This study focuses on the problems of unstable body center of gravity and easy foot slipping when quadruped robots move on slopes. This article proposes a mathematical model of vestibular reflex suitable for slope motion of a full elbow quadruped robot. First, a control network of central pattern generator (CPG) is built in Simulink with a Hopf oscillator, and the quadruped robot model is imported into Simulink. Trot gait is generated by adjusting the phase relationship of the CPG network, and the uphill simulation experiment is conducted with this gait. In order to achieve adaptability and stability of slope motion, a mathematical model of vestibular biological reflection is introduced into the control network, enabling the robot to adjust its body posture and maintain a stable center of gravity based on slope information during slope motion. The simulation results show that the vestibular reflex effectively reduces the vibration and slip of the robot body during slope motion, ultimately proving that the organic integration of the vestibular biological reflex mathematical model and CPG can effectively improve the stability of quadruped robots during slope motion.