arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2607.06323cs.RO
收藏

LAMP:用于灵巧手部操作的潜在运动先验引导的现实世界学习

LAMP: Latent Motion Prior-Guided Real-World Learning for Dexterous Hand Manipulation

Fudan University(复旦大学);Tsinghua University(清华大学);Peking University(北京大学);PsiBot

Xinye Yang, Zhiyuan Ma, Hongze Yu, Yuanpei Chen, Yaodong Yang, Xiaojie Chai, Xinlei Chen, Chao Yu

AI总结:

研究针对灵巧手部现实世界学习易出错的问题,提出含潜在运动先验模块的三阶段学习框架LAMP,先预训练模块,再训练视觉运动策略并通过在线残差RL改进,在真实机器人任务中取得高成功率,提升了学习效果。

摘要:

灵巧手部的现实世界学习仍然很脆弱,因为高维手部动作会放大模仿误差,使强化学习探索容易出现接触中断运动。虽然将模仿学习(IL)与在线强化学习(RL)相结合可以减少人工监督,但在原始手部动作空间中进行无约束探索对物理硬件来说样本效率低且有风险。我们引入了一个潜在运动先验模块(\prior{}),它将最近的手部动作历史映射到一个紧凑的、基于历史条件的潜在先验,并将连续的潜在命令解码为可执行的高维手部目标。基于此先验,\method{}是一个三阶段的现实世界灵巧学习框架:它从演示中预训练\prior{},训练一个视觉运动策略来预测原生手臂命令和潜在手部动作偏移,并在相同的潜在手部动作空间中通过在线残差RL改进策略。这个共享的、可解码的接口让残差探索在已演示的、接触一致的手部动作附近进行局部校正,而不是独立地扰动每个手指关节。我们在四个真实机器人灵巧操作任务上针对原始的、线性的和离散的手部动作接口评估了\method{}。从特定任务的小演示集开始,\method{}实现了56.25%的平均IL成功率,在线RL后提高到98.75%,在三个任务上最终成功率达到100%,在其余任务上达到95%。

英文摘要:

Real-world learning for dexterous hands remains brittle because high-dimensional hand actions amplify imitation errors and make reinforcement-learning exploration prone to contact-breaking motion. While combining imitation learning (IL) with online reinforcement learning (RL) can reduce manual supervision, unconstrained exploration in raw hand-action spaces is sample-inefficient and risky for physical hardware. We introduce a latent motion prior module (\prior{}) that maps recent hand-action histories to a compact, history-conditioned latent prior and decodes continuous latent commands into executable high-dimensional hand targets. Built on this prior, \method{} is a three-stage real-world dexterous learning framework: it pretrains \prior{} from demonstrations, trains a visuomotor policy that predicts native arm commands and latent hand-action offsets, and improves the policy with online residual RL in the same latent hand-action space. This shared, decodable interface lets residual exploration make local corrections near demonstrated, contact-consistent hand motions rather than perturbing every finger joint independently. We evaluate \method{} on four real-robot dexterous manipulation tasks against raw, linear, and discrete hand-action interfaces. Starting from small task-specific demonstration sets, \method{} achieves a 56.25\% average IL success rate and raises it to 98.75\% after online RL, reaching 100\% final success on three tasks and 95\% on the remaining task.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