Research On Visual Servo Control System Of Robot PUMA560

As a robot control technology, visual servo has been widely accepted. The first visual servo system can be traced back to the late 1970s and early 1980s. With the image processing technology and the related hardware rapidly developing, research papers in the direction have been increasing. From the mid-1990s to the present, as visual servo relates to many elemental areas, so it is a challenging research topic. This paper focuses on the classification of visual servo system, modeling, designing controller. With robot visual servo system studied in depth, the classification of visual servo system in accordance with the different standards is various. The systems with different structural have their own advantages and disadvantages. The thesis consulted a lot of reference materials and summarized the current classifications and their characteristics of visual servo system. To clarify the characteristics of different types system, it is very important for further research.The features of 3D objects in the vision system can be mapped to the two-dimensional image planes, We then use the image information to control the robot. It is needed to relate the changes of feature vector in image plane with the change of robot's posture. This relation is called image Jacobian matrix, which can be used to control robot to the desired position. In view of the three main control structure include of position-based, image-based and 2^-|2D, we use the method of the task function and select the task variables, then drived the Jacobian matrix detailedly. After knowing the relationship between the image information and space configuration, two kind of controller is designed.Accurate calibration to the camera is a relatively complicated process. Scholarshave used many ways to avoid this process and most of them used the way that is based on-line estimate method to determine the unknown parameters. The thesis designed an adaptive controller for PUMA560 robot aimed at the structure of eye-in-hand robots with cameras, estimated the unknown parameters of tasks Jacobian matrix and the unknown parameters camera and manipulator relative transform matrix online, a stability analysis is included. Simulation results can be seen that such a control method that can avoid calibrating and enable camera movement from the initial pose to the desired pose.Epipolar geometry discussion is the relationship between two camera image planes. Epipolar geometry is applied to the visual servo control in the thesis. And finding the relation of the images obtained from the current camera position and orientation and the images of hope. Then the relation is used to control robot. From the simulation results, this control method can make the error vector convergence to zero of images, the robot can reach the desired position and orientation.