Research Of Binocular Visual Servo Control Based On The Recurrent Neural Networks

With the constant improvement of the image processing method and the computer performance, the robot visual servo system has achieved rapid development, and it has been widely used in the welding and sealing, conveyor belt of the workpiece assembly, tracking and navigation and so on. As the robot manipulator is consisted of turning link rod, this leads the mathematical model of the robot manipulator has strongly coupling and nonlinear, so the control problem of the robot manipulator in the scientific community has always been a hot research direction. In view of the control problems of the robot manipulator visual servo system, the main research contents of this article is as follows:Firstly, because of the complexity of the robot manipulator itself structure and external factors, the reality model of the robot manipulator can't be obtained. But the inaccuracy dynamics model of the robot manipulator will affect the control performance of the visual servo system. Aiming at the inaccuracy in robotic manipulator modeling, this study leads diagonal recurrent neural networks in the visual servo control system of robotic manipulator. In order to achieve the robotic manipulator positioning control, this study uses the diagonal recurrent neural networks to identify inaccurate item and compensates the uncertainties of the model in the control law.Secondly, as the approximation ability of diagonal recurrent neural network in system identification is poor, the precision of robot manipulator in visual servo control system remains to be further improved. Echo state network is a new recurrent neural network. This structure of the network is simple, and just need to adjust output weights. The convergence speed of the echo state network is fast, and the learning precision of the network has enhanced. So, aim at the inaccuracy in robotic manipulator modeling, this study leads echo state network in the visual servo control system of robotic manipulator. In order to achieve the robotic manipulator positioning control, this study uses the echo state networks to identify inaccurate item and compensates the uncertainties of the model in the control law.Finally, this paper use MATLAB simulation tools to verify the proposed controller of the visual servo system. The simulation result shows that the visual servo system can reach the desired target, so the proposed controller of the visual servo system is effective.