Non-time reference gait planning and optimization for a stable bipedal walking

Abstract

A non-time reference gait planning method is proposed. The usual reference variable, time, is substituted by a non-time variable in gait, so the whole gait-planning phase can be divided into two phases, (1) planning the space walking path: Taking the forward locomotion of upper-body as reference variable, considering the constraint of the environment, the walking path of a robot without collision with other objects is designed, thus the relating locomotion of the parts of the robot is obtained; (2) planning the trajectory of the non-time reference variable: according the constraint of ZMP stability, design the forward locomotion of upper-body. The gait-planning problem is changed to the optimization problem. The optimized gait with good stabilities is obtained by Genetic Algorithm (GA). This non time reference gait planning method has advantages in passing obstacles, climbing upstairs or downstairs and other similar situation in which the walking path is specified.