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.

翻译：演示学习（LfD）作为一种高效框架，能够将类人技能传授给机器人。然而，设计一种能够在动态环境中无缝模仿、泛化并对干扰做出响应的LfD框架，以应对长时域操作任务，仍然是一个挑战。为应对这一挑战，我们提出了逻辑动态运动基元（Logic-DMP），该方法将任务与运动规划（TAMP）与DMP的最优控制公式相结合，使我们能够纳入运动层面的经由点规范，并处理动态环境中任务层面的变化或干扰。我们通过三个长时域操作任务，对所提方法与多种基线方法进行了比较分析，评估了其泛化能力和响应性。实验结果表明，Logic-DMP在处理长时域操作任务中的任务层面变体与干扰时，具有快速的泛化能力和响应性。