Trajectory Tracking Control of a Redundantly Actuated Parallel Robot Using Diagonal Recurrent Neural Network

Abstract

Parallel robots have good performance in terms of rigidity, accuracy and dynamic characteristics. In this paper, a 2-DOF redundantly actuated parallel robot is taken as the object of study. Diagonal recurrent neural network (DRNN) is known for its dynamic mapping and fit for nonlinear dynamical systems. A neural network PID controller which is composed of the conventional PID control and the DRNN neural network is proposed. The DRNN neural network makes up the deficiency of the conventional PID control, and strengthens the adaptivity of the whole system. The conventional PID controller is applied to compare with the proposed controller. The two controllers are used to track a straight line under the trapezoidal velocity planning. The obtained results confirm the theoretical findings, i.e., the neural network PID controller can make further reduction on tracking errors. The neural network PID controller can be an effective control approach to improve the trajectory tracking performance of parallel robotic systems.