Virtual-model-based compliance control for pushing recovery of position controlled humanoid robots

Abstract

Motion of humanoid robots in human environments is subject to various unknown disturbances. Standing and walking in such environments, especially under unknown external disturbances, has always been a problem. We hope to achieve full-body compliance control for pushing recovery of position controlled humanoid robots under unknown disturbances. So, we propose a compliance controller for standing maintenance based on virtual model control. The controller uses the concept of impedance and admittance control. An AMPM (Angular Momentum inducing inverted Pendulum Model)-based virtual model with variable gain is adopted to generate the recovery force while a viscoelasticity-based admittance controller is used to apply the recovery force to the desired trajectory, which could be adjusted to meet the performance of the robot. Simulations in V-REP by exerting external force demonstrate the effectiveness and robustness of our method.