arXiv：cs.LG· Haoran Hao, Shahram Najam Syed, Jeffrey Ichnowski, Jeff Schneider·· 6 小时前AI 评分40
FAR：面向测试时恢复与持续策略改进的失败感知重试框架
FAR: Failure-Aware Retry for Test-Time Recovery and Continual Policy Improvement
AI 导读
研究人员提出失败感知重试框架 FAR，让机器人在测试时从先前失败中学习并自主完成任务，被 CoRL 2026 接收。FAR 结合失败对比偏好自适应与轻量动作扰动，并将成功恢复轨迹纳入持续策略改进训练循环。仿真和真实操作任务中，其成功率较标准扩散策略平均提升 17.6% 和 11.7%，并改善了重置与时间步预算下的数据效率。
正文
Abstract:Robot policies inevitably encounter failures when deployed in real environments. Naive retries often repeat the same mistakes, while many existing recovery methods rely on human intervention. In this paper, we propose Failure-Aware Retry (FAR), a framework that enables robots to learn from previous failures at test time, adapt their behavior accordingly, and eventually complete the task autonomously. FAR combines Failure-Contrastive Preference Adaptation, which constructs preference learning data from failures to steer the policy away from previously unsuccessful behaviors, with lightweight action perturbations during retries to encourage local exploration. We further incorporate successful recovery trajectories into a training loop for continual policy improvement. Experiments in both simulation and real-world manipulation tasks show that FAR substantially improves success rates and robustness, with average gains of 17.6% over the standard diffusion policy in simulation and 11.7% in the real world. In addition, FAR improves data efficiency under both reset and timestep budgets during continual policy improvement by exploiting informative failure cases. Videos and code are available at this https URL.
| Comments: | Accepted by CoRL 2026. Project Page: this https URL |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2607.01111 [cs.RO] |
| (or arXiv:2607.01111v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2607.01111 arXiv-issued DOI via DataCite |
Submission history
From: Haoran Hao [view email]
[v1]
Wed, 1 Jul 2026 16:01:54 UTC (2,263 KB)
[v2]
Wed, 7 Oct 2026 03:21:38 UTC (2,501 KB)
来源：arXiv：cs.LG · arxiv.org