Model Predictive Control with Improved Current Loop Cascaded for Manipulator Systems

In medical manipulator control system, torque-position control with high precision and fast dynamic response is required. In this paper, a control system that cascades model predictive control and improved PI control is proposed. Firstly, in order to balance the gravity of the robot arm, a gravity learning algorithm separating the fictional damping is designed. Then, the disturbance of current loop caused by speed variation is eliminated through the improved PI control with state feedback decoupling. The position control is optimized and the reference current trajectory is obtained through model predictive control. Torque-position control with gravity balance and high dynamic performance is realized through this control method.