Application of variable zero-moment point in walking control of the biped robot

Abstract

Using the predictive control based on zero-moment point (ZMP), the biped robot can walk comparatively stably. However, the problems such as lack of self-adaptivity are also highlighted mainly on account of modeling errors and environmental perturbations; specifically, the tracking errors of ZMP are generated, leading to a reduced walking stability. To address this problem, in the present work, the expected ZMP was decomposed into the reference ZMP which is pre-planned offline, and the variable ZMP which can be varied in real time. With the addition of the variable ZMP, the outside interferences can be eliminated. By combining the predictive control system and the inverse system of variable ZMP, the walking pattern of the robot with favorable self-adaptivity can be achieved. Finally, the simulation results indicate that the self-adaptivity of the robot can be effectively improved using the proposed control system.