Biped Fast Walking Gait Shaping via Evolutionary Multi-Objective Optimization

Abstract

The multi-objective optimization of the fast walking gait using a multi-objective genetic algorithm (MOGA) is applied to the real biped robot to get the optimal set of walking parameters based on the desired walking performances such as walking speed, swaying in the saggittal plane, and power level during the walk. The robot, which walks in a dynamically stable manner based on the inverted pendulum model, is used as an objective function evaluator in the MOGA process. The method can be used as a biped gait shaping process or a recipe to extract the best performance of the existing robot.