-on-inatable-membrane problem. 2102

ˆ Solar sail boom-membrane attachment: Solar sail designs (e.g., IKAROS, NEA 2103

Scout) attach rigid booms to thin-lm membranes via reinforced corner ttings. The 2104

stress concentration and load distribution at these attachment points have been anal- 2105

ysed in the solar sail literature Fernandez et al. [2014]. 2106

The gap remains genuine: none of these analogues addresses the full combination of vac- 2107

uum, thermal cycling, atomic oxygen, micrometeoroid exposure, and zero-gravity dynamics 2108

on an inatable pressure-stabilised substrate. The adjacent work provides starting points 2109

for analysis but not validated solutions. 2110

Table 21 summarises the technology readiness of assembly interface approaches. 2111

The closest ight analog is the BEAM-ISS interface: a rigid berthing ring connects the 2112

inatable module to the ISS Node 3 (Tranquility) common berthing mechanism. This is a 2113

single rigid-to-inatable joint at the berthing interface, not a distributed attachment system 2114

across the membrane surface. No demonstrated technology exists for attaching multiple rigid 2115

Table 21: Assembly interface technology readiness for space structures.

Interface Type TRL Heritage Notes

Rigid-to-rigid (PDGF) 9 ISS Operational since 2001 Rigid-to-rigid (androgynous) 34 Ground demo Chen et al. 2024 Rigid-to-exible (hardpoint) 23 BEAM ring Conceptual only Rigid-to-exible (distributed) 12 None No published work

subsystems (reaction wheels, solar array drives, communications antennas) to an inatable 2116

membrane at distributed locations. This is a novel nding of this survey and represents a 2117

critical research gap. 2118

12.4 Assembly-Enabled Inatable Platforms: Design Requirements 2119

Based on the analysis in Sections 12.212.3, a set of design requirements for assembly-enabled 2120

inatable platforms can be identied: 2121

1. Embedded rigid attachment rings: Metallic rings (0.51 m diameter) must be sewn 2122

into the inatable fabric at pre-determined assembly points during manufacturing, with 2123

integrated load-spreading plates to distribute reaction forces over sucient membrane 2124

area. The stress concentration factor at such embedded hardpoints (25× local stress 2125

amplication) must be accounted for in the membrane structural design. 2126

2. Compliance layer: A 35 mm silicone or elastomeric foam layer between each rigid 2127

attachment ring and the membrane accommodates local deformation and provides 2128

vibration isolation, preventing point-load damage to the fabric. 2129

3. Pre-integration requirement: Retrotting hardpoints onto an already-deployed 2130

inatable is impractical. All assembly interfaces must be designed in and manufactured 2131

as part of the inatable structure before launch. This implies that the assembly concept 2132

of operations must be fully dened before the inatable is manufactureda signicant 2133

systems engineering constraint. 2134

4. Active vibration isolation: Small dampers or isolation mounts between each E- 2135

Walker grapple point and the membrane surface attenuate reaction forces from assem- 2136

bly operations, reducing excitation of global membrane vibration modes. 2137

5. Pressure-aware operations: Assembly operations that change the mass distribu- 2138

tion (adding subsystems) alter both the inertia tensor and the natural frequencies of 2139

the inatable structure. AOCS must accommodate these time-varying dynamics 2140

connecting to the gap identied in Section 11.4. 2141

The E-Walker on an inatable platform is conditionally feasible with pre-integrated hard- 2142

points, compliance layers, and active vibration isolation. However, none of these solutions has 2143

been demonstrated even at component level for space applications. A ground demonstration 2144

programmeanalogous to NASA Langley's CIRAS/TALISMAN truss assembly demonstra- 2145

tions but on an inatable test articlewould represent a signicant advance toward closing 2146

this gap. 2147

13 Challenges, Open Questions, and Research Roadmap 2148

The preceding eight technology surveys (Sections 512) have documented a paradox that 2149

denes the current state of soft inatable robotic systems for space: individual enabling tech- 2150

nologies have reached moderate-to-high readiness levelsVectran restraint layers at TRL 9 2151

(Section 5), shape memory alloy deployment actuators at TRL 89 (Section 7.5), bre Bragg 2152

grating sensors on rigid spacecraft at TRL 78 (Section 8.1)yet no integrated soft inat- 2153

able robotic system has been demonstrated in space. This section consolidates the research 2154

gaps identied throughout the survey, assesses their severity and interdependence, proposes 2155

a structured research roadmap spanning 5-year and 15-year horizons, and identies the most 2156

viable path to a near-term ight demonstration. 2157

13.1 Critical Research Gaps 2158

A systematic analysis of the technology areas reviewed in Sections 512 reveals 5 critical 2159

gaps, 9 moderate gaps, and 10 minor gaps. Here we consolidate the 5 critical gaps, each of 2160

which represents a showstopper for at least one major application domain. 2161

C1: Absence of Quantitative Soft-versus-Rigid Fragmentation Comparison. The 2162

central motivation for soft capture in active debris removal (Section 3.2) rests on the propo- 2163

sition that compliant mechanisms reduce fragmentation risk relative to rigid robotic arms. 2164

Qualitative evidence supports this argument: Wang et al. Wang et al. [2023] identied the 2165

potential to generate fragments during the capturing phase for rigid systems; Chen et 2166

al. Chen et al. [2024a] concluded that single contact-based caging is excessively risky for 2167

fast-tumbling targets; and the RemoveDebris harpoon test demonstrated structural fail- 2168

ure of a carbon bre boom at 20 m s−1 impact Aglietti et al. [2020]. The e.deorbit mission 2169

study computed peak joint torques of 195 N m for capture of an 8-tonne ENVISAT tumbling 2170

at 5 ◦s−1 Flores-Abad et al. [2014]. However, no published study provides a quantitative 2171

fragmentation probability as a function of contact compliance. The catastrophic fragmenta- 2172

tion threshold (10 J g−1 specic energy from the IMPACT model Johnson et al. [2001]) has 2173

never been applied to a soft-versus-rigid capture force comparison. The fragmentation risk 2174

is physically plausible and supported by qualitative assessmentsparticularly for degraded 2175

appendages (solar panels, thermal blankets, antennas) that may have lost 3060% of their 2176

original strength through decades of space environment exposurebut remains experimen- 2177

tally unquantied. This survey adopts the precautionary principle that compliant capture 2178

is preferred until quantitative data become available, on the basis that the consequences of 2179

inadvertent fragmentation are severe enough to warrant risk-averse technology selection. We 2180

propose this as the single highest-priority experimental investigation the community should 2181

undertake, requiring hypervelocity and low-velocity impact testing with debris surrogates at 2182

varying contact compliance levels. 2183

C2: No Soft Robotic Capture System Has Flown in Space. Despite eight distinct 2184

soft or compliant capture approaches documented in Section 3.2gecko adhesive (TRL 4 2185

5), DEMES grippers (TRL 34), bistable soft grippers (TRL 23), cryogenic metallic cable 2186

robots (TRL 3), inatable origami arms (TRL 3), ytrap origami (TRL 23), thermally 2187

qualied multi-layer grippers (TRL 2), and the INSIDeR system concept (TRL ∼4)none 2188

has own. The gecko adhesive gripper of Jiang et al. Jiang et al. [2017] achieved microgravity 2189

validation with 100% capture success rate on spherical targets and capacity exceeding 400 kg, 2190

making it the most mature candidate. However, this gripper operates on a rigid robotic arm 2191

platform and is more accurately classied as a compliant end-eector on a conventional 2192

manipulator (Section 3.1). The gap between ground/parabolic-ight demonstration and or- 2193

bital ight requires addressing space environment qualication (vacuum outgassing, thermal 2194

cycling, radiation exposure over mission-duration timescales) for which limited data exist. 2195

C3: Rigid-to-Flexible Assembly Interface Lacks Specic Published Research. 2196

Section 12.3 identied that no published work specically addresses distributed rigid-module 2197

attachment to inatable membranes in the space environment, though adjacent work in 2198

tensegrity structures Skelton and de Oliveira [2009], Sabelhaus et al. [2015], deployable an- 2199

tenna feed supports Santiago-Prowald and Rodrigues [2018], and solar sail boom-membrane 2200

attachments Fernandez et al. [2014] provides relevant design heritage. All heritage docking 2201

interfacesISS PDGF, Common Berthing Mechanism, ClearSpace-1 capture arms, and the 2202

androgynous interfaces reviewed by Chen et al. Chen et al. [2024b]assume rigid-to-rigid 2203

mating. At the 100-metre scale required for large inatable debris shields (Section 11.3) or 2204

solar power platforms, the inatable structure becomes a platform onto which functional 2205

modules must be assembled in orbit Nair et al. [2024], Li et al. [2019]. The reaction force 2206

problemhow to apply assembly torques to a membrane that deforms under the applied 2207

loadhas no published solution specic to the space inatable context. Embedded metallic 2208

hardpoint rings represent a plausible design concept informed by the tensegrity and antenna 2209

feed analogues, but require detailed nite element analysis of stress concentration at the 2210

rigid-exible interface, none of which has been published. 2211

C4: No Published AOCS Theory for Pressure-Stabilized Inatable Structures. 2212

The control-structure interaction literature reviewed in Section 11.1 addresses rigid trusses, 2213

mesh antennas, and mechanically stiened deployable arraysstructures with inherent sti- 2214

ness independent of pressurization. Pressure-stabilized inatable structures exhibit funda- 2215

mentally dierent dynamics: stiness is a function of ination pressure (a time-varying pa- 2216

rameter), membranes wrinkle under compression introducing piecewise-linear stiness non- 2217

linearity, fabric is anisotropic, and internal gas couples to structural modes D'Eleuterio 2218

and Hughes [1984], Jenkins [2001]. The D'EleuterioHughes gyroelastic body framework 2219

