Adaptive Locomotion Control of a Hexapod Robot via Bio-Inspired Learning

• DOI: 10.3389/fnbot.2021.627157
• Publication Date: 2021-01-26T06:51:07Z
• Authors: Wenjuan Ouyang; Haozhen Chi; Jiangnan Pang; Wenyu Liang; Qinyuan Ren
• Keywords: Hexapod; Adaptability; Robot locomotion

In this paper, an adaptive locomotion control approach for a hexapod robot is proposed. Inspired from biological neuro systems, 3D two-layer artificial center pattern generator (CPG) network adopted to generate the of robot. The first layer CPG responsible generating several basic patterns and functional configuration determined through kinematics analysis. second controls limb behavior adapt environment change in specific pattern. To enable adaptability controller, reinforcement learning (RL)-based employed tune parameters. Owing symmetrical structure robot, only two parameters need be learned iteratively. Thus, proposed can used practice. Finally, both simulations experiments are conducted verify effectiveness approach.