cs.RO提交于 2026-10-30已译

闭环机器人软件的学习与迁移

Learning and Transferring Closed-Loop Robot Software

So Kuroki · Kento Kawaharazuka · Yujin Tang

中文摘要

闭环机器人策略需要观测处理、状态管理和依赖情境的分支，人工设计与调参成本高昂。编码智能体虽日益支持控制代码生成与优化，但源任务上改进的实现能否支持新任务的策略获取仍不清楚。将完整闭环实现视为可复用的执行经验：为学习每个源策略，编码智能体从少量成功演示生成策略代码，并利用仿真反馈迭代改进，验证选出的实现存入软件档案。迁移到新任务时，智能体结合档案实现、目标演示与执行反馈生成并改进策略。在 RoboCasa 的四个源任务上，迭代优化将成功率从初始生成策略代码的 28.3% 提升至验证选出策略代码的 64.2%，并在物理 Unitree G1 上验证。在 RoboCasa 另外九个目标任务上，无参考时成功率 45.2%，使用源任务初始生成策略代码为 41.5%，使用其验证选出策略代码为 57.0%。结果表明改进源任务代码也能改善新任务上的学习。

关键要点

01闭环机器人策略需观测处理、状态管理与分支，人工设计调参成本高
02编码智能体从少量演示生成策略代码，并用仿真反馈迭代改进
03RoboCasa 四个源任务成功率从 28.3% 升至 64.2%，并在 Unitree G1 验证
04九个目标任务迁移成功率：无参考 45.2%、初始代码 41.5%、验证选出代码 57.0%
05改进源任务代码可提升新任务学习，但初始代码迁移反而低于无参考

解读

尚无解读。

原始英文摘要

arXiv:2609.19906v2 Announce Type: replace 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.