D'Eleuterio and Hughes [1984, 1986, 1987] provides the most promising theoretical founda- 2220

tion, but requires four extensions: (i) pressure-dependent constitutive model for membrane 2221

elements, (ii) wrinkling constraints reecting piecewise-linear stiness transitions, (iii) or- 2222

thotropic fabric constitutive laws, and (iv) gas-structure coupling for internal atmosphere 2223

dynamics. Each extension constitutes a substantial theoretical undertaking; collectively they 2224

dene a research programme of 1015 years. 2225

C5: Inatable-Power Integration Gap. The PowerSphere programme (Section 9.2) 2226

demonstrated thin-lm photovoltaic integration with an inatable substrate using amor- 2227

phous silicon cells, achieving 7.25 W kg−1 at 10% cell eciency Cadogan et al. [2003]. The 2228

programme has been inactive since approximately 2009, and no successor has been identied. 2229

Meanwhile, perovskite/CIGS tandem cells have achieved 2100 W kg−1 with 25 µm substrates 2230

and greater than 85% power retention after more than 50 years of LEO-equivalent proton irra- 2231

diation Lang et al. [2020]. The technology exists to revive inatable-integrated photovoltaics 2232

at 20300× the specic power of the original PowerSphere, yet no programme is pursuing 2233

this integration. The gap is institutional rather than technical: exible PV researchers and 2234

inatable structure researchers operate in separate communities with no overlap programme. 2235

Vectran restraint Kevlar MMOD Nextel bumper Zylon (interior) SMP rigidisation

Materials &

Structures

BEAM inflation InflateSail LOFTID Origami packaging Active controlled

Deployment

Mechanics

SMA hinges Tendon-driven DEA/DEMES Vacuum-gap electrostatic Jamming (vacuum)

Actuation

FBG (heritage) FBG in webbing Multicore FOSS Distributed impact Capacitive soft

Sensing &

SHM

ROSA/iROSA CIGS thin-film

Power Systems

Li-ion batteries Perovskite PV PowerSphere-type

MLI (heritage) JWST sunshield VO2 coatings LHP deployed PCM in fabric

Thermal Management

CMG (heritage) Distributed CMG

AOCS

EP drag comp. CSI for inflatables

Gyroelastic body

Canadarm2 Walking robots

In-Orbit Assembly

Docking i/f Rigid-flex i/f

Autonomous assembly Concept (TRL 1-3)

Flight proven

Validated (TRL 4-6)

(TRL 7-9)

1 2 3 4 5 6 7 8 9 Technology Readiness Level (TRL)

Figure 12: Technology readiness landscape across the eight enabling technology areas re- viewed in Sections 512. Each marker represents a specic sub-technology; colour indicates TRL band (red: concept TRL 13; orange: validated TRL 46; green: ight-proven TRL 7 9). While heritage components (Vectran, FBG, ROSA, MLI, Canadarm2) have reached TRL 79, the integrative technologies required for soft inatable robotic systemsvacuum- gap actuators, jamming in vacuum, rigid-to-exible interfaces, distributed momentum man- agement, and PCM in fabricremain at TRL 23.

13.2 Integration Challenges at System Level 2236

Beyond individual technology gaps, the fundamental barrier to ight-ready soft inatable 2237

robotic systems is system integration. The preceding sections documented integration decits 2238

across multiple interfaces: 2239

ˆ ActuationStructure: Vacuum-gap electrostatic actuators (Section 7.2) achieve >4 N 2240

force at 0.7 g mass Sirbu et al. [2025] using thin-lm polymer multilayer construction 2241

that is structurally analogous to inatable membrane wall architecturesyet no study 2242

has attempted to laminate actuator layers into an inatable arm liner. Similarly, the 2243

jamming-in-vacuum concept (Section 7.6) has a sound physical basis Fitzgerald et al. 2244

[2020] but zero experimental validation in relevant conditions. 2245

ˆ SensingStructure: FBG sensors woven into Vectran webbing have been demon- 2246

strated at NASA JSC on 0.61 m and 2.74 m test articles (TRL 45) Bally Ribbon 2247

Mills and Luna Innovations [2020], while multicore bre optic shape sensing achieves 2248

0.64 mm position accuracy in soft actuators Galloway et al. [2019]. The same FBG 2249

technology could provide both structural health monitoring for inatable walls and 2250

proprioceptive sensing for inatable robotic armsa unied sensing architecture that 2251

has not been proposed or demonstrated. 2252

ˆ PowerThermalStructure: A large inatable membrane with thin-lm PV on the 2253

sun-facing surface, MLI on the space-facing surface, and variable-emissivity coatings 2254

for thermal regulation represents a multi-functional surface that would merge the power 2255

and thermal subsystems into a single membrane layer. The PowerSphere concept ap- 2256

proached this integration using 2004-era materials Cadogan et al. [2003]; 2025-era per- 2257

ovskite/CIGS cells on Kapton or Mylar substrates would share the same polymer base 2258

as inatable MLI layers Lang et al. [2020], making the integration pathway plausible. 2259

ˆ AOCSDeployment: BEAM's deployment anomaly (25 ination bursts over 7 hours; 2260

Section 6.3) illustrates that deployment is a dynamic event with angular momen- 2261

tum consequences. For a free-ying 100-metre inatable, each ination pulse imparts 2262

momentum to the structure, and as the structure changes shape during deployment 2263

its modal frequencies shiftpotentially crossing into the AOCS controller bandwidth 2264

D'Eleuterio and Hughes [1984]. No published work addresses the coupled deployment 2265

AOCS problem for inatables. 2266

ˆ DragPowerThermal Cascade: At 500 km altitude, a 100-metre broadside inat- 2267

able experiences drag forces of 0.3521 N depending on solar activity (Section 11.3). To 2268

illustrate the cascade quantitatively, consider a worked example for the solar-minimum 2269

broadside case (FD = 0.35 N) and the solar-maximum broadside case (FD = 21 N): 2270

Step 1  Thrust: Hall thruster at Isp = 3,000 s, exhaust velocity ve = g0Isp = 2271

29,430 m/s. 2272

Step 2  Power: Pthrust = FDve/(2η) where η = 0.6. Solar-min broadside: P = 2273

0.35 × 29,430/1.2 = 8.6 kW. Solar-max broadside: P = 21 × 29,430/1.2 = 515 kW. 2274

Step 3  Solar array: At 300 W m−2 (BOL, triple-junction) and 100 W kg−1 system- 2275

level specic power: solar-min requires 29 m2 / 86 kg; solar-max requires 1,717 m2 / 2276

5,150 kgexceeding the entire platform mass budget. 2277

Step 4  Waste heat: At 40% combined losses (thruster + PPU): solar-min generates 2278

3.4 kW waste; solar-max generates 206 kW waste. 2279

Step 5  Radiator: At 200 W m−2 radiator capacity: solar-min requires 17 m2; solar- 2280

max requires 1,030 m2. 2281

This cascade demonstrates that the solar-maximum broadside scenario is infeasible 2282

without active attitude control to reduce Aeff, conrming that drag budget and AOCS 2283

capability are inextricably coupled. Edge-on operation at solar minimum (0.007 N 2284

drag, ∼0.17 kW power, <1 m2 array) is feasible; all other scenarios require either ac- 2285

tive attitude management, altitude selection, or both. The 150 kW range previously 2286

stated applies to solar-minimum conditions with partial attitude control. No pub- 2287

lished analysis traces this full cascade end-to-end for inatable platforms, and a com- 2288

plete parametric study spanning altitude, solar cycle, attitude strategy, and propulsion 2289

technology is identied as a future research need. 2290

A unifying observation emerges: the integration barriers are not gaps within individual 2291

technology disciplines but gaps between disciplines. The soft robotics community, the inat- 2292

able structures community, the space power community, and the GNC community each have 2293

mature capabilities; the intersections remain unexplored. This fragmentation of the research 2294

landscape is itself a structural challenge that programmatic measures (cross-disciplinary 2295

funding calls, joint ground demonstrators) must address. 2296

13.3 Proposed Research Roadmap: 5-Year and 15-Year Horizons 2297

Based on the gap analysis above and the technology readiness levels documented in Sec- 2298

tions 512, we propose a two-horizon research roadmap. The 5-year horizon (20262031) 2299

targets ground validation and component-level ight demonstration; the 15-year horizon 2300

(20262041) targets system-level ight demonstration and initial operational capability. 2301

5-Year Horizon (20262031). Five priority activities are identied, each addressing one 2302

or more critical or moderate gaps: 2303

1. Jamming-in-vacuum experimental validation (addresses M1). Ground experi- 2304

ment: vacuum chamber with sealed granular/layer jamming specimen connected to a 2305

pressurized chamber simulating an inatable interior. Measure stiness ratio versus 2306

pressure dierential and compare to terrestrial baselines. Space-compatible granular 2307

media candidates include hollow glass microspheres and metallic powder. This ex- 2308

periment is well-dened, moderate-cost, and publishable regardless of outcome. If 2309

successful, it validates variable-stiness robotic elements that are simpler in orbit than 2310

on Eartha paradigm inversion for soft space robotics. 2311

2. FBG-in-Vectran-webbing ight demonstration (addresses M6). Current ground 2312

demonstrations at NASA JSC Bally Ribbon Mills and Luna Innovations [2020] have 2313

