Principle analysis and simulation for biomimetic biped walking

Abstract

One of the main challenges in the control of biped walking is to translate the understanding from human walking to robot walking. In this paper, firstly, some key principles of biomimetic walking are investigated. A novel two-point gait is proposed according to the structure of human feet and the characteristics of human walking. And the influence of length variation of stance leg on walking efficiency is studied. Secondly, the control method of biomimetic biped called passive/active hybrid control is presented. The robot locomotes in a passive state since there is just one contact point between the stance foot and the ground in every moment. In contrast, the motion of the swing leg subjects to a active control, which is referred by the prescribed trajectory. Furthermore, a finite state based control architecture is developed to implement the coordination of walking. To verify the proposed theory, a 2D simulation system with an 8 DOF planar biped robot is developed. The simulation exhibits an efficient walking style.