Pith. sign in

REVIEW 1 cited by

Risk-Guided Diffusion: Toward Deploying Robot Foundation Models in Space, Where Failure Is Not An Option

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.17601 v1 pith:QQL7LUAY submitted 2025-06-21 cs.RO cs.AI

Rohan Thakker , Adarsh Patnaik , Vince Kurtz , Jonas Frey , Jonathan Becktor , Sangwoo Moon , Rob Royce , Marcel Kaufmann
show 5 more authors
Georgios Georgakis Pascal Roth Joel Burdick Marco Hutter Shehryar Khattak
This is my paper · ORCID
classification cs.ROcs.AI
keywords diffusionfailuremodelsnavigationrisk-guidedroboticsafetyspace
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Safe, reliable navigation in extreme, unfamiliar terrain is required for future robotic space exploration missions. Recent generative-AI methods learn semantically aware navigation policies from large, cross-embodiment datasets, but offer limited safety guarantees. Inspired by human cognitive science, we propose a risk-guided diffusion framework that fuses a fast, learned "System-1" with a slow, physics-based "System-2", sharing computation at both training and inference to couple adaptability with formal safety. Hardware experiments conducted at the NASA JPL's Mars-analog facility, Mars Yard, show that our approach reduces failure rates by up to $4\times$ while matching the goal-reaching performance of learning-based robotic models by leveraging inference-time compute without any additional training.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Embodied AI: Emerging Risks and Opportunities for Policy Action

    cs.CY 2025-08 conditional novelty 4.0 of 10

    A policy analysis arguing that embodied AI risks are real, under-covered by current US/EU/UK frameworks, and best handled through certification, benchmarks, clarified liability, and economic adaptation.

Reference graph

Works this paper leans on

27 extracted references · 15 canonical work pages · cited by 1 Pith paper

  1. [1]

    Daniel,Thinking, fast and slow, 2017

    K. Daniel,Thinking, fast and slow, 2017

    work page 2017

  2. [2]

    To boldly go where no robots have gone before–part 1: Eels robot to spearhead a new one-shot exploration paradigm with in-situ adaptation,

    M. Onoet al., “To boldly go where no robots have gone before–part 1: Eels robot to spearhead a new one-shot exploration paradigm with in-situ adaptation,” inAIAA Scitech F orum, 2024

    work page 2024

  3. [3]

    To boldly go where no robots have gone before– part 4: Neo autonomy for robustly exploring unknown, extreme environments with versatile robots,

    R. Thakkeret al., “To boldly go where no robots have gone before– part 4: Neo autonomy for robustly exploring unknown, extreme environments with versatile robots,” inAIAA SCITECH F orum, 2024

    work page 2024

  4. [4]

    Open- vla: An open-source vision-language-action model,

    M. J. Kim, K. Pertsch, S. Karamcheti, T. Xiao, A. Balakrishna, S. Nair, R. Rafailov, E. Foster, G. Lam, P. Sanketiet al., “Open- vla: An open-source vision-language-action model,”arXiv preprint arXiv:2406.09246, 2024

    arXiv 2024

  5. [5]

    Open x-embodiment: Robotic learning datasets and rt-x models,

    O.-X. E. Collaboration, A. Padalkar, A. Pooley, A. Jain, A. Bewley, A. Herzog, A. Irpan, A. Khazatsky, A. Rai, A. Singhet al., “Open x-embodiment: Robotic learning datasets and rt-x models,”arXiv preprint arXiv:2310.08864, 2023

    arXiv

  6. [6]

    Gnm: A general navigation model to drive any robot,

    D. Shah, A. Sridhar, A. Bhorkar, N. Hirose, and S. Levine, “Gnm: A general navigation model to drive any robot,” in2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 7226–7233

    2023

  7. [7]

    STEP: Stochastic Traversability Evaluation and Planning for Risk-Aware Off-road Navigation; Results from the DARPA Subterranean Challenge

    A. Dixit, D. D. Fan, K. Otsu, S. Dey, A.-A. Agha-Mohammadi, and J. W. Burdick, “Step: Stochastic traversability evaluation and planning for risk-aware off-road navigation; results from the darpa subterranean challenge,”arXiv preprint arXiv:2303.01614, 2023

    work page Pith review arXiv 2023

  8. [8]

    Fast Traversability Estimation for Wild Visual Navi- gation,

    J. Frey, M. Mattamala, N. Chebrolu, C. Cadena, M. Fallon, and M. Hutter, “Fast Traversability Estimation for Wild Visual Navi- gation,” inProceedings of Robotics: Science and Systems, Daegu, Republic of Korea, July 2023

    work page 2023

Show all 27 references
  1. [9]

    Scaling test- time compute without verification or rl is suboptimal,

    A. Setlur, N. Rajaraman, S. Levine, and A. Kumar, “Scaling test- time compute without verification or rl is suboptimal,”arXiv preprint arXiv:2502.12118, 2025

    2025 arXiv

  2. [10]

