Robot motion remix based on motion capture data towards human-like locomotion of humanoid robots

This paper presents preliminary results on generating walking and turning motion of a humanoid robot based on human motion data obtained by using motion capturing system. The target humanoid robot is recently released HRP-4C, which looks like a Japanese woman with realistic geometry. One of the main uses of this robot is to be entertainment. Therefore it is important that the robot's motions are easily created, and feasible for the robot. Captured human motion data are used as a starting point, to which the following modifications are added: step length, speed, rotational angle during turn, and so on. Dynamic balance is also considered. Verification is conducted through both simulation and experiment with HRP-4C. A turning performance based on human motion is successfully demonstrated.