The Design And Research Of Control Method On Robot Visual Servoing

Visual servo control technology is one of the important research topics in machine vision. The emergence of the technology can provide a reference for solving the problem of robot autonomous action problem in an unknown environment; greatly improve the intelligence, accuracy and flexibility of the robot to complete the task in an unknown environment. Because of its’ advantage of efficient, flexible, etc., the robot with visual servo function wins more and more attention in the scientific research and industrial production.In this paper, our research mainly aims at the visual controller of the visual servoing system to resolve the limitations in the application environment, and put forward relevant control algorithms, including visual controller based on image moment, visual controller based on Homography matrix and visual controller based on switching rules. The main work and results of this paper were as follows:1. Unlike the local geometry features of the image, this paper used the global features which we called image moment as the feature information of the image, and constructed the visual controller based on image moment. The results indicate that, due to using all the image data as the feature information, the ideal control precision could be realized in unknown depth of object and camera focus. This controller can also works well when strong noises exist, so it has strong robustness to noises.2. Aiming at getting the depth information in image Jacobin matrix and overcoming the possible singularity point in visual servoing system, a visual controller was designed based on homography matrix. The controller overcomes the problem of solving the depth information in two-dimensional image data, and enlarges robot’s application area in visual servoing system. It could achieve better convergence effect when external noise is weak.3. A switch controller based on hybrid dynamic systems theory was designed in visual servoing system. The controller consists of the above two sub-controllers (the controller based on image moment and the controller based on homography matrix) and an upper decision controller. The decision controller switches between the two sub-controllers according to the switching rule. The switching system switches between the global and local features of the visual images, overcomes the limitations of each sub-controller on the application environment, and takes full advantages of two controllers to get better effect of servo.4. In order to verify the effect of the above controllers, the paper uses six degrees of freedom PUMA560robot which was provided by robotics toolbox (Robotics Toolbox for Matlab) as instance to do some experiments. The simulation results indicate that the controllers we designed can make the system get better steady-state characteristics and control effect.