ReCoVLA: VLM引导的奖励编译用于视觉-语言-动作策略的故障恢复
ReCoVLA: VLM-Guided Reward Compilation for Failure Recovery in Vision-Language-Action Policies
AI总结：
提出ReCoVLA框架，通过冻结预训练VLA策略，利用外部VLM推断故障模式并编译结构化奖励，训练残差恢复策略，实现零样本仿真到真实部署，在多种操作任务中提升成功率。
摘要：
视觉-语言-动作（VLA）策略为语言条件操作提供了强大的先验知识，但在需要针对性恢复的非标称状态下仍然脆弱。我们提出ReCoVLA——一种故障条件的残差恢复框架，它保持预训练的VLA策略冻结，使用外部视觉-语言模型（VLM）推断故障模式和恢复阶段，并从任务相关组件编译结构化奖励。ReCoVLA并非使用VLM直接生成动作或奖励，而是将其作为语义奖励选择器：它预测恢复描述符和奖励掩码，用于仿真中的残差策略训练，随后将训练好的恢复策略零样本部署到真实世界。这解耦了高层故障理解与低层纠正控制，以支持不同的VLA。在短时域、长时域和接触丰富的操作任务上的实验表明，ReCoVLA在平均性能上优于测试的基线。在仿真中，我们的奖励编译器将微调$π_{0.5}$基线的平均成功率从36.7%提升到66.7%。在物理零样本仿真到真实实验中，ReCoVLA取得了最佳平均性能，成功率为61.7%。
英文摘要：
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 $π_{0.5}$ baseline to 66.7%. In physical zero-shot sim-to-real experiments, ReCoVLA achieves the best average performance, with 61.7% success.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
