Study of human-like locomotion for humanoid robot based on human motion capture data

Abstract

In order to realize the stable and human-like walking gait of the humanoid robot, we propose a gait planning method based on both the human motion capture data (HMCD) and the zero moment point (ZMP). Firstly, we draw the motion patterns of human walking gait in the joint space, which is the angle data of 6 DOFs, namely the pitch, roll, and yaw of the hip joint, the pitch of the knee joint, and the pitch and the roll of the ankle joint, of one leg, and apply these motion patterns to the humanoid robot supposing that the motion of the left and the right leg are symmetric and phase reversal. Secondly, we employ ZMP as a balance index to fine-tuning the motion of the humanoid, so that we can achieve a robust walking of humanoid in the situation of disturbance. We validate our approach via simulation in Webots.