Active disturbances rejection control of a quadruped robot leg

This paper is focused on the control of a biologically-inspired quadruped robot leg. After a brief introduction of mechanical construction, the leg´s dynamic equations in operational space are built and active disturbances rejection controllers are designed for three degrees of freedom (DOF). The extended Jacobian matrix method is used to optimize redundant DOF. The core of this work is the comparison of performance between Active Disturbances Rejection Control (ADRC) and conventional PD control, and the robustness of ADRC in disturbance or model uncertainty conditions. Simulation results suggest that ADRC can solve the contradiction between rapid responses and overshoot well and improve the end positioning accuracy by an order of magnitude. Its behavior worsens little when model offsets or disturbed by measuring noise.