Optimal walking trajectory generation for a biped robot using genetic algorithm

Abstract

Information of suitable additive via-points, which determine the shapes of trajectories, such as velocities and accelerations at the via-points, are needed for smooth biped walking. Also, incorrect information can cause indifferentiability on the trajectory and result in discontinuous walking motion. In the paper optimal via-points data are found using a genetic algorithm which minimizes the sum of deviation of velocities and accelerations as well as a jerk. As a result, we obtain the continuity on the whole trajectory interval, the energy distribution through optimum velocities and accelerations. Smooth and stable walking with the leg trajectory optimized by genetic algorithm is shown by walking experiments.