Abstract: To realize the stable and rhythmic walking of quadruped robot, proportion gait control strategy based on central pattern generator (CPG) is explored. The oscillator is modeled by two mutually inhibited neurons. Furthermore, quadruped hip joint CPG network is constructed on the basis of this kind of oscillator model. According to the simulated relation between oscillator parameters and output of the CPG network, a group of parameters of CPG network are determined and a proportion gait is proposed. Gait with different duty factor can be realized by threshold adjustment. The reflex characteristic of gait based CPG network is analyzed. Experiments show that the stable walking can be achieved through this control method based on CPG with good obstacle reflex performance and environmental adaptation.

Key words: Central pattern generator, Gait, Quadruped, Robot