arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2404.16138cs.RO
收藏

面向动态环境中多步操作任务的演示逻辑学习

Logic Learning from Demonstrations for Multi-step Manipulation Tasks in Dynamic Environments

Idiap Research Institute(伊迪亚普研究所);EPFL(洛桑联邦理工学院)

Yan Zhang, Teng Xue, Amirreza Razmjoo, Sylvain Calinon

AI总结:

针对动态环境长时序操作任务的LfD框架难兼顾模仿、泛化与抗扰性的问题,提出结合TAMP与DMP最优控制的Logic-DMP,经三项任务实验验证其具备快速泛化与响应能力。

摘要:

从演示中学习(LfD)是向机器人传授类人技能的高效框架。然而,设计一款能够在动态环境中的长时序操作任务上实现流畅模仿、泛化并对干扰做出响应的LfD框架仍是一项挑战。为解决该问题,我们提出逻辑动态运动基元(Logic-DMP),它将任务与运动规划(TAMP)与DMP的最优控制公式相结合,能够纳入运动层的途经点规范,并应对动态环境中的任务层变化或干扰。我们将所提方法与多个基线进行对比分析,在三项长时序操作任务中评估其泛化能力与响应性。实验表明,Logic-DMP在处理长时序操作任务中的任务层变体与干扰时具备快速泛化能力与响应性。

英文摘要:

Learning from Demonstration (LfD) stands as an efficient framework for imparting human-like skills to robots. Nevertheless, designing an LfD framework capable of seamlessly imitating, generalizing, and reacting to disturbances for long-horizon manipulation tasks in dynamic environments remains a challenge. To tackle this challenge, we present Logic Dynamic Movement Primitives (Logic-DMP), which combines Task and Motion Planning (TAMP) with an optimal control formulation of DMP, allowing us to incorporate motion-level via-point specifications and to handle task-level variations or disturbances in dynamic environments. We conduct a comparative analysis of our proposed approach against several baselines, evaluating its generalization ability and reactivity across three long-horizon manipulation tasks. Our experiment demonstrates the fast generalization and reactivity of Logic-DMP for handling task-level variants and disturbances in long-horizon manipulation tasks.

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。

↑