HuggingFace Daily Papers（社区热门论文）· HuggingFace Daily Papers（社区热门论文）·· 9 天前AI 评分47
Recursive Harness Distillation：跨智能体蒸馏经验提升机器人操作成功率
AI 导读
研究提出 Recursive Harness Distillation，让强智能体把干预经验蒸馏成 playbook 供轻量智能体使用，并借助轻量智能体的执行反馈递归优化，无需更新模型参数。
正文
Abstract:A central goal in robotics is to enable manipulation across changing tasks and environments. Vision-language-action (VLA) models provide broad manipulation capabilities but can struggle when execution requires diagnosing failures and adapting behavior. Strong agents can discover effective interventions through interaction with these policies. We propose Recursive Harness Distillation to accumulate this experience as reusable guidance across agents. A strong agent distills its experience into a playbook for a light agent, then recursively refines the playbook using the light agent's execution feedback. The resulting playbook enables agents to reuse accumulated intervention knowledge in new task instances without updating model parameters. In real-world manipulation, the harness improves success from 37.3% to 64.0%. On SimplerEnv Bridge, the light agent with the playbook achieves 66.7% success, compared with 41.7% for the GR00T-only baseline, and outperforms the strong agent without a playbook. The same playbook also benefits the strong agent, which reaches 79.2% success. These results demonstrate the feasibility of harness distillation for robotics: intervention experience can be accumulated, refined through execution, and reused across agents to improve manipulation.
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2609.33378 [cs.RO] |
| (or arXiv:2609.33378v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2609.33378 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Seungyeon Kim [view email]
[v1]
Sun, 27 Sep 2026 09:01:15 UTC (9,264 KB)
来源：HuggingFace Daily Papers（社区热门论文） · arxiv.org