Logic Learning From Demonstrations for Multi-Step Manipulation Tasks in Dynamic Environments

• Published in: IEEE robotics and automation letters Vol. 9; no. 8; pp. 7214 - 7221
• Main Authors: Zhang, Yan, Xue, Teng, Razmjoo, Amirreza, Calinon, Sylvain
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.08.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Disturbances; Dynamics; Learning; Learning from demonstrations; Logic; Manipulator dynamics; Motion planning; Movement; Optimal control; Planning; reactive long-horizon manipulation planning; Skeleton; Switches; Task analysis; task and motion planning; Trajectory
• ISSN: 2377-3766; 2377-3766
• DOI: 10.1109/LRA.2024.3418276

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-LfD, which combines Task and Motion Planning (TAMP) with an optimal control formulation of Dynamic Movement Primitives (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-LfD for handling task-level variants and disturbances in long-horizon manipulation tasks.