Humanoid Soccer Gait Generation and Optimization Using Probability Distribution Models

In this paper, humanoid gait generation is formulated as a multi-objective optimization problem with multiconstraint based on probability distribution models. Under this framework, an estimation of distribution algorithm (EDA) based gait optimization approach has been developed to speed up searching in high dimensional coupling space constructed by the permutation of optimization parameters to establish a periodic orbit in biped locomotion. To better understand how information are transferred between these parameters, a factorized distribution algorithm (FDA) based gait optimization method using maximum entropy solution principle has been proposed so that we may progress toward better understanding human locomotion and extend the results to design of humanoid robots. The proposed probability based estimation algorithms have been successfully used to generate and optimize various types of basic soccer-playing humanoid gaits for our humanoid soccer robot Robo-Erectus which has been one of pioneering humanoid soccerplaying robots in the Humanoid League since 2002. Index Terms – Humanoid soccer, gait generation and optimization, probability distribution, evolutionary optimization