Push Recovery Control Based on Model Predictive Control of Hydraulic Quadruped Robots.

This work is aimed at addressing the balance problem of hydraulic quadruped robots, trotting on even terrain, which is impacted by lateral disturbance. The ability of push recovery means that a robot can restore stability when the roll angle of the body is too large after a strong side impact. To maintain the balance of the impacted robot, three strategies are proposed inspired by the human response to external disturbance, including supporting leg adjustment strategy, one-step motion of swinging legs strategy, and N-step motion of swinging legs strategy. Quadruped robots can be considered as humanoids owing to the nature of their trotting gait. Thus, the contributions of this artile are as follows. A simplified dynamic model of a quadruped robot is established based on linear inverted pendulum (LIP), and the idea of capture point (CP) and zero moment point (ZMP). A push recovery control system based on model predictive controller (MPC) is established according to the requirement of the control strategy. Finally, the effectiveness of the push recovery control system is verified by simulation and experiment.