Online humanoid whole-body teleoperation framework with variable time delay for imitating human motion

Abstract

This paper proposes an online humanoid whole-body teleoperation framework with variable time delay for imitating human motion. In order to imitate human motion and maintain stability of the robot, the framework generates the reference motion for the robot based on human motion with a variable time delay according to the admissible conditions for imitating human motion. The human motion is converted to reference motion suitable for the robot through the kinematic mapping process. Then the robot is controlled in the direction of minimizing reference motion error while considering joint limitation and self-collision. The whole-body human motion was measured utilizing the custom developed data acquisition, and teleoperation experiments were performed using a humanoid platform HUBO2+. Successful experiments through the proposed framework confirms that the humanoid robot can reliably imitate human motion which require exact positioning and manipulating.