    Dtg : Diffusion-based trajectory generation for mapless global navigation,

    J. Liang, A. Payandeh, D. Song, X. Xiao, and D. Manocha, “Dtg : Diffusion-based trajectory generation for mapless global navigation,” IEEE International Conference on Intelligent Robots and Systems (IROS), 2024

    2024

  3. [11]

    Safediffuser: Safe planning with diffusion probabilistic models,

    W. Xiao, T.-H. Wang, C. Gan, and D. Rus, “Safediffuser: Safe planning with diffusion probabilistic models,”arXiv preprint arXiv:2306.00148, 2023

    2023 arXiv

  4. [12]

    Control barrier function based quadratic programs for safety critical systems,

    A. D. Ames, X. Xu, J. W. Grizzle, and P. Tabuada, “Control barrier function based quadratic programs for safety critical systems,”IEEE Transactions on Automatic Control, vol. 62, no. 8, pp. 3861–3876, 2016

    2016

  5. [13]

    Gr00t n1: An open foundation model for generalist humanoid robots,

    J. Bjorck, F. Casta ˜neda, N. Cherniadev, X. Da, R. Ding, L. Fan, Y . Fang, D. Fox, F. Hu, S. Huanget al., “Gr00t n1: An open foundation model for generalist humanoid robots,”arXiv preprint arXiv:2503.14734, 2025

    2025 arXiv

  6. [14]

    Helix: A vision-language-action model for generalist humanoid control,

    A. Figure, “Helix: A vision-language-action model for generalist humanoid control,”Figure AI News, 2024

    2024

  7. [15]

    Monte carlo tree diffusion for system 2 planning,

    J. Yoon, H. Cho, D. Baek, Y . Bengio, and S. Ahn, “Monte carlo tree diffusion for system 2 planning,”arXiv preprint arXiv:2502.07202, 2025

    2025

  8. [16]

    Nomad: Goal masked diffusion policies for navigation and exploration,

    A. Sridhar, D. Shah, C. Glossop, and S. Levine, “Nomad: Goal masked diffusion policies for navigation and exploration,” inIEEE International Conference on Robotics and Automation (ICRA), 2024

    2024

  9. [17]

    Diffusion policy: Visuomotor policy learning via action diffusion,

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burchfiel, R. Tedrake, and S. Song, “Diffusion policy: Visuomotor policy learning via action diffusion,”The International Journal of Robotics Research, 2024

    2024

  10. [18]

    Diffusion models beat gans on image synthesis,

    P. Dhariwal and A. Nichol, “Diffusion models beat gans on image synthesis,”Advances in neural information processing systems, vol. 34, pp. 8780–8794, 2021

    2021

  11. [19]

    Constrained synthesis with projected diffusion models,

    J. K. Christopher, S. Baek, and N. Fioretto, “Constrained synthesis with projected diffusion models,”Advances in Neural Information Processing Systems, 2024

    2024

  12. [20]

    Score-based generative modeling through stochastic differential equations,

    Y . Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” inInternational Conference on Learning Rep- resentations (ICLR), 2021

    2021

  13. [21]

    The complexity of robot motion planning,

    J. Canny, “The complexity of robot motion planning,”MIT press, 1988

    1988

  14. [22]

    S. M. LaValle,Planning algorithms. Cambridge university press, 2006

    2006

  15. [23]

    Autonomous spot: Long-range autonomous ex- ploration of extreme environments with legged locomotion,

    A. Boumanet al., “Autonomous spot: Long-range autonomous ex- ploration of extreme environments with legged locomotion,” inIEEE International Conference on Intelligent Robots and Systems (IROS), 2020. APPENDIXI BASELINE IMPLEMENTATION DETAILS

    2020

  16. [24]

    We adhere to the normalization parameters provided in the NoMaD framework and apply velocity-based unnormalization to obtain the final action outputs from the diffusion model

    Vanilla NoMaD : For the Vanilla NoMaD baseline, we utilize the publicly available pre-trained weights released by the original authors and evaluate the model in a zero-shot setting across both simulated and real- world environments. We adhere to the normalization parameters pr...

  17. [25]

    For data preprocessing, we adopt a distance-based waypoint sampling strategy applied across both our collected datasets and the original NoMaD datasets

    Finetuned NoMaD : We collect approximately one hour of driving data in both simulated and real-world environments via teleoperation of the robot, follow- ing the methodology employed in the open-source datasets used by the NoMaD framework. For data preprocessing, we adopt a di...

  18. [26]

    This approach guarantees that the resulting trajectory remains entirely within safe bounds

    Safety Filter: To ensure safety, we implement a simple truncation-based safety filter that truncates the output trajectory at the waypoint immediately preceding the first predicted collision. This approach guarantees that the resulting trajectory remains entirely within safe bounds

  19. [27]

    In simulation, we sample a total of 50 trajectories, while in the real- world setting, we sample 8 trajectories

    Risk Guidance Diffusion : We implement the projected risk guidance mechanism as described in this work, uti- lizing the risk map generated using [7]. In simulation, we sample a total of 50 trajectories, while in the real- world setting, we sample 8 trajectories. Risk-guidance ...

Pith tools