Recova: Agent-Guided Failure Recovery for Autonomous Robotic Manipulation
作者: Isabella Liu, An-Chieh Cheng, Johan Bjorck, Zhiding Yu, Hongxu Yin, Jan Kautz, Linxi Fan, Yuke Zhu, Sifei Liu
分类: cs.RO
发布日期: 2026-10-01
备注: Project page: https://www.liuisabella.com/Recova
💡 一句话要点
提出Recova框架以解决自主机器人操作中的失败恢复问题
🎯 匹配领域: 支柱一:机器人控制 (Robot Control)
关键词: 自主机器人 操作失败 恢复机制 数字双胞胎 代理引导 DAgger训练 自适应学习
📋 核心要点
- 现有方法在处理操作失败时缺乏有效的恢复机制,导致任务无法继续执行。
- Recova框架通过代理引导的方式,在数字双胞胎中联合开发任务执行与失败恢复,提升了系统的自适应能力。
- 在多个实验设置中,Recova的成功率显著高于基线方法,且人类干预需求大幅降低。
📝 摘要(中文)
操作失败可能导致任务策略无法恢复的场景。学习纠正行为需要可扩展的失败探索和物理基础。本文提出了Recova,一个代理引导的框架,联合开发任务执行和恢复,利用重建的数字双胞胎进行验证和优化。在数字双胞胎中,代理诊断失败,测试纠正程序,并收集成功的任务和恢复回放。在实际部署中,Recova监控进度,调用学习或程序化恢复,验证场景恢复并恢复执行。当没有合适的恢复方案时,人类示范将解决失败并进入学习循环,从而扩展系统的恢复能力。实验结果显示,Recova在多个设置中显著提高了成功率,减少了人类干预。
🔬 方法详解
问题定义:本文旨在解决自主机器人操作中的失败恢复问题,现有方法在面对操作失败时常常无法有效恢复,导致任务中断。
核心思路:Recova框架通过构建数字双胞胎,利用代理引导的方式进行失败诊断和恢复策略的学习,从而实现任务执行与恢复的协同优化。
技术框架:Recova的整体架构包括数字双胞胎的构建、失败诊断模块、恢复策略测试、成功回放收集、实际部署监控和人类示范学习等多个阶段。
关键创新:Recova的主要创新在于将代理引导的恢复机制与数字双胞胎结合,形成一个自我学习和适应的闭环系统,显著提高了恢复能力。
关键设计:在设计中,采用了DAgger训练策略,将物理回放和人类示范数据路由到相应的策略中,优化了恢复能力的学习过程。
🖼️ 关键图片
📊 实验亮点
在六个LIBERO-Pro设置和四个MolmoSpaces类别中,Recova的平均成功率分别达到了78.8%和64.9%,相比最强基线的71.7%和38.0%有显著提升。通过DAgger微调,成功率从23.8%提升至77.5%,而恢复技能进一步提升至87.5%。人类干预需求在四轮收集中从87.5%降至0%。
🎯 应用场景
Recova框架在自主机器人操作领域具有广泛的应用潜力,特别是在制造、物流和服务机器人等场景中。通过提升机器人在复杂环境中的操作能力,减少人类干预,能够显著提高工作效率和安全性,推动智能自动化的发展。
📄 摘要(原文)
Manipulation failures can leave scenes in states from which a task policy cannot recover. Learning corrective behaviors requires scalable failure exploration and physical grounding. We present Recova, an agent-guided framework that jointly develops task execution and recovery in a reconstructed digital twin, then verifies and refines both through real-world experience. In the twin, the agent diagnoses failures, tests corrective programs, and collects successful task and recovery rollouts for separate policies. During deployment, it monitors progress, invokes a learned or programmatic recovery, verifies scene restoration, and resumes execution. When no suitable recovery is available, a human demonstration resolves the failure and enters the learning loop, allowing the system to expand its recovery capabilities. Physical rollouts and human demonstrations are routed to the corresponding policy for DAgger training. Across six LIBERO-Pro settings and four MolmoSpaces categories, Recova achieves 78.8% and 64.9% mean success, compared with 71.7% and 38.0% for the strongest baselines. With parallel collection across four real-robot workstations, DAgger fine-tuning raises mean success from 23.8% to 77.5%, and recovery skills further raise it to 87.5%. Over four collection rounds on one task, observed human intervention falls from 87.5% to 0%. Together, these results show how agent-guided recovery turns failures into reusable capabilities, improving robustness while progressively reducing human intervention. Project page: https://www.liuisabella.com/Recova