Now

5-year milestones

2026 2028

Jamming-in-vacuum validation

2027 2030

FBG flight demonstration

2026 2029

Perovskite fold-deploy testing

2027 2030

Rigid-flexible interface prototype

2026 2031

Gyroelastic theory extension

15-year milestones

2029 2035

Soft gripper flight demo

2031 2037

10 m inflatable with PV

2033 2039

Assembly robot on inflatable

5-year milestone

15-year milestone

2035 2041

AOCS-qualified inflatable

Dependency

2026 2028 2030 2032 2034 2036 2038 2040 Year

Figure 13: Research roadmap for soft inatable robotic space systems spanning 5-year and 15-year horizons. Near-term milestones focus on ground validation of critical unknowns (jamming-in-vacuum, FBG ight, perovskite fold-deploy, rigid-exible interface); long-term milestones target integrated ight demonstrations (soft gripper capture, 10 m inatable with PV, assembly robot on inatable substrate, AOCS-qualied inatable).

reached TRL 45. The next step is a ight experiment on an ISS external payload 2314

platform (e.g., MISSE or Bartlett) exposing FBG-instrumented Vectran webbing to the 2315

LEO environment (atomic oxygen, UV, thermal cycling, MMOD) for 1224 months. 2316

Success would advance the technology to TRL 67 and establish the ight heritage 2317

base for inatable SHM. 2318

3. Perovskite/CIGS fold-deploy-power testing (addresses C5, M5). Deposit per- 2319

ovskite/CIGS tandem cells on 25 µm polymer substrates identical to those used for 2320

inatable MLI. Subject samples to 1000 fold/deploy mechanical cycles, 1000 thermal 2321

vacuum cycles (−100 ◦C to 120 ◦C), and atomic oxygen exposure at LEO-equivalent 2322

uences. Measure power output degradation after each environmental stress. This 2323

establishes whether the remarkable radiation hardness of perovskite/CIGS Lang et al. 2324

[2020] survives the additional mechanical and environmental stresses of inatable in- 2325

tegration. 2326

4. Rigid-to-exible interface ground prototype (addresses C3). Design, fabricate, 2327

and test embedded metallic load-spreader rings sewn into representative multi-layer 2328

inatable fabric during manufacture. Characterize load distribution, stress concentra- 2329

tion factors, and modal response under simulated assembly loading. Compare FEA 2330

predictions with experimental measurements. This ground programme would produce 2331

the rst published dataset on rigid-to-exible assembly interfaces for space inatables. 2332

5. Gyroelastic theory extension for pressure-stabilized membranes (addresses 2333

C4). Mathematical extension of the D'EleuterioHughes framework D'Eleuterio and 2334

Hughes [1984, 1986] incorporating pressure-dependent stiness and fabric orthotropy. 2335

Numerical validation against commercial FEM codes for representative inatable ge- 2336

ometries (cylinder, torus, sphere). Publication of the extended theory would establish 2337

the foundational AOCS framework that any 100-metre-class inatable mission will 2338

require. 2339

15-Year Horizon (20262041). Four system-level demonstrations dene the long-term 2340

roadmap: 2341

1. Soft gripper ight for debris capture (addresses C1, C2). A CubeSat or small- 2342

satellite class mission demonstrating compliant capture of a cooperative (then non- 2343

cooperative) target in LEO. The gripper subsystem (gecko adhesive, DEMES, or suc- 2344

cessor technology) operates on an inatable arm with integrated FBG sensing. This 2345

mission provides the rst orbital data on soft capture dynamics and validates the frag- 2346

mentation risk reduction argument with ight telemetry. 2347

2. 10-metre inatable with integrated photovoltaics (addresses C5). A free-ying 2348

technology demonstrator deploying a 10-metre-class inatable membrane with lami- 2349

nated perovskite/CIGS cells, demonstrating fold/deploy survival and power generation 2350

in the orbital environment. This bridges the gap between ROSA-class rigid-boom ex- 2351

ible arrays (TRL 9) and the 100-metre inatable solar platforms envisioned for future 2352

missions. 2353

3. Assembly robot on inatable substrate (addresses C3). A ground or parabolic- 2354

ight demonstration of a walking or crawling robot (E-Walker class Nair et al. [2024]) 2355

operating on an inatable test article, attaching and detaching rigid modules via em- 2356

bedded hardpoint interfaces. This validates the rigid-to-exible assembly concept in 2357

representative (reduced) gravity conditions. 2358

4. AOCS-qualied pressure-stabilized inatable (addresses C4). A free-ying in- 2359

atable structure (310 metre scale) with onboard AOCS demonstrating three-axis at- 2360

titude control of a pressure-stabilized membrane in LEO. This validates the extended 2361

gyroelastic theory and provides the rst ight data on control-structure interaction for 2362

inatable spacecraft. 2363

Drag Power Thermal Cascade for 100 m Inflatable at 500 km

P = F·ve/2

( = 60%) PSA = 300 W/m2

EP 100%

waste 60%

Solar Array

EP Thrust

Electrical

Thermal Waste

Best case (solar min,

Drag Force

Required

Power

Heat

Area

edge-on)

3.3 m2

0.35 N

0.35 N

1.0 kW

0.6 kW

Radiator

Area

0.2 m2

Solar Array

EP Thrust

Electrical

Thermal Waste

Worst case (solar max,

Drag Force

Required

Power

Heat

Area

broadside)

165+ m2

21 N

21 N

50+ kW

30+ kW

Radiator

Area

10+ m2

Figure 14: Drag-power-thermal cascade analysis for a 100 m-class inatable structure in LEO, illustrating how atmospheric drag drives propulsion power requirements, which in turn drive solar array sizing and thermal dissipation budgets. The cascade quanties the interdependence of the AOCS, power, and thermal subsystems.

13.4 The Path to Flight Demonstration 2364

Among the roadmap milestones, the most ight-ready near-term demonstrator can be iden- 2365

tied by selecting the highest-TRL components from each technology area and integrating 2366

them into a single mission concept. The analysis in Sections 58 suggests the following 2367

combination: 2368

ˆ Capture mechanism: Gecko adhesive gripper (TRL 45, microgravity validated, 2369

400 kg capacity) Jiang et al. [2017], noting that this is a compliant end-eector on a 2370

conventional arm rather than a fully soft system. 2371

ˆ Arm structure: Inatable multi-link arm based on the POPUP concept (TRL 3) 2372

Palmieri et al. [2023], using Vectran fabric links with FBG-instrumented webbing. 2373

ˆ Structural health monitoring: FBG sensors in Vectran webbing (TRL 45 ground) 2374

Bally Ribbon Mills and Luna Innovations [2020], providing both SHM and propriocep- 2375

tive shape sensing via multicore FOSS principles Galloway et al. [2019]. 2376

ˆ Deployment: SMA-based hinge deployment for arm segments (TRL 89) Costanza 2377

and Tata [2020]. 2378

This combination achieves an estimated system TRL of 34, limited by the inatable 2379

arm structure. A CubeSat-class (12U16U) demonstrator could validate the complete soft 2380

capture conceptdeploy inatable arm, acquire cooperative target, demonstrate FBG-based 2381

shape sensing during capturewithin a 35 year development timeline from programme ini- 2382

tiation. The mission would produce the rst orbital dataset on: (i) inatable arm deployment 2383

dynamics, (ii) FBG sensor performance in the LEO environment on a exible structure, and 2384

(iii) compliant capture contact dynamics. These three datasets address critical gaps C2, M6, 2385

and partially C1, making this demonstrator the highest-value single mission for advancing 2386

the eld. 2387

The key technical risk is the inatable arm structure: POPUP-class arms have been 2388

demonstrated only in simulation Palmieri et al. [2023], and the transition from analytical 2389

design to space-qualied ight hardware requires a focused engineering programme. However, 2390

the constituent technologiesVectran fabric, SMA deployment mechanisms, FBG sensors 2391

each have independent space heritage that de-risks the integration challenge. 2392

A critical observation from the roadmap analysis is that the fragmentation paradox (Sec- 2393

tion 3.1) will not be resolved by the ight demonstrator alone. The proposed CubeSat mission 2394

validates soft capture mechanics but does not generate fragmentation data. Resolving gap 2395

C1 requires a parallel ground campaign: hypervelocity and low-velocity impact testing with 2396

debris surrogate materials (solar panel fragments, aluminium honeycomb, carbon bre com- 2397

posite) at representative contact forces, comparing rigid grasp, compliant grasp, and soft 2398

envelopment capture modes. Parabolic ight campaigns can provide microgravity validation 2399

of the ground results. Together, the ight demonstrator and the ground fragmentation study 2400

would establish the quantitative evidence base that the soft ADR proposition currently lacks. 2401

14 Conclusions 2402

This survey has reviewed the state of the art in soft inatable robotic systems for space 2403

applications, covering eight enabling technology areas across 14 sections and synthesizing 2404

ndings from the active debris removal, space exploration, and robotic assembly domains. 2405

Four key ndings emerge from this comprehensive analysis. 2406

Finding 1: The Fragmentation Paradox Demands Soft Capture Solutions. The 2407

space debris environment has reached a critical state: over 54,000 tracked objects larger than 2408

10 cm, an estimated 140 million fragments between 1 mm and 1 cm, and a total orbital mass 2409

exceeding 15,800 tonnes ESA Space Debris Oce [2025]. Active debris removal at the rate of 2410

at least 5 large objects per year is required to stabilize the LEO population Liou et al. [2010]. 2411

