Phase-Amplitude Reduction-Based Imitation Learning

• DOI: 10.48550/arxiv.2406.03735
• Publication Date: 2024-06-06
• Authors: Satoshi Yamamori; Jun Morimoto
• Keywords: FOS: Computer and information sciences; Computer Science - Robotics; Computer Science - Machine Learning; Robotics (cs.RO); Machine Learning (cs.LG)

In this study, we propose the use of phase-amplitude reduction method to construct an imitation learning framework. Imitating human movement trajectories is recognized as a promising strategy for generating range human-like robot movements. Unlike previous dynamical system-based approaches, our proposed allows not only imitate limit cycle trajectory but also replicate transient from initial or disturbed state cycle. Consequently, offers safer approach that avoids unpredictable motions immediately after disturbances specified state. We first validated by reconstructing simple limit-cycle attractor. then compared with conventional on lemniscate tracking task simulated arm. Our findings confirm can more accurately generate movements converge target periodic attractor standard approach. Subsequently, applied real arm