闭环机器人软件的学习与迁移：编码智能体借助软件档案将 RoboCasa 迁移成功率提升至 57.0%
Learning and Transferring Closed-Loop Robot Software
AI 导读
研究人员提出把完整闭环策略实现当作可复用的执行经验，由编码智能体根据少量成功演示生成策略代码，并用仿真反馈迭代优化，验证通过的实现存入软件档案。在 RoboCasa 四个源任务上，迭代优化将成功率从初始代码的 28.3% 提升到验证筛选代码的 64.2%，并在实体 Unitree G1 上验证了该学习方法。
正文 · 原文
Abstract:Closed-loop robot policies require observation processing, state management, and situation-dependent branching, making them costly to design and tune manually. Although coding agents increasingly support control-code generation and optimization, it remains unclear whether implementations improved on source tasks also support policy acquisition for new tasks. We study this question by treating complete closed-loop implementations as reusable execution experience. To learn each source policy, a coding agent generates policy code from a few successful demonstrations and iteratively improves it using simulation feedback. The validation-selected implementations are retained in a software archive. To transfer the acquired software to new tasks, the agent generates and improves policies using archived implementations, target demonstrations, and execution feedback. We evaluate policy learning and its relationship to cross-task transfer. For learning, across four source tasks in RoboCasa, iterative optimization increases the success rate from 28.3% with initially generated policy code to 64.2% with validation-selected policy code. We further validate this learning approach on a physical Unitree G1. For transfer, across nine additional target tasks in RoboCasa, success rates are 45.2% without references, 41.5% using initially generated policy code from source tasks, and 57.0% using their validation-selected policy code. These results show that improving source-task code also improves learning on new tasks.
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.19906 [cs.RO] |
| (or arXiv:2609.19906v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2609.19906 arXiv-issued DOI via DataCite |
Submission history
From: So Kuroki [view email]
[v1]
Thu, 17 Sep 2026 08:51:12 UTC (1,237 KB)
[v2]
Thu, 29 Oct 2026 10:18:41 UTC (3,039 KB)
来源：arXiv cs.AI · arxiv.org