Human Motion Imitation

Abstract

Reproducing human behaviors and movement has long been the inspiration behind humanoid robotics research. This chapter introduces the field of imitation learning, and the key methods for observing and modeling human movement, and reproducing it by humanoid robots. A key challenge when reproducing human movement is to account for the differences in the kinematic and dynamic properties between the human and the robot and ensure that the robot maintains postural stability while reproducing the movement as closely as possible. This chapter overviews the typical data processing pipeline, starting from human observation and progressing through motion modeling, and reproduction. Next, the various approaches for modeling human movement, adapting it to ensure feasibility and controlling the robot to replicate the desired movement are overviewed. This chapter ends with a discussion of open problems and current research directions.