Optimal Gait Synthesis of a 34-DOF Humanoid Robot on Uneven Ground

Abstract

This paper describes kinematics, dynamics, body trajectory generation, and gait optimization of a 34-Degrees of freedom (DOF) Humanoid robot (HR) for a balanced walk on 3D uneven terrain. The equation of motion of 34-DOF HR is simplified by a systematic analysis of the flow of kinematics and dynamics (KAD) information through the kinematics chains (KCs) elucidated by KAD information flow diagrams. Thirty parameters parametrize hip and foot-trajectory, joint trajectories of the upper body, and the gait duration. A multi-objective genetic algorithm (GA) solves the HR optimal gait generation problem by minimizing the energy required to traverse one step of gait and dynamic force and torque at the mid hip due to the upper body motion and satisfying the constraint of the dynamic balance. We propose a new method to characterize the HR gait by incorporating the terrain unevenness by three angle parameters representing the HR support foot orientation in a reference frame attached to the horizontal ground-plane and HR initial and final configuration. The proposed method generalizes the gait representation on uneven terrain and covers all possible walk conditions within the HR kinematics and dynamics limitation. The proposed method is verified by numerical simulation of the HR gait on uneven terrain.