Compliance control method for robot joint with variable stiffness

Abstract

Aiming at the disadvantages of insufficient flexibility and poor stability of the traditional control methods of variable stiffness joint robots, a new multi-degree-of-freedom robot joint compliance control method is proposed. The multi-degree-of-freedom robot joint system is introduced, and the multi-degree-of-freedom robot dynamic model is constructed using the Lagrangian method. On this basis, the control algorithm based on feedback linearisation and adaptive RBF neural network realises the compliance control of the multi-degree-of-freedom robot manipulator wrist joint. First, the dynamic model of the robot joint is analysed, and the nonlinear state-space model is linearised using the feedback linearisation method. Then, the fourth-order Runge-Kutta method is used to improve the flexibility of robot joint control when solving the dynamic model, and carried out simulation verification. The simulation results show that the proposed method can converge faster in the control process of the desired angle and the desired stiffness of the variable stiffness joint, and it is robust to the uncertainty of the robot system and the changing external interference.