Biologically Inspired Energy Efficient Walking for Biped Robots

Abstract

Prior methods of bipedal humanoid walking through high gain position control is energy intensive. This paper introduces a biologically inspired method for robot locomotion that improves energy efficiency. Neural oscillators were used as the central pattern generator of the gait pattern and were able to entrain the natural frequency for robot locomotion. The impedance of the legs was modulated so that the legs can swing more freely during swing phase and control position accurately during support phase. Dynamic simulations of humanoid robot walking with the proposed method resulted in energy efficiency improvement of 40.5%. Also, the modulation of impedance resulted in a more stable contact with the ground during landing of the foot.