Binocular visual servoing based on PID neural network

Abstract

The PID neural network was introduced into the control of the robot for image-based binocular visual servo control system with hand-eye model, and a controller was designed which combined the PI motion controller with the PID neural network controller in this paper. PI motion controller gives the desired velocity of the robot joints based on the image errors, and obtains the joint torque from the neural network PID controller. And the torque drives the robot reaching a desired position and orientation. The Lyapunov theory is used to prove asymptotic stability of the PI motion controller. And we compared the tracking performance of BP neural network with PID neural network. Simulation results show the effectiveness of this method.