mlsanityData to 13 Oct 2026

Uncovering Vulnerability of Vision-Language-Action Models under Joint-Level Physical Faults

arXiv 2606.10501 ↗First posted 9 Jun 2026cs.ROcited by 2

On the map: Vision-language-action models → Failure recovery and robustness

Abstract (arXiv)

Deploying Vision-Language-Action (VLA) models in real robotic systems requires robustness not only to semantic and perceptual variations, but also to embodiment-side faults that change how actions are physically realized. Real robots can experience joint-level changes caused by actuator degradation, hardware faults, safety limits, collision damage, or wear-induced friction. These faults are critical because they alter the action-to-motion interface of a policy, disrupting the learned closed-loop relationship between commanded actions, realized motion, and subsequent observations. In this work, we study realistic joint-level physical faults and show that VLA models are vulnerable when predicted actions are executed through a perturbed robot body. Our analysis reveals joint-dependent effects, with heterogeneous degradation in task success across affected joints. We also show that performance drops cannot be attributed solely to physical infeasibility, since feasible faults such as increased joint friction can still substantially reduce success rates and induce closed-loop execution mismatch. Motivated by these findings, we propose Joint-level Physical-fault Aware Residual Calibrator (J-PARC), a lightweight residual calibration framework built on top of a frozen VLA policy. J-PARC infers a latent joint-fault regime from recent joint dynamics and conditions a shared residual calibrator on this regime, enabling adaptive action correction across faulty joints. Experiments show that J-PARC improves robustness under joint-level faults while preserving fault-free environment performance.

Key facts. Cited by 2 papers in the map (data through 13 Oct 2026). Placed in Failure recovery and robustness, one of 13 directions in Vision-language-action models. Builds on 25 of its 37 references that are in the map. First posted on arXiv 9 Jun 2026.

Builds on

25 of 37 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Models15 Oct

Other references · 24

  1. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,022
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,823
  3. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,028
  4. 2020Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problemscited by 1,889
  5. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,699
  6. 2024Octo: An Open-Source Generalist Robot Policycited by 1,450
  7. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,368
  8. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,197
  9. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 862
  10. 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 656
  11. 2025UniVLA: Learning to Act Anywhere with Task-centric Latent Actionscited by 397
  12. 2024A Survey on Vision-Language-Action Models for Embodied AIcited by 277
12 more
  1. 2025A Survey on Vision-Language-Action Models: An Action Tokenization Perspectivecited by 96
  2. 2024Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Roboticscited by 72
  3. 2024VLATest: Testing and Evaluating Vision-Language-Action Models for Robotic Manipulationcited by 44
  4. 2025BadVLA: Towards Backdoor Attacks on Vision-Language-Action Models via Objective-Decoupled Optimizationcited by 38
  5. 2025Eva-VLA: Evaluating Vision-Language-Action Models' Robustness Under Real-World Physical Variationscited by 24
  6. 2025RobustVLA: On Robustness of Vision-Language-Action Model against Multi-Modal Perturbationscited by 24
  7. 2025When Alignment Fails: Multimodal Adversarial Attacks on Vision-Language-Action Modelscited by 19
  8. 2025RobustVLA: Robustness-Aware Reinforcement Post-Training for Vision-Language-Action Modelscited by 17
  9. 2024Probing the Robustness of Vision-Language Pretrained Models: A Multimodal Adversarial Attack Approachcited by 7
  10. 2024Multi-Task Learning of Active Fault-Tolerant Controller for Leg Failures in Quadruped robotscited by 4
  11. 2024Adaptive Compensation for Robotic Joint Failures Using Partially Observable Reinforcement Learningcited by 3
  12. 2026Moving On, Even When You're Broken: Fail-Active Trajectory Generation via Diffusion Policies Conditioned on Embodiment and Taskcited by 2

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Beyond Patch Removal: Persistent Adversarial Effects in Vision-Language-Action Policiescited by 0
  2. 2026CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encodingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Taming VLAs under Robot Execution Errors: Self-Compensation and Stress Testing
  2. 2026Blackout vs. Freeze: Analyzing Physical Failure Modes of VLAs under Camera Faults
  3. 2026ReconVLA: An Uncertainty-Guided and Failure-Aware Vision-Language-Action Framework for Robotic Control
  4. 2026FATE-VLA:Failue-aware test generation for vision-language-action models
  5. 2026Training-Free Action Correction for VLA Model Failures via Language Feedback
  6. 2026PhysVLA: Towards Physically-Grounded VLA for Embodied Robotic Manipulation
  7. 2026ForesightSafety-VLA: A Unified Diagnostic Safety Benchmark for Vision-Language-Action Models
  8. 2026Benchmarking Vision-Language-Action Models on SO-101: Failure and Recovery Analysis