Adaptive Walking on Slope of Quadruped Robot Based on CPG

Abstract

Compared with the traditional wheeled robot, the largest advantage of quadruped robot is that it has better adaptability to complex terrain environment. In order to adapt to walking from plane to slope, the vestibulospinal reflex model and flexion reflex model are constructed by using foot contact force sensor and impedance controller. At the same time, the reflex models are integrated into CPG network and combined with the kinematics of the robot to obtain environmental information. By adjusting the equilibrium position of each leg of the robot, the quadruped robot can adapt to the slope and complete the smooth transition and walking from flat ground to slope. In this paper, CPG network is constructed based on the Hopf oscillator model. The results verified by the co-simulation of MATLAB and ADAMS show that the proposed method can make the stable transition of the quadruped robot from flat ground to slope and realize the adaptation to the slope.