Robust approach for humanoid joint control based on a disturbance observer

Abstract

This study develops a robust method to control the humanoid joint based on a disturbance observer (DOB). In controlling the bipedal robot, it is usually difficult to precisely compensate for the gravity torques, friction forces, model mismatch and so on. The authors regard these as disturbances and employ the DOB to design a robust controller for the humanoid joint. The requirements for joint control are firstly discussed; a single joint model, that is, inverted pendulum model, is illustrated and the open-loop transfer function of the ankle joint is also presented. The authors consider the DOB for compensation, develop the design criterions of the filter and propose the robust servocontrol structure of the joint controller. For a given simulated robot, the performance and robustness of the proposed controller are demonstrated comparing with the proportional–integral–differential controller, and the effect of model perturbations is especially considered in case of the system performance. The results show the validation and feasibility of the proposed method.