Optimal Multi-criteria Humanoid Robot Gait Synthesis — an Evolutionary Approach

Humanoid robots operating in everyday life environments must generate the gait based on the environmental conditions. Often the gait has to satisfy different objectives. In this paper, we present a new method for humanoid robot gait generation based on multiobjective evolutionary algorithms. In our method, we consider two different conflicting objectives for the humanoid robot gait generation: minimum energy and minimum torque change. In the difference from single objective genetic algorithm, the multiobjective evolutionary algorithm converges in a set of nondominated Pareto optimal gaits. Based on the environmental conditions and the user requirements, the appropriate humanoid robot gait can be selected. Simulation and experimental results using the “Bonten-Maru” humanoid robot show a good performance in the proposed method.