A Novel Biologically Inspired Neural Network and Its Application on the Locomotion Control of Quadruped Robot

This paper proposed a novel Biologically Inspired Neural Networks-central pattern generator (CPG) model with proprioceptive and reflex mechanism. It not only contains inner sensory information but also contains the information of the actuators and automatically tunes the parameter of CPG correspond to the actuators information and inner sensory information. The position of the joints linked directly with the output of CPG is introduced to the CPG to found its proprioceptive system, spontaneously making the robot can realize the actuator working status, furthermore changing CPG output to fit the change and decrease influence of the problematic joints or actuators on the robot controlled. The reflex mechanism can tune the CPG output to fit the environment to avoiding obstacle and adapting the ground. Its application on the locomotion control of a quadruped robot demonstrates the effectiveness of the proposed approach.