Abstract
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.
Abstract (translated)
学习演示(LfD)作为一种将人类技能传递给机器人的有效框架,在长时程操作任务中为机器人提供了重要的价值。然而,设计一个能够无缝模拟、推广和应对干扰的LfD框架仍然具有挑战性。为了解决这个问题,我们提出了逻辑动态运动元启发式(Logic-DMP),它结合了任务和动作规划(TAMP)和DMP的最佳控制公式,允许我们通过级联运动点规格,并处理动态环境中的任务级别差异或干扰。我们对所提出的方案进行了与几个基线的比较分析,评估了其在长时程操作任务中的泛化能力和反应性。我们的实验结果表明,Logic-DMP在处理长时程操作任务中的任务级别变异和干扰方面具有快速泛化能力和反应性。
URL
https://arxiv.org/abs/2404.16138