Yet the dominant ADR approachrigid robotic capture, as exemplied by ClearSpace-1 2412

carries an unquantied but non-trivial fragmentation risk for tumbling targets (Section 3.1). 2413

Rigid capture of a debris object could generate new fragments, potentially exacerbating the 2414

very problem it aims to solve. Soft and compliant capture mechanisms (Section 3.2), by ab- 2415

sorbing kinetic energy rather than transmitting contact impulses, oer a system-level safety 2416

margin that rigid capture cannot provide. The absence of a quantitative soft-versus-rigid 2417

fragmentation comparison (gap C1) is the single most important open research question 2418

identied by this survey. Until this comparison is performed, the ADR community is select- 2419

ing capture mechanisms without the fundamental dataset needed for informed technology 2420

selection. 2421

Finding 2: Inatable Habitats Are Flight-Proven, with a Clear Path to Deep- 2422

Space Application. BEAM's 8+ years of continuous operation on the International Space 2423

Station has conclusively demonstrated that pressure-stabilized inatable modules can sur- 2424

vive the LEO environment at TRL 9 (Section 4.1). The mass eciency advantage is decisive: 2425

39 kg m−3 for TransHab versus 137205 kg m−3 for metallic ISS modules Valle et al. [2019a]. 2426

Vectran-based restraint layers provide specic strengths exceeding 2300 kN m kg−1, an or- 2427

der of magnitude beyond aerospace metals (Section 5.1). Current commercial programmes 2428

(Sierra Space LIFE) have demonstrated full-scale burst pressures of 77 psi, exceeding NASA 2429

structural requirements by 27% (Section 4.2). The path from BEAM to deep-space habitats 2430

requires addressing three challenges: radiation shielding (BEAM's 810× higher SPE dose 2431

versus metallic modules; Section 4.4), autonomous deployment reliability (BEAM's 25-burst, 2432

7-hour deployment was rescued by ISS crew; Section 6.3), and the 19× volume scale-up from 2433

BEAM's 16 m3 to a 300+ m3 deep-space transit habitat. Each challenge is substantive but 2434

bounded, with identied mitigation strategies (water-wall radiation shielding, deployment 2435

sequencing control, and multi-layer restraint engineering, respectively). 2436

Finding 3: The Space Vacuum Is a Resource, Not Merely an Obstacle. The tra- 2437

ditional framing of the space environment as hostile to soft roboticspneumatic actuation 2438

loses its working medium, elastomers outgas, lubricants evaporateis being overturned by 2439

three developments. First, vacuum-gap electrostatic actuators Sirbu et al. [2025] achieve 2440

>4 N force at 0.7 g mass with >100 Hz bandwidth by using internal vacuum gaps as func- 2441

tional elements; these actuators require vacuum and are simpler in orbit than on Earth 2442

(Section 7.2). Second, the jamming-in-vacuum principle exploits the ambient orbital vac- 2443

uum as the external low-pressure reservoir for granular or layer jamming, eliminating the 2444

vacuum pump required in terrestrial implementations (Section 7.6); this remains a logical 2445

deduction requiring experimental validation (gap M1), but the physics is straightforward. 2446

Third, the very existence of pressure-stabilized inatable structures depends on the vacuum 2447

environment providing the pressure dierential that creates structural stiness. Together, 2448

these observations suggest that soft inatable robotic systems for space constitute a distinct 2449

engineering disciplinenot merely terrestrial soft robotics adapted for space, but a eld 2450

where the space environment enables capabilities impossible on Earth. 2451

Finding 4: The Critical Barrier Is System Integration, Not Individual Technol- 2452

ogy Maturity. Perhaps the most signicant nding of this survey is negative: no single 2453

technology gap is a showstopper for the eld. Vectran and Kevlar are ight-proven for inat- 2454

able structures (TRL 9). SMA deployment mechanisms are ight-proven (TRL 89). FBG 2455

sensors have own on Proba-2 (TRL 78). iROSA-class exible photovoltaics power the 2456

ISS (TRL 9). Loop heat pipes transport multi-kilowatt thermal loads (TRL 9). Reaction 2457

wheels provide attitude control for the largest operational spacecraft (TRL 9). The barrier 2458

is at the interfaces: no programme has integrated FBG sensors into an inatable structure 2459

for ight; no programme is developing photovoltaics on inatable substrates; no theory ad- 2460

dresses AOCS for pressure-stabilized membranes; no interface enables rigid module assembly 2461

onto exible platforms. The eld suers from a fragmentation of its ownnot of debris, but 2462

of research communities. Soft roboticists, inatable structure engineers, space power spe- 2463

cialists, and GNC researchers each advance their disciplines without the cross-disciplinary 2464

programmes needed to integrate their outputs into ight-ready systems. 2465

This survey has attempted to bridge that fragmentation by reviewing all eight enabling 2466

technology areas through a single lens: the unifying thesis that the same high-strength fabric 2467

technologies (Vectran, Kevlar, Nextel) serve both active debris removal and space exploration 2468

applications. The cross-domain connections identied throughoutthermal management 2469

informing actuator design (Section 10), MMOD protection materials serving as actuation 2470

substrates (Section 5), FBG sensing unifying habitat SHM and robotic proprioception (Sec- 2471

tion 8.1), and the dragpowerthermal cascade governing 100-metre-class platform architec- 2472

ture (Section 11.3)are insights that emerge only from the breadth of an integrative review. 2473

They cannot be seen from within any single technology discipline. 2474

The research roadmap proposed in Section 13.3 identies concrete near-term actions: 2475

jamming-in-vacuum validation, FBG ight demonstration on inatable webbing, perovskite/CIGS 2476

fold-deploy testing, rigid-exible interface prototyping, and gyroelastic theory extension. The 2477

most ight-ready integrated demonstratora gecko-adhesive gripper on an inatable arm 2478

with FBG structural health monitoringcould y within 35 years of programme initia- 2479

tion, generating the rst orbital dataset on soft inatable robotic capture. The longer-term 2480

visiona 10-metre inatable with integrated photovoltaics, assembly robots operating on 2481

inatable platforms, and AOCS-qualied pressure-stabilized structuresdenes a 15-year 2482

trajectory toward operational capability. 2483

The space debris crisis demands action on a timescale shorter than the 15-year technology 2484

roadmap allows. ClearSpace-1 and its successors will y rigid capture missions within this 2485

decade. The soft robotics and inatable structures communities must move from component- 2486

level demonstration to system-level integration with urgency commensurate with the prob- 2487

lem. The technologies exist; the integration does not. Closing the integration gaps identied 2488

in this survey is the dening challenge for the next generation of space robotics research. 2489

References 2490

ESA Space Debris Oce. ESA space environment report 2025. Technical report, European 2491

Space Agency, 2025. URL https://www.esa.int/Space_Safety/Space_Debris. Current 2492

debris environment statistics: 54,000 tracked objects, 15,800 tonnes, Starlink avoidance 2493

manoeuvre data. 2494

Donald J. Kessler and Burton G. Cour-Palais. Collision frequency of articial satellites: 2495

The creation of a debris belt. Journal of Geophysical Research, 83(A6):26372646, 1978. 2496

doi: 10.1029/JA083iA06p02637. Seminal paper predicting the debris cascade (Kessler 2497

syndrome). 2498

Jer-Chyi Liou and Nicholas L. Johnson. Risks in space from orbiting debris. Science, 311 2499

(5759):340341, 2006. doi: 10.1126/science.1121337. LEGEND simulation demonstrating 2500

LEO instability even with zero future launches. 2501

Jer-Chyi Liou and Nicholas L. Johnson. Instability of the present LEO satellite popula- 2502

tions. Advances in Space Research, 41(7):10461053, 2008. doi: 10.1016/j.asr.2007.04.081. 2503

Extended 200-year LEGEND projections conrming irrecoverable LEO instability. 2504

Jer-Chyi Liou, Nicholas L. Johnson, and N. M. Hill. Controlling the growth of future LEO 2505

debris populations with active debris removal. Acta Astronautica, 66(56):648653, 2010. 2506

doi: 10.1016/j.actaastro.2009.08.005. Quanties the 5 large objects per year ADR require- 2507

ment to stabilise LEO. 2508

Alexander Ledkov and Vladimir Aslanov. Review of contact and contactless active space 2509

debris removal approaches. Progress in Aerospace Sciences, 134:100858, 2022. doi: 2510

10.1016/j.paerosci.2022.100858. PiAS ADR review covering nets, harpoons, ion beam 2511

shepherd, tethers; soft/compliant approaches absent. 2512

Guglielmo S. Aglietti et al. The active space debris removal mission RemoveDebris: Part 2513

2, in orbit operations. Acta Astronautica, 168:310322, 2020. doi: 10.1016/j.actaastro. 2514

2019.09.001. Documents RemoveDebris harpoon boom failure at 20 m/s contact velocity; 2515

in-orbit results. 2516

Gerald D. Valle et al. System integration comparison between inatable and metallic 2517

spacecraft structures. In 2019 IEEE Aerospace Conference, pages 110, 2019a. doi: 2518

10.1109/AERO.2019.8742000. Denitive mass/volume comparison: TransHab 39 kg/m3 vs 2519

Columbus 137 kg/m3 vs Tranquility 205 kg/m3. Vectran specic strength 2,330 kN-m/kg. 2520

