Research On Visual Servo System Of Intelligent Controller

Visual servo control is a frontier field and also an important research direction in robotics. The research on robot visual servo has great significance for improving the robot intelligence level. Expanding the application fields of robots and even for the development of robotics subject.The robot visual servo control combines many research areas, such as robotics, automatic control theory, computer vision and computer science.we present a new kind of visual controller based on CMAC. It can optimize the nonlinear relationship of image jacobian matrix(IJM), compensate for uncertainties of mapping relationships which is due to calibration error. The main work content and research results are as follows:1.The dissertation introduced the classification and the key research of visual servo system, then explore the research methods and the development status of intelligent control. Showed the mathematical model of visual servo system. Build the jacobian matrix. established visual servo system model by the information above by matlab/Simulink.2. To the problem of IJM estimation, the dissertation uses Celebellar Model Articulation controller(CMAC) as the visual controller directly. Presented an optimization CMAC learning rules which has a good ability to change learning rate in running, and proved the convergence of the rule. The simulation result shows that the improved CMAC intelligent control algorithm can effectively approximate the relationship of jacobian mapping. Then compared the performance of improved CMAC visual controller and traditional CMAC visual controller, shows that the improved CMAC visual controller learning faster.3. To the disadvantage of PID and CMAC, the dissertation adopted CMAC feedforward control structure, combined it with PID controller, Presented CMAC+PID composite visual controller.finally, We did a static targeting simulation analysis by the model built above. It proved that the proposed control algorithm can quickly make the tracking error tends to zero, and the created IBVS has a good ability in visual servo system control.