论文arXiv 2611.04226今天

DiMuRe框架实现机器人扰动恢复81.9%成功率及LIBERO-Disturb基准

提出DiMuRe诊断引导多级恢复框架，结合程序复用与VLA抓取应对执行扰动。在LIBERO-Disturb基准801个测试用例中，完整系统任务成功率达81.90%，较原程序提升23个百分点，且物理步数更少。

AI 整理

仿真与训练
具身模型
评测与安全

本文由 AI 根据公开内容整理，可能有误，请以原文为准：DiMuRe: Diagnosis-Guided Multilevel Recovery for Robotic Task Execution（arXiv 2611.04226）

核心内容

针对机器人任务执行中扰动导致前置条件失效问题，研究提出DiMuRe诊断引导多级恢复框架。该系统利用阶段特定证据与方法前置条件，动态选择程序复用或预定义恢复技能（含受限VLA抓取）。在原始程序退出后，多模态大模型可基于图像与执行反馈生成修正代码，并通过本地验证检查恢复结果。研究同步发布LIBERO-Disturb基准，专门评估智能体在执行中途受扰下的鲁棒性与恢复能力。

在LIBERO-Disturb基准的801个计划扰动Encore开发案例中，仅依赖程序复用与恢复技能的Local DiMuRe任务完成率达77.65%，相比同等物理步数限制下的原程序58.68%显著提升。引入终端重规划、额外步数允许及早期交接机制的Full DiMuRe进一步将成功率推至81.90%。实验表明Local DiMuRe在每个计划回合中消耗的物理步数均少于原程序，验证了该框架在保持任务连续性的同时具备更高的执行效率与容错边界。

关键看点

多级恢复机制：基于阶段证据与前置条件动态切换程序复用、预定义技能及有界VLA抓取策略
性能量化提升：Full DiMuRe在801个扰动用例中达81.90%成功率，较原程序58.68%提升23.22%
执行效率优化：Local DiMuRe在同等物理步数限制下成功率77.65%，且单回合消耗步数更少
专用评测基准：发布LIBERO-Disturb数据集，支持程序化智能体与VLA策略的中途扰动恢复评估

工程与产业观察

该工作确立了“诊断引导+多级回退”作为具身智能长程任务鲁棒性的有效范式。通过结构化复用既有程序而非全盘重规划，显著降低了LLM/VLA在扰动恢复中的算力开销与幻觉风险。LIBERO-Disturb基准填补了中途扰动恢复评测空白，为验证机器人在非结构化环境下的工程落地能力提供了标准化标尺。

英文摘要 (Abstract)

Disturbances can invalidate the conditions required by a robot's next action even when much of its task program remains useful. A recovery system must therefore choose a response suited to the current state. We present DiMuRe (diagnosis-guided multilevel recovery), a framework for robot-use agents that uses phase-specific evidence and method preconditions to select program reuse or predefined recovery skills, including bounded vision-language-action (VLA) grasping. After the original program exits, a multimodal large language model (MLLM) can generate corrective code from images and execution feedback. Local verification checks the outcome of each recovery attempt. We also introduce LIBERO-Disturb, a benchmark for robustness and recovery under mid-execution disturbances, supporting program-based agents and VLA policies. On 801 scheduled disturbed Encore development cases, Local DiMuRe (program reuse and recovery skills) achieves 77.65% complete-task success, compared with 58.68% for the original program under the same physical-step allowance. Full DiMuRe achieves 81.90% with terminal replanning, an additional step allowance, and early handoff. Local DiMuRe also uses fewer physical steps per scheduled episode than the original program.

作者：Renkai Yi, Yuwei Wu, Haoyu Zhang, Penghao Dai, Zhi Gao, Jin Chen, Che Sun