Sensor-Based Biped Walking: A Sagittal Plane Research

This paper aims to explore the principles of biomimetic biped walking in the sagittal plane. Firstly, a new walking pattern denoted as two-point-foot walking pattern is proposed according to the structure of the foot of human being. Secondly, the energy cycle of the new walking pattern is investigated. Based on the features of the energy cycle, a multi-scale predict algorithm is developed to estimate the state of walking. The algorithm can be used in the control of the energy input to ensure the continuity and stability of walking. Then, a sensor based control structure is developed to implement the control of walking-start, acceleration and steady walking. Finally, a 2D simulation system with an 8 DOF planar biped robot is developed in this paper. The result of simulation exhibits a very naturally walking pattern close to that of human being´s slow walking.