Genetic design of biped walking fuzzy logic controller

Abstract

This paper presents the automatic design, through a multi-objective genetic algorithm, of a fuzzy logic controller (FLC) applied to walking of a biped robot. The design approach of the FLC comprises two aspects: tuning the antecedents and consequents membership functions parameters and obtain the antecedents and consequents fuzzy partition of a rule base with a fixed structure such that allow to perform tracking of a walking pattern. Optimization results of the FLC design for a 3-DOF biped robot and the tracking results are provided.