NASA Johnson Space Center. BEAM ISS year-one science results. Technical Report JSC- 2521

CN-39950, NASA, 2017. BEAM Year-One data: 16 m3, 1,415 kg, deployment anomaly 2522

(25 bursts over 7 hours), September 2017 SPE radiation data (2-2.5 mGy vs 0.25 mGy in 2523

metallic modules). 2524

Angel Flores-Abad, Ou Ma, Khanh Pham, and Steve Ulrich. A review of space robotics 2525

technologies for on-orbit servicing. Progress in Aerospace Sciences, 68:126, 2014. doi: 10. 2526

1016/j.paerosci.2014.03.002. Foundational PiAS review establishing the four-phase capture 2527

framework; pre-dates soft robotics wave. 2528

Zheyao Zhao et al. Robotic manipulators for in-orbit servicing and active debris removal: 2529

Review and comparison. Progress in Aerospace Sciences, 152:101046, 2024. doi: 10.1016/ 2530

j.paerosci.2024.101055. PiAS rigid manipulator review; soft and inatable manipulators 2531

are outside scope. 2532

Yongchang Zhang, Pengchun Li, Jiale Quan, et al. Progress, challenges, and prospects of 2533

soft robotics for space applications. Advanced Intelligent Systems, 5(3):2200071, 2023a. 2534

doi: 10.1002/aisy.202200071. Most relevant prior soft-robotics-for-space survey; does not 2535

cover inatable platforms, power, thermal, AOCS, or dual application principle. 2536

Ion-Dan Sirbu et al. Vacuum-gap electrostatic multilayer actuators for soft robotics in 2537

space. Nature Communications, 16, 2025. doi: 10.1038/s41467-025-66232-7. Break- 2538

through: vacuum-gap actuators require vacuum to function. 0.7 g mass, >4 N force, 2539

>100 Hz bandwidth, 1.4 kW/kg. TRL 3-4; no space environment testing reported. 2540

Siobhan G. Fitzgerald, Gareth W. Delaney, and David Howard. A review of jamming ac- 2541

tuation in soft robotics. Actuators, 9(4):104, 2020. doi: 10.3390/act9040104. Review of 2542

jamming mechanisms and variable stiness. Stiness ratios >25:1 achievable. Basis for 2543

space jamming-for-free insight. 2544

O. Araromi, I. Gavrilovich, J. Shintake, S. Rosset, M. Richard, V. Gass, and H. Shea. 2545

Rollable multisegment dielectric elastomer minimum energy structures for a deployable 2546

microsatellite gripper. IEEE/ASME Transactions on Mechatronics, 20(1):438446, 2015. 2547

doi: 10.1109/TMECH.2014.2329367. DEMES gripper for CleanSpace One: 860,000+ 2548

cycles, 0.8-2.2 mN force, passive negative feedback in microgravity. 2549

NASA. NASA Systems Engineering Processes and Requirements. NPR 7123.1B, 2020. 2550

Denes the Technology Readiness Level (TRL) scale used throughout this survey. 2551

Xiang Wang et al. Compliant capture approaches for active debris removal missions. 2552

Aerospace, 10(1):60, 2023. States explicitly that rigid manipulation has the potential 2553

to generate fragments during capture. 2554

Yan Chen et al. Risks of contact-based caging for tumbling debris targets. arXiv preprint 2555

arXiv:2405.00943, 2024a. Characterises single contact-based caging as excessively risky 2556

for fast-tumbling targets. 2557

Hao Jiang et al. A robotic device using gecko-inspired adhesives can grasp and ma- 2558

nipulate large objects in microgravity. Science Robotics, 2(7):eaan4545, 2017. doi: 2559

10.1126/scirobotics.aan4545. Microgravity-validated gecko adhesive gripper: 100% suc- 2560

cess rate for spheres, up to 400 kg objects. Highest TRL soft capture approach (TRL 2561

4-5). 2562

Paolo Palmieri et al. An inatable 7-DOF space robotic arm for active debris removal. In 2563

CISM International Centre for Mechanical Sciences, Courses and Lectures, 2023. POPUP 2564

inatable 7-DOF arm for ADR; Vectran fabric links; TRL 3. 2565

DuPont. Kevlar 49 Aramid Fiber Technical Data Sheet, 2019. Filament-level UTS = 3.6 2566

GPa; density = 1.44 g/cm3; specic strength = 2,500 kN-m/kg at lament level. 2567

Sierra Space Corporation. LIFE habitat ultimate burst pressure test results. Corporate press 2568

release, 2024. LIFE 285: full-scale burst at 77 psi; LIFE 10 (1/3 scale): burst at 255 psi, 2569

FOS 16× LEO, 23× lunar. Note: corporate press release, not peer-reviewed. 2570

Kriss J. Kennedy. Lessons from TransHAB: An architect's experience. In AIAA Space 2571

Architecture Symposium, number AIAA 2002-6105, 2002. TransHab design philosophy, 2572

burst tests to 4× operating pressure, restraint-layer preloading for deployment. Primary 2573

reference for TransHab programme. 2574

John W. Norbury et al. Passive shielding materials for space radiation protection: A review. 2575

Materials, 2025. Polyethylene achieves 27.8% mass saving vs aluminium for equivalent 2576

proton shielding. Three-layer shields: up to 70% TID improvement for electrons. 2577

Timothy L. Weadon. Long term loading properties of Vectran fabric for inatable space 2578

habitats. Master's thesis, West Virginia University, 2013. Vectran creep characterisation: 2579

exponential sensitivity to load level, manufacturing variability ±10% UTS. Primary data 2580

source for restraint layer lifetime prediction. 2581

Roberto Destefanis et al. Enhanced space debris shields for manned spacecraft. In 2582

International Journal of Impact Engineering, volume 29, pages 217228, 2003. doi: 2583

10.1016/j.ijimpeng.2003.09.019. Nextel/Kevlar stued Whipple shield: protects against 2584

2× larger projectiles at equal areal density vs standard Whipple. 2585

Jun-Ho Cha et al. Inatable origami multi-shock shield (IMSS) structure for space habitation 2586

module. In Proceedings of the 34th Congress of the International Council of Aeronautical 2587

Sciences (ICAS), 2024. IMSS: waterbomb tessellation, UHMWPE/Kevlar/Nextel, 90% 2588

volume savings vs rigid Whipple. Demonstrates origami geometry serving both packaging 2589

and debris protection simultaneously. 2590

Douglas A. Litteken. Inatable technology: Using exible materials to make large space 2591

structures. In Proceedings of SPIE 10966, Earth and Space 2018, volume 10966, 2019. 2592

doi: 10.1117/12.2500091. Comprehensive heritage review from NASA JSC SME: Echo to 2593

BEAM, including Mars lander airbags with Vectran, structural design principles. 2594

Koryo Miura. Method of packaging and deployment of large membranes in space. Technical 2595

Report 618, Institute of Space and Astronautical Science, 1985. Foundational Miura-ori 2596

fold pattern: single-actuator, rigid-foldable, unlimited compaction. 2597

Mark Schenk, Andrew D. Viquerat, Keith A. Seen, and Simon D. Guest. Review of inat- 2598

able booms for deployable space structures: Packing and rigidization. Journal of Spacecraft 2599

and Rockets, 51(3):762778, 2014. doi: 10.2514/1.A32598. Review of packaging ratios for 2600

inatable booms; cylindrical adaptations of Miura-ori; rigidization classication. 2601

Felix Lang et al. Proton radiation hardness of perovskite tandem photovoltaics. Joule, 4(5): 2602

10541069, 2020. doi: 10.1016/j.joule.2020.03.012. Perovskite/CIGS tandem: 2,100 W/kg 2603

with 25 µm substrate, >85% retention after >50 year LEO-equivalent proton irradiation. 2604

Key enabler for inatable-integrated PV. 2605

William Foster-Hall, David Harvey, Ling Yin, and Rini Akmeliawati. Soft robotics for space 2606

applications: Cryogenic performance of modular metallic cable structures. Soft Robotics, 2607

2025. doi: 10.1177/21695172251364758. Metallic cable-based soft robots: only 5% sti- 2608

ness increase over 100 cycles at −196 ◦C. Solves elastomer embrittlement for deep-space 2609

applications. 2610

Wei Zhang, Feng Li, Junlin Li, and Qinkun Cheng. Review of on-orbit robotic arm active de- 2611

bris capture removal methods. Aerospace, 10(1):13, 2022. doi: 10.3390/aerospace10010013. 2612

Andrea Stol, Paolo Gasbarri, and Marco Sabatini. A combined impedance-pd approach 2613

for controlling a dual-arm space manipulator in the capture of a non-cooperative target. 2614

Frontiers in Robotics and AI, 4:34, 2017. doi: 10.1016/j.actaastro.2017.07.014. 2615

Aerospace Corporation. IMPACT satellite breakup model. Technical report, The Aerospace 2616

Corporation, 2020. Catastrophic fragmentation threshold: 10 J/g specic energy. 2617

ClearSpace SA and European Space Agency. ClearSpace-1: ESA's rst space debris re- 2618

moval mission. https://www.esa.int/Space_Safety/ClearSpace-1, 2020. Contract 2619

value ¿86M; target changed from VESPA to Proba-1; launch ∼2029. 2620

Nicholas L. Johnson, Paula H. Krisko, J.-C. Liou, and Phillip D. Anz-Meador. NASA's new 2621

breakup model of EVOLVE 4.0. Advances in Space Research, 28(9):13771384, 2001. doi: 2622

