Design and torque control base on neural network PID of a variable stiffness joint for rehabilitation robot

Abstract

Variable stiffness joints have been gradually applied in rehabilitation robots because of their intrinsic compliance and greater ability to adjust mechanical stiffness. This paper designs a variable stiffness joint for upper limb rehabilitation training. The joint adopts the variable stiffness principle based special curved surface. The trapezoidal lead screw in the variable stiffness module has a self-locking function, and the stiffness can be maintained without the continuous output torque of the motor. In the aspect of control, back propagation (BP) neural network PID control strategy is used to control the torque of variable stiffness joint. Experiments show that this control method can effectively improve the torque control performance of variable stiffness joints in the case of low stiffness, and the isotonic centripetal resistance training can be realized by using the joints and control methods designed in this paper.