Biped Robot Walking using a Combination of Truncated Fourier Series and GALA (Genetic Algorithm parameters adaption using Learning Automata)

Abstract

Controlling a biped robot with a high degree of freedom to achieve stable, straight and fast movement patterns is one of the most complex problems. With growing computational power of computer hardware, simulation of such robots in high resolution real time environment has become more applicable. This paper introduces a novel approach to Generate Bipedal gait for humanoid locomotion. In this scene, first we have used a modified Truncated Fourier Series (TFS) to generate angular trajectories, then to find the best angular trajectory we built an improved Genetic Algorithm (GA). One of the major difficulties of GAs is choosing appropriate values for mutation and crossover parameters. Hence, we present GALA (Genetic Algorithm parameters adaption using Learning Automata) to adjust these parameters by recruiting Learning Automata. As results show, my approach could generate better values for angular trajectories for biped walking, hence in my approach the robot could walk with high stability and faster than other approaches. Evaluations performed on Simulated NAO robot in RoboCup 3D soccer simulation environment.