10.1016/S0273-1177(01)00423-9. Original experimental basis for the 10 J/g catastrophic 2623

fragmentation threshold. 2624

Jun Shintake, Vito Cacucciolo, Dario Floreano, and Herbert Shea. Soft robotic grippers. 2625

Advanced Materials, 30(29):1707035, 2018. doi: 10.1002/adma.201707035. 2626

Yuhang Liu, Kai Luo, Shuai Wang, Xiping Song, and Zhijuan Zhang. A soft and bistable 2627

gripper with adjustable energy barrier for fast capture in space. Soft Robotics, 10(1):7787, 2628

2022. doi: 10.1089/soro.2021.0147. 2629

Yongchang Zhang, Jiale Quan, Pengchun Li, Wenping Song, and Guozheng Zhang. A ytrap- 2630

inspired bistable origami-based gripper for rapid active debris removal. Advanced Intelli- 2631

gent Systems, 5(7):2200468, 2023b. doi: 10.1002/aisy.202200468. 2632

Miguel Ruiz Vincueria et al. A thermally-resilient multi-layered soft gripper for space appli- 2633

cations. arXiv preprint, 2023. arXiv:2311.08942; accepted IROS 2024. 2634

CT Ingénierie, AirCaptif, and CNES. INSIDeR: Innovative Net and Space Inatable struc- 2635

ture for active Debris Removal. In Proceedings of the 7th European Conference on Space 2636

Debris (SDC7), number Paper 714. ESA, 2017. 2637

CT Ingénierie, AirCaptif, CNES, and ESA. INSIDeR: Ground demonstrator development 2638

and net capture simulations. In Proceedings of the 8th European Conference on Space 2639

Debris (SDC8), number Paper 91. ESA, 2021. 2640

J. L. Arnold, Eric L. Christiansen, A.J. Davis, James L. Hyde, Dana M. Lear, J.-C. Liou, 2641

Frankel Lyons, Thomas G. Prior, George Studor, Martin Ratli, Shannon Ryan, F. Gio- 2642

vane, and B. Corsaro. Handbook for designing MMOD protection. Technical Report 2643

NASA/TM-2009-214785, NASA, 2009. 2644

Roberto Destefanis, Frank Schaefer, Michel Lambert, and Michel Faraud. Selecting enhanced 2645

space debris shields for manned spacecraft. International Journal of Impact Engineering, 2646

33(112):219230, 2006. doi: 10.1016/j.ijimpeng.2006.09.065. 2647

NASA. NextSTEP-2 partners develop ground prototypes for deep space habitats. https: 2648

//www.nasa.gov/nextstep, 2016. Six companies selected August 2016: Bigelow, Boeing, 2649

Lockheed Martin, Orbital ATK, Sierra Nevada, NanoRacks. 2650

Lockheed Martin. Bursting the bubble with inatable habi- 2651

tats. https://www.lockheedmartin.com/en-us/news/features/2022/ 2652

bursting-the-bubble-with-inflatable-habitats.html, 2022. Burst pressure 2653

285 psi; Waterton Canyon facility; NextSTEP-2 prototype. 2654

Hassell Studio and European Space Agency. Inatable lunar habitat at Shackleton crater, 2655

2024. Scalable inatable pod system; regolith 3D printing; capacity up to 144 people. 2656

Skidmore, Owings & Merrill and European Space Agency. Moon village habitat concept, 2657

2019. Semi-inatable shell; 4-person crew; 300-day missions. 2658

European Space Agency. Pneumocell: Inatable lunar habitat for regolith burial, 2018. 2659

Designed for burial under 45 m of regolith for radiation protection. 2660

NASA. LOFTID: Low-earth orbit ight test of an inatable decelerator. https://www. 2661

nasa.gov/mission/loftid/, 2022. 6 m diameter; orbital reentry at Mach 30; November 2662

2022. 2663

ESA/ESTEC. First European workshop on inatable space structures, 2002. ESA-WPP-200; 2664

Noordwijk, The Netherlands. 2665

G. D. Valle, D. Litteken, and T. Jones. Review of habitable softgoods inatable design, 2666

analysis, testing, and potential space applications. In AIAA SciTech 2019 Forum, 2019b. 2667

doi: 10.2514/6.2019-1528. AIAA 2019-1528. 2668

R. Destefanis et al. Space environment characterisation of Kevlar. In ESA-ESTEC ISME-09, 2669

2009. 2670

Toyobo Co., Ltd. ZYLON (PBO ber) technical information. Technical report, Toyobo Co., 2671

Ltd., 2005. 2672

M. A. Said, B. Dingwall, A. Gupta, A. M. Seyam, G. N. Mock, and T. Theyson. Investigation 2673

of ultra violet (UV) resistance for high performance bers. Advances in Space Research, 2674

37(11):20522058, 2006. Independent UV degradation data for PBO (Zylon) bre. 2675

E. Christiansen and B. Davis. Heat-cleaned Nextel in MMOD shielding. In NASA/JSC, 2676

2019. NTRS 20190033497. 2677

Roger C. Linton, Ann F. Whitaker, and Miria Finckenor. Space environment durability of 2678

beta cloth in LDEF thermal blankets. In NASA Conference Publication, 1993. NTRS 2679

19940026510. 2680

B. A. Banks, K. K. de Groh, and S. K. Miller. Low Earth orbital atomic oxygen interactions 2681

with materials. Technical report, NASA Glenn Research Center, 2004. NASA/TM-2004- 2682

213400. 2683

M. M. Finckenor and D. Dooling. Multilayer insulation material guidelines. Technical report, 2684

NASA Marshall Space Flight Center, 1999. NASA/TP-1999-209263. 2685

K. J. Kennedy. TransHab inatable habitat case study. Technical report, NASA, 2016. 2686

NTRS 20160011581. 2687

J. Edgecombe, H. DeLaFuente, and Gerald D. Valle. Damage tolerance testing of a NASA 2688

TransHab derivative woven inatable module. In NASA, 2009. doi: 10.2514/6.2009-2167. 2689

NTRS 20090014832. 2690

D. G. Gilmore. Spacecraft Thermal Control Handbook, Volume I: Fundamental Technologies. 2691

The Aerospace Press, 2nd edition, 2002. 2692

D. Cadogan and S. E. Scarborough. Rigidizable materials for use in gossamer space inatable 2693

structures. In 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and 2694

Materials Conference, 2001. AIAA 2001-1417. 2695

Andrew D. Viquerat et al. InateSail de-orbit ight demonstration results and follow-on 2696

drag sail applications. Acta Astronautica, 162:203219, 2019. doi: 10.1016/j.actaastro. 2697

2019.06.005. InateSail (2017): 1 m Al-Mylar boom, 10 m2 sail, 72-day deorbit from 505 2698

km. TRL 8-9 for strain-rigidised CubeSat booms. 2699

C. Underwood, A. Viquerat, M. Schenk, S. Fellowes, B. Taylor, Chiara Massimiani, Richard 2700

Duke, Brian Stewart, C. Bridges, D. Masutti, and A. Denis. The InateSail CubeSat 2701

mission: The rst European demonstration of drag-sail de-orbiting. In 4S Symposium, 2702

2017. 2703

B. Defoort, V. Peypoudat, M. Bernasconi, K. Chuda, and X. Coqueret. Recent advances in 2704

the rigidization of gossamer structures. In 45th AIAA/ASME/ASCE/AHS/ASC Struc- 2705

tures, Structural Dynamics, and Materials Conference, 2005. doi: 10.1007/1-4020-3317-6_ 2706

16. AIAA 2004-1561. 2707

R. Allred et al. UV rigidizable carbon-reinforced isogrid inatable booms. In 43rd 2708

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Confer- 2709

ence, 2002. AIAA 2002-1202. 2710

Adherent Technologies Inc. Rigidization on command (ROC) materials for inatable space- 2711

craft. In SAE Technical Paper, 2001. doi: 10.4271/2001-01-2220. 2001-01-2220. 2712

Vikas Rastogi, Sanjay H. Upadhyay, and Kripa Sankar Singh. Rigidization analysis of SMA- 2713

based inatable toroidal space structures. Mechanics Based Design of Structures and 2714

Machines, 52(11), 2024. doi: 10.1080/15397734.2024.2345375. 2715

R. Verker, E. Keren, N. Refaeli, Y. Carmiel, A. Bolker, D. David, S. Katz, E. Sagi, D. Bashi, 2716

I. Finkelstein, T. Nahum, A. Haran, A. Shemesh Sadeh, M. Ariel, I. Gouzman, O. Amrani, 2717

Y. Simhony, and M. Murat. Measurements of material erosion in space by atomic oxygen 2718

using the on-orbit material degradation detector. Acta Astronautica, 2023. doi: 10.1016/ 2719

j.actaastro.2023.07.020. 2720

R. Zhai, Xiaoning Yang, Lixiang Jiang, Hong Gao, Yuxin Zhang, and Zi-Long Jiao. Syner- 2721

gistic eects of atomic oxygen and thermal cycling in low earth orbit on polymer-matrixed 2722

space material. Polymers, 2023. doi: 10.1016/j.heliyon.2023.e17431. PMC 10404884. 2723

