arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2310.16014cs.ROcs.AIcs.CVcs.LG
收藏

人在环中的任务与运动规划用于模仿学习

Human-in-the-Loop Task and Motion Planning for Imitation Learning

NVIDIA(英伟达);Georgia Institute of Technology(佐治亚理工学院)

Ajay Mandlekar, Caelan Garrett, Danfei Xu, Dieter Fox

AI总结:

提出HITL-TAMP系统,结合TAMP与人类遥操作,通过门控机制提升数据收集效率并训练出高性能的模仿学习策略。

摘要:

从人类演示中进行模仿学习可以教会机器人复杂的操作技能,但十分耗时且劳动密集。相比之下,任务与运动规划(TAMP)系统是自动化的,擅长解决长时程任务,但难以应用于富接触任务。在本文中,我们提出了人在环中的任务与运动规划(HITL-TAMP),这是一种利用两种方法优势的新颖系统。该系统采用TAMP门控控制机制,选择性地向人类遥操作员移交和收回控制权。这使人类遥操作员能够管理一支机器人队伍,最大化数据收集效率。收集到的人类数据随后与模仿学习框架相结合,以训练TAMP门控策略,与在完整任务演示上训练相比,这带来了更优的性能。我们将HITL-TAMP与传统的遥操作系统进行了比较——在相同时间预算下,用户收集的演示数量是原来的3倍以上。此外,仅用10分钟的非专家遥操作数据,就能训练出熟练的智能体(75%+成功率)。最后,我们使用HITL-TAMP在12个富接触、长时程任务上收集了2.1K个演示,并表明该系统通常能产生近乎完美的智能体。视频和更多结果请见 https://hitltamp.github.io 。

英文摘要:

Imitation learning from human demonstrations can teach robots complex manipulation skills, but is time-consuming and labor intensive. In contrast, Task and Motion Planning (TAMP) systems are automated and excel at solving long-horizon tasks, but they are difficult to apply to contact-rich tasks. In this paper, we present Human-in-the-Loop Task and Motion Planning (HITL-TAMP), a novel system that leverages the benefits of both approaches. The system employs a TAMP-gated control mechanism, which selectively gives and takes control to and from a human teleoperator. This enables the human teleoperator to manage a fleet of robots, maximizing data collection efficiency. The collected human data is then combined with an imitation learning framework to train a TAMP-gated policy, leading to superior performance compared to training on full task demonstrations. We compared HITL-TAMP to a conventional teleoperation system -- users gathered more than 3x the number of demos given the same time budget. Furthermore, proficient agents (75\%+ success) could be trained from just 10 minutes of non-expert teleoperation data. Finally, we collected 2.1K demos with HITL-TAMP across 12 contact-rich, long-horizon tasks and show that the system often produces near-perfect agents. Videos and additional results at https://hitltamp.github.io .

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。

↑