Controlling Human-like Locomotion of a Biped by a Biologically Motivated Approach

Controlling biped robots in a human-like way is still in the air in robotics. Due to the existing of shortcomings of technical control methods, researchers began to try to find help from biomechanics, neuroscience, human movement analysis and other biological fields. This paper presents a biologically inspired method of controlling the locomotion of biped robots based on the neurological and biomechanical research of human walking. The control concept looks inside of the human motion control to generate an efficient, robust control approach for bipedal dynamic walking. The framework of the control approach consists of a hierarchical architecture of control network, motor patterns and reflexes that works locally and distributed to exploit the inherent dynamics of of the system. In this paper, the approach is validated on a simulated anthropomorphic biped with 21 DoFs. Several locomotion like standing, walking initiation and cyclic walking have been deeply studied. This paper will study the biomechanical aspect and motion analysis of termination of gait and its application in a biped robot. The corresponding reflexes and motor patterns will be also introduced in this paper.