The Research On The Neural Network Control Strategies For Robotic Visual Servo System

Robotic visual servo control is an hot research topic in the robotics field. Non-linearity, coupling, and uncertainty are key obstacles to improve the performance of robotic visual servo control system. Much attention has been paid on this problem by the researchers all around the world. For solving this problem, neural network control strategies for robotic visual servo system are studied in this paper for improving its intelligence, robustness and dynamic performance.The researches of the robotic visual servo and neural network technology are reviewed. The system structure, coordinate transformation, and imaging principle of robotic visual servo system are described in detail. And several types of neural network controller are presented and applied on robotic visual servo system to conquer its nonlinearity, coupling, and uncertainty.Based on the ability of the error back–propagation network (BP network) to approximate any non-linearity function on any accuracy, a BP network visual servo controller is designed, in which BP network is used to approximate inverse Jacobin matrix of robotic visual servo system.Cerebella model articulation controller (CMAC) neural network has fast learning speed and can be used very well for real time system. Hence, it is introduced into this paper to construct a study control strategy for robotic visual servo system. In this strategy, CMAC is used as feedforward controller to compensate a general feedback controller.Using the good local performance of wavelet transformation, a wavelet network visual servo controller is designed by using wavelet functions as its nerve cells'activations function.For each proposed control strategy, simulation experiments are made. The results of the simulation experiments show that all proposed intelligent neural network models have good intelligence. When they are applied to control robotic manipulators, they are effective for achieving good performance.