aiqbench analysis
ReCoVLA addresses the brittleness of Vision-Language-Action (VLA) policies by introducing a framework that leverages an external VLM to compile structured rewards for failure recovery, rather than generating actions directly. By keeping the pretrained VLA frozen and decoupling high-level failure understanding from low-level control, the method achieves significant improvements in success rates across both simulation and zero-shot physical deployment. This approach offers a promising pathway for enhancing the robustness of robotic manipulation systems in off-nominal states without requiring full retraining of the base policy.
ReCoVLA: VLM-Guided Reward Compilation for Failure Recovery in Vision-Language-Action Policies
arXiv:2606.09630v1 Announce Type: cross Abstract: Vision-language-action (VLA) policies provide strong priors for language-conditioned manipulation, but remain brittle in off-nominal states requiring targeted recovery. We propose ReCoVLA -- a failure-conditioned residual recovery framework that keeps a pretrained VLA policy frozen, uses an external vision-language model (VLM) to infer the failure mode and recovery stage, and compiles a structured reward from task-relevant components. Rather than using the VLM to generate actions or rewards directly, ReCoVLA uses it as a semantic reward selector: it predicts a recovery descriptor and reward mask for in-simulation residual-policy training, followed by zero-shot sim-to-real deployment of the trained recovery policies. This decouples high-level failure understanding from low-level corrective control to support different VLAs. Experiments across short-horizon, long-horizon, and contact-rich manipulation tasks show that ReCoVLA outperforms the tested baselines on average. In simulation, our reward compiler improves average success from 36.7% for the fine-tuned $\pi_{0.5}$ baseline to 66.7%. In physical zero-shot sim-to-real experiments, ReCoVLA achieves the best average performance, with 61.7% success.
Abstract:Vision-language-action (VLA) policies provide strong priors for language-conditioned manipulation, but remain brittle in off-nominal states requiring targeted recovery. We propose ReCoVLA -- a failure-conditioned residual recovery framework that keeps a pretrained VLA policy frozen, uses an external vision-language model (VLM) to infer the failure mode and recovery stage, and compiles a structured reward from task-relevant components. Rather than using the VLM to generate actions or rewards directly, ReCoVLA uses it as a semantic reward selector: it predicts a recovery descriptor and reward mask for in-simulation residual-policy training, followed by zero-shot sim-to-real deployment of the trained recovery policies. This decouples high-level failure understanding from low-level corrective control to support different VLAs. Experiments across short-horizon, long-horizon, and contact-rich manipulation tasks show that ReCoVLA outperforms the tested baselines on average. In simulation, our reward compiler improves average success from 36.7% for the fine-tuned $\pi_{0.5}$ baseline to 66.7%. In physical zero-shot sim-to-real experiments, ReCoVLA achieves the best average performance, with 61.7% success.
Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2606.09630 [cs.RO]
(or arXiv:2606.09630v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.09630 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 257 words
