Design of the Anti-disturbance Virtual Model Controller for Quadruped Robot

The virtual model control (VMC) framework proposed by Marc Raibert (Boston Dynamics) reduces the dimension of the multi-point support balance control problem in quadruped robot, which decouples the stability control of the quadruped robot into height, attitude and speed. The traditional virtual model is constructed by the PD feedback controller (PD-VMC), which can realizes reliable movement on complex terrain. However, the PD controller can not overcome the disturbances cause by system's non-linearity, modeling error and leg swing inertia. The fixed feedback gain of PD controller also makes the poor system adaptive ability on real robot. In this paper, an anti-disturbance virtual model controller (D-VMC) is constructed by LADRC. The federal LESO observer (F-LESO) is proposed to improve the accuracy of state observation by using multiple measurements. Simulations and experiments have verified the effectiveness of the proposed D-VMC controller. It has the highly anti-disturbance ability, which effectively improve the tracking accuracy of the desired attitude, and can effectively guarantee the stable running of the quadruped robot at 1m/s. In addition, the F-LESO observer can estimate the unknown load quality and improve the height control stability.