人体步行规律与仿人机器人步态规划

Abstract

After years of evolution, human walking can be characterized as energy-efficient, stable, natural and coordinated.The humanoid robots have to possess such characteristics in order to adapt to the complicated environment and serve the humankind.This paper analyzes the human walking mechanism from the biomimetic viewpoint, and proposes a method of designing the gait pattern of humanoid robots based on this mechanism.Firstly, the human motion data were captured with the optical motion capture system and analyzed in order to obtain the relationship of the human gait pattern parameters and the determining method of these parameters was proposed with the evaluation index of walking similarity.Then the humanoid robot torque compensation strategy in the yaw direction based on coordinated motion of the arms and waist was proposed by analyzing the principle how a human being succeeded in canceling the yaw torque while walking.Finally, we tested and improved the method through dynamics simulation.The simulation and the actual experiments on the platform of "BHR" humanoid robot demonstrated that more smooth trajectories can be generated and the slip occurred in fast walking before can be avoided with this method.In the future, we will further our studies on the human walking mechanism and explore the methods for fast walking gait planning based on this mechanism.