M. Schenk, S. Kerr, A. M. Smyth, and S. Guest. Inatable cylinders for deployable space 2724

structures. In Proc. Transformables 2013, 2013. 2725

C. H. M. Jenkins. Gossamer Spacecraft: Membrane and Inatable Structures Technology for 2726

Space Applications, volume 191 of Progress in Astronautics and Aeronautics. AIAA, 2001. 2727

doi: 10.2514/4.866616. 2728

A. C. Horn. A low cost inatable CubeSat drag brake utilizing sublimation. Master's thesis, 2729

Old Dominion University, 2017. 2730

D. P. Cadogan, S. E. Scarborough, J. K. Lin, and G. H. Sapna. PowerSphere multifunctional 2731

ultraviolet-rigidizable inatable structures. Journal of Spacecraft and Rockets, 43(5), 2006. 2732

Na Li, Haijun Peng, and Fei Li. Instantaneous optimal control of inatable folded structures. 2733

Acta Astronautica, 194:5262, 2022a. doi: 10.1016/j.actaastro.2022.02.024. 2734

S. J. Hughes, R. A. Dillman, B. R. Starr, R. A. Stephan, M. C. Lindell, C. J. Player, and F. M. 2735

Cheatwood. Inatable re-entry vehicle experiment (irve) design overview. In 18th AIAA 2736

Aerodynamic Decelerator Systems Technology Conference, 2005. doi: 10.2514/6.2005-1636. 2737

AIAA 2005-1636. 2738

Thomas W. Murphey, Jeremy A. Banik, and Sergio Pellegrino. TRAC boom technology. 2739

AIAA SciTech Forum, (AIAA 2015-0434), 2015. Triangular Rollable and Collapsible 2740

boom: primary competitor to inatable booms for CubeSat-class deployables. 2741

Jeremy A. Banik and Thomas W. Murphey. Performance validation of the triangular rollable 2742

and collapsible mast. In AIAA Structures, Structural Dynamics, and Materials Conference, 2743

number AIAA 2010-2585, 2010. 2744

C. Sickinger and L. Herbeck. Deployment strategies, analyses, and tests for the CFRP booms 2745

of a solar sail. In ESA SP-561, 2004. CFRP boom deployment for solar sails. 2746

J. Santiago-Prowald and G. Rodrigues. Deployable antennas. In Handbook of Antenna 2747

Technologies. Springer, 2018. Comprehensive comparison of mesh, solid, and inatable 2748

antenna architectures. 2749

O. A. Araromi, I. Gavrilovich, J. Shintake, S. Rosset, M. Richard, V. Gass, and H. R. 2750

Shea. Towards a deployable satellite gripper based on multisegment dielectric elastomer 2751

minimum energy structures. In Proc. SPIE 9056, Electroactive Polymer Actuators and 2752

Devices (EAPAD) 2014, page 90562G, 2014. doi: 10.1117/12.2044667. 2753

Haopeng Liang, Yong Zhao, Bingxiao Du, Wensheng Wu, and Xiaoqian Chen. Design, fabri- 2754

cation and modeling of a dielectric elastomer tridimensional minimum energy structure for 2755

space mission. Sensors and Actuators A: Physical, 2023. doi: 10.1016/j.sna.2023.114747. 2756

A. Punning, K. J. Kim, V. Palmre, F. Vidal, C. Plesse, N. Festin, A. Maziz, K. Asaka, 2757

T. Sugino, G. Alici, et al. Ionic electroactive polymer articial muscles in space applica- 2758

tions. Scientic Reports, 4:6913, 2014. doi: 10.1038/srep06913. 2759

J. S. Mehling et al. NASA tendril: A slender continuum robot for inspection. Technical 2760

report, NASA, 2006. 2761

X. Ouyang, D. Meng, X. Wang, C. Wang, B. Liang, and N. Ding. Hybrid rigid-continuum 2762

dual-arm space robots: Modeling, coupling analysis, and coordinated motion planning. 2763

Aerospace Science and Technology, 116:106861, 2021. doi: 10.1016/j.ast.2021.106861. 2764

Fernando Ruíz, Begoña C. Arrue, and Anbal Ollero. Thermally-resilient soft gripper for 2765

on-orbit operations. In IEEE/RSJ International Conference on Intelligent Robots and 2766

Systems (IROS), 2023. doi: 10.48550/arxiv.2311.08942. 2767

G. Costanza and M. E. Tata. Shape memory alloys for aerospace, recent developments, and 2768

new applications: A short review. Materials, 13(8), 2020. doi: 10.3390/ma13081856. 2769

Alain Blanc, Damien Chassoulier, Fabrice Champandard, and Xavier François. Solar array 2770

root hinge based on shape memory alloy (SMA) actuator. In ESMATS, 2013. 2771

Y. Zhang, P. Li, J. Quan, L. Li, G. Zhang, and D. Zhou. Progress, challenges, and prospects 2772

of soft robotics for space applications. Advanced Intelligent Systems, 5(3):2200071, 2023c. 2773

doi: 10.1002/aisy.202200071. 2774

Ahmad Ataka, Peng Qi, Ali Shiva, Ali Shafti, Helge A. Wurdemann, Hongbin Liu, and 2775

Kaspar Althoefer. Real-time pose estimation and obstacle avoidance of a pneumatic soft 2776

growing robot. In IEEE International Conference on Robotics and Automation (ICRA), 2777

2020. Closed-loop pneumatic eversion arm for space-relevant inspection. 2778

Elliot W. Hawkes, Laura H. Blumenschein, Joseph D. Greer, and Allison M. Okamura. A 2779

soft robot that navigates its environment through growth. Science Robotics, 2(8), 2017. 2780

doi: 10.1126/scirobotics.aan3028. 2781

Jianglong Guo, Troy Bamber, Kirk Hovell, et al. Electroadhesion technologies for robotics: 2782

A comprehensive review. Soft Robotics, 2020. doi: 10.1109/tro.2019.2956869. Electroad- 2783

hesion gripping for non-cooperative surfaces. 2784

Yoonho Kim, Hyunwoo Yuk, Ruike Zhao, Shawn A. Chester, and Xuanhe Zhao. Printing 2785

ferromagnetic domains for untethered fast-transforming soft materials. Nature, 558:274 2786

279, 2018. doi: 10.1038/s41586-018-0185-0. 2787

I. Mckenzie, S. Ibrahim, E. Haddad, S. Abad, A. Hurni, and Lun K. Cheng. Fiber optic 2788

sensing in spacecraft engineering: An historical perspective from the European Space 2789

Agency. Frontiers in Physics, 9:719441, 2021. doi: 10.3389/fphy.2021.719441. 2790

S. Girard et al. Radiation eects on ber Bragg gratings: Vulnerability and hardening 2791

studies. Sensors, 2022. 2792

Tamana Baba, N. Saidin, N. F. Hasbullah, and Ishaq Ahmad. Radiation tolerant ber 2793

Bragg gratings: Review of FBG sensing. Journal of Optics, 2025. doi: 10.1007/ 2794

s12596-025-02861-x. 2795

Bally Ribbon Mills and Luna Innovations. Inatable space habitats use sensors embedded in 2796

webbing for structural health monitoring. Technical report, NASA SBIR Program, 2020. 2797

K. C. Galloway, Y. Chen, E. Templeton, B. Rife, I. S. Godage, and E. J. Barth. Fiber optic 2798

shape sensing for soft robotics. Soft Robotics, 6(5):671684, 2019. doi: 10.1089/soro.2018. 2799

0131. 2800

Davide Paloschi, Sergei Bronnikov, Kirill Raspopin, et al. 3D shape sensing with multicore 2801

optical bers. IEEE Sensors Journal, 20(24):1470214710, 2020. doi: 10.1109/metroind4. 2802

0iot48571.2020.9138221. 2803

Shahriar Sefati, Rachel A. Hegeman, Farshid Alambeigi, et al. FBG-based large deection 2804

shape sensing of a continuum manipulator. IEEE Transactions on Medical Robotics and 2805

Bionics, 2021. Multicore FBG shape sensing on continuum robots. 2806

Ginu Rajan, Madhav Ramakrishnan, Yuliya Semenova, Eliathamby Ambikairajah, Ger- 2807

ald Farrell, and Gang-Ding Peng. Overview of ber optic sensor technologies for 2808

strain/temperature sensing applications in composite materials. Sensors, 21(4):125, 2021. 2809

Comprehensive review of FBG sensing for structural health monitoring. 2810

A. M. White, Isaac Little, Anastasiya Artyuk, Nicholas McKibben, Fereshteh Rajabi Kouchi, 2811

Claire Chen, David Estrada, and Zhangxian Deng. On-demand fabrication of piezoelectric 2812

sensors for in-space structural health monitoring. Advanced Materials, 2024. doi: 10.1088/ 2813

1361-665x/ad3d16. 2814

Pengfei Li, Suyu Xu, Chao Xu, Fei Du, and Shancheng Cao. Development of a miniature 2815

dynamic stiness measurement prototype toward structural health monitoring of space 2816

inatable structures. Measurement, 2022b. doi: 10.1016/j.measurement.2022.111051. 2817

Brian R. Spence, Steven White, Michael LaPointe, Steven Kiefer, Paul LaCorte, Jeremy 2818

Banik, Darren Chapman, and Jay Merrill. International space station (ISS) roll-out so- 2819

lar array (ROSA) spaceight experiment mission and results. In 2018 IEEE Aerospace 2820

Conference, pages 113. IEEE, 2018. doi: 10.1109/AERO.2018.8396796. 2821

Baochen Yan et al. Development and challenges of large space exible solar arrays. Space 2822

Solar Power and Wireless Transmission, 2:3342, 2025. doi: 10.1016/j.sspt.2024.12.001. 2823

