Whole-body motion imitation using human modeling

Abstract

The purpose of this study is to develop a methodology that enables a humanoid robot to imitate a whole body motion of a human. In the communication and interaction with a human being with motions and gestures, a humanoid robot needs not only to look like a human but to behave as a human does to make sure the meanings of motions and gestures. To act like a human, the humanoid robot has to imitate the kinematical and dynamical movement of human. For this, a methodology for a humanoid robot to imitate human whole body motions, is presented. Human motions are acquired from a motion capture system. The proposed methodology first converts the captured motions into the motions for a humanoid robot considering kinematical differences between human and robot. In addition, the methodology proposes a simplified human model to analyze a human ZMP trajectory using an optimization scheme. The methodology computes a stable ZMP trajectory for the humanoid robot based the ZMP trajectory of this simplified human model. From the computed ZMP trajectory, the whole body motion for the humanoid robot is generated through this off-line process. A human whole body motion, a dancing motion, is converted to examine the developed methodology.