Abstract: A seven-link biped robot model was chosen for gait stability optimization in climbing upstairs. Motions of the feet during a regular step were designed by the 3^rd order spline; motions of the hip were designed through acceleration analysis, and the locus of ZMP was obtained by computing the dynamic model. Thus the gait stability of the biped robot could be shown as a multi variable optimal problem in connection with the placement and the pose of the upper body of the robot. Finally, genetic algorithm (GA) was applied to get the optimal solution. The simulations of the biped robot climbing upstairs showed that the method work satisfactorily.

Key words: biped robot, genetic algorithm, gait optimization, upstairs