Robust Model Predictive Control for humanoids standing balancing

Abstract

This paper presents the implementations of Model Predictive Control for the standing balance control of a humanoid to reject external disturbances. The strategies allow the robot to have a compliant behaviour against external forces resulting in a stable and smooth response. The first, ZMP based controller, compensates for the center of mass deviation while the second, attitude controller, regulates the orientation of the body to counterbalance the external disturbances. These two control strategies are combined as an integrated stabilizer, which further increases the effectiveness. Simulation studies on the COMAN humanoid are presented and the data are analysed. The simulations show significant improvements in rejection of external disturbances compared to an existing compliant stabilizer.