Using virtual model control and genetic algorithm to obtain stable bipedal walking gait through optimizing the ankle torque

Abstract

This paper presents a new approach of controlling the biped robot to achieve stable dynamic walking by optimizing the stance-leg ankle behavior. In this study, the motion control in the sagittal plane of the robot is formulated using a control framework known as Virtual Model Control. Genetic Algorithm is used to search for optimal gains of the controllers such that stable walking gait with smooth velocity profile can be achieved. The stability of the biped is checked using the zero-moment-point criterion. Simulation results show that the obtained walking gait is stable and its forward velocity profile is very smooth. © 2007 World Scientific Publishing Co. Pte. Ltd.