Perceptive Planning and Control for Biped Robot Walking with Unexpected Disturbance

Abstract

Biped humanoid robots are unstable intrinsically with unexpected disturbances. Traditional locomotion algorithms use time as reference for cooperative motions of feet, center of mass and other parts of body with proper dynamic model and Zero Moment Point (ZMP) based balance criteria. Yet such cooperation can fail due to unforeseen disturbances on one of robot's parts. In this paper, a new perceptive planning and control scheme is proposed with non-time, but balance related reference for bipedal walking to meet this challenge. Different from traditional gait planning and control, in which trajectories use time as a global reference in joint level, trajectories of left, right foot and center of mass (CoM), as multi-robot, use ZMP related reference with the continuous perceptive ability of unexpected disturbance. The proof-of-concept tests by simulation and hardware of a full-size humanoid robot are demonstrated.