Hyun-Deok Kim et al. Flexible perovskite/CIGS tandem solar cells with record specic 2824

power. Joule, 2024. 23.6% eciency, 6,150 W/kg cell-level specic power, 100,000 bending 2825

cycles. 2826

David P. Cadogan, Christina Withrow, John Keller, and Barmac K. Taleghani. Development, 2827

design, and testing of PowerSphere multifunctional ultraviolet-rigidizable inatable struc- 2828

tures. Journal of Spacecraft and Rockets, 43(5):10491056, 2003. doi: 10.2514/1.19882. 2829

David Scheiman, Phillip Jenkins, David Brinker, and Joseph Appelbaum. Low-intensity 2830

low-temperature (LILT) measurements and coecients on new photovoltaic structures for 2831

the PowerSphere concept. Thin Solid Films, 480481:6166, 2005. 2832

Phillip P. Jenkins et al. PowerSphere thermal cycling results. Technical Report NASA/TM- 2833

2009-215511, NASA Glenn Research Center, 2009. 2834

Kneev Sharma and A. Santasalo-Aarnio. Energy storage systems for space applications. 2835

Journal of Energy Storage, 120:116315, 2025. doi: 10.1016/j.est.2025.117131. 2836

J. Arenberg, J. Flynn, A. Cohen, R. Lynch, and J. Cooper. Status of the JWST sunshield 2837

and spacecraft. In Proc. SPIE 8442, Space Telescopes and Instrumentation 2012. SPIE, 2838

2016. doi: 10.1117/12.928154. 2839

Hyunsu Kim, Kevin Cheung, Raymond C. Y. Auyeung, David E. Wilson, Kristin M. Chari- 2840

par, Alberto Piqué, and Nicholas A. Charipar. VO2-based switchable radiator for space- 2841

craft thermal control. Scientic Reports, 9:11329, 2019. doi: 10.1038/s41598-019-47572-z. 2842

Almamoun Hendaoui, Nath Emond, Mohamed Chaker, and Emile Haddad. VO2-based smart 2843

coatings with improved emittance-switching properties for an energy-ecient near room- 2844

temperature thermal control of spacecrafts. Solar Energy Materials and Solar Cells, 117: 2845

494498, 2013. doi: 10.1016/j.solmat.2013.07.023. 2846

Yury F. Maydanik. Loop heat pipes. Applied Thermal Engineering, 25(56):635657, 2005. 2847

doi: 10.1016/j.applthermaleng.2004.07.010. 2848

B. Diaconu, M. Cruceru, and L. Anghelescu. Phase change materials in space systems: Fun- 2849

damental applications, materials and special requirementsa review. Acta Astronautica, 2850

214:6077, 2023. doi: 10.1016/j.actaastro.2023.12.040. 2851

F. Nicassio, Daniele Fattizzo, Michele Giannuzzi, G. Scarselli, and G. Avanzini. Attitude 2852

dynamics and control of a large exible space structure by means of a minimum complexity 2853

model. Acta Astronautica, 200, 2022. doi: 10.1016/j.actaastro.2022.05.047. 2854

Gabriele M. T. D'Eleuterio and Peter C. Hughes. Dynamics of gyroelastic continua. Journal 2855

of Applied Mechanics, 51(2):415422, 1984. doi: 10.1115/1.3167639. 2856

Gabriele M. T. D'Eleuterio and Peter C. Hughes. Modal parameter analysis of gyroelastic 2857

continua. Journal of Applied Mechanics, 53(4):918, 1986. doi: 10.1115/1.3171874. 2858

Gabriele M. T. D'Eleuterio and Peter C. Hughes. Dynamics of gyroelastic spacecraft. Journal 2859

of Guidance, Control, and Dynamics, 10(4):401405, 1987. doi: 10.2514/3.20238. 2860

Christopher J. Damaren and Gabriele M. T. D'Eleuterio. Optimal control of large space 2861

structures using distributed gyricity. Journal of Guidance, Control, and Dynamics, 12(5): 2862

723731, 1989. doi: 10.2514/3.20474. 2863

Pedro Cachim, Wolfgang Kraus, Zachary Manchester, Pedro Lourenço, and Rodrigo Ventura. 2864

Optimal attitude control of large exible space structures with distributed momentum 2865

actuators. In 2025 IEEE Aerospace Conference, 2024. doi: 10.1109/AERO58975.2025. 2866

10957060. arXiv:2410.07376. 2867

Yifan Jiang, Jun Zhang, Peng Tian, Tengfei Liang, and Zhihui Li. Aerodynamic drag anal- 2868

ysis and reduction strategy for satellites in very low earth orbit. Aerospace Science and 2869

Technology, 126:107389, 2022. doi: 10.1016/j.ast.2022.108077. 2870

Tommaso Andreussi, Enrico Ferrato, et al. A review of air-breathing electric propulsion: 2871

From mission studies to technology verication. Journal of Electric Propulsion, 2:24, 2872

2022. doi: 10.1007/s44205-022-00024-9. 2873

Lee H. Sentman. Free molecule ow theory and its application to the determination of aero- 2874

dynamic forces. LMSC Technical Report, (LMSC-448514), 1961. Standard free-molecular- 2875

ow drag coecient models. 2876

Kenneth Moe and Mildred M. Moe. Gas-surface interactions and satellite drag coecients. 2877

Planetary and Space Science, 53(8):793801, 2005. doi: 10.1016/j.pss.2005.03.005. 2878

Manuel Stein and John M. Hedgepeth. Analysis of partly wrinkled membranes. Technical 2879

Report TN D-813, NASA, 1961. Foundational tension-eld theory for partially wrinkled 2880

membranes. 2881

D. G. Roddeman, J. Drukker, C. W. J. Oomens, and J. D. Janssen. The wrinkling of thin 2882

membranes: Part Itheory. Journal of Applied Mechanics, 54:884887, 1987. 2883

H. Norman Abramson. The Dynamic Behavior of Liquids in Moving Containers. Number 2884

SP-106. NASA, 1966. Foundational reference for fuel-sloshing dynamics analogy. 2885

Tristan Leclercq and Emmanuel de Langre. Vortex-induced vibrations of cylinders bent by 2886

the ow. Journal of Fluids and Structures, 83:371395, 2018. doi: 10.1016/j.juidstructs. 2887

2018.03.008. Gas-structure coupling relevant to inated membrane dynamics. 2888

NASA. In-space servicing, assembly, and manufacturing (ISAM) state of play2025 edition. 2889

Technical Report NASA/TM-20250008988, NASA Space Technology Mission Directorate, 2890

2025. 2891

Wenjiang Li, Li Zhong, Weijun Zhu, Zhiyou Xu, Qiang Tang, and Wei Zhan. A survey 2892

of space robotic technologies for on-orbit assembly. Space: Science & Technology, 2022: 2893

9849170, 2022c. doi: 10.34133/2022/9849170. 2894

William Doggett et al. NASA puts in-space assembly robots to the test 2895

(CIRAS/TALISMAN/SAMURAI/NINJAR). Technical report, NASA Langley Research 2896

Center, 2018. 2897

Mathieu Rognant et al. Autonomous assembly of large structures in space (PULSAR 2898

project). In 8th European Conference for Aeronautics and Space Sciences (EUCASS), 2899

2019. doi: 10.13009/eucass2019-685. 2900

Aadarsh Nair et al. Design engineering a walking robotic manipulator for in-space assembly 2901

missions (E-Walker). Frontiers in Robotics and AI, 9, 2022. doi: 10.3389/frobt.2022. 2902

941672. 2903

Aadarsh Nair et al. The new era of walking manipulators in space: Feasibility and operational 2904

assessment of assembling a 25 m LAST in orbit. Acta Astronautica, 2024. doi: 10.1016/j. 2905

actaastro.2024.10.002. 2906

Shuai Liu, Enyang Zhang, Zhenbang Xu, and Jingxu Zhang. Design of docking interfaces for 2907

on-orbit assembly of large structures in space. Sensors, 24(20):6534, 2024. doi: 10.3390/ 2908

s24206534. 2909

Robert E. Skelton and Mauricio C. de Oliveira. Tensegrity Systems. Springer, 2009. doi: 2910

10.1007/978-0-387-74242-7. Foundational reference for rigid-to-exible structural connec- 2911

tions. 2912

Andrew P. Sabelhaus, Jonathan Bruce, Ken Caluwaerts, et al. System design and locomotion 2913

of SUPERball, an untethered tensegrity robot. In IEEE International Conference on 2914

Robotics and Automation (ICRA), 2015. NASA Ames tensegrity robot with rigid-bar to 2915

cable-net interfaces. 2916

Juan M. Fernandez, Vaios J. Lappas, and Andrew J. Daton-Lovett. Completely stripped 2917

solar sail concept using bi-stable reeled composite booms. Acta Astronautica, 69:7885, 2918

2014. Boom-membrane attachment analysis for solar sails. 2919

Huaiming Chen et al. Design and analysis of androgynous docking interface for in-space 2920

assembly. Sensors, 24(20), 2024b. 2921

Wei-Jie Li, Da-Yi Cheng, Xin-Gang Liu, Yan-Bin Wang, Wen-Hao Shi, Zhi-Xiang Tang, 2922

Feng Gao, Fa-Ming Zeng, Hong-Yu Chai, Wen-Bo Luo, Qi Cong, and Zhong-Liang Gao. 2923

On-orbit service (OOS) of spacecraft: A review of engineering developments. Progress in 2924

Aerospace Sciences, 108:32120, 2019. doi: 10.1016/j.paerosci.2019.01.004. 2925