Research On Uncalibrated Visual Servo System Of Microassembly Robot System

Robot applications have broad prospects for development in the field of micro, micro robot is of small scale assembly parts for assembly robot. Micro assembly robot system platform is presented in this paper, by the double light path microscope provides visual image signal, three manipulator and gripper performs the specific operation, through uncalibrated visual servoing method, by the image signal to extract the characteristics of the robot motion control.The manipulator end gripper and parts for target recognition, this paper adopted the image recognition and the ORB feature points based on contour matching two ways. Based on contour recognition method is applied to the target during the initial positioning and mobile positioning, first of all to grayscale image, filtering, edge detection, morphological processing and contour detection, and the external contour is rectangle, then according to the outline and the external is the area of the rectangle and vertex coordinate information for identification. When the target chamber and grooves for precise positioning, using feature point matching algorithm, in the comparison of the ORB, SIFT and SURF the principle of three kinds of commonly used algorithms and performance, this paper chose the fastest ORB algorithm. Experiments prove that this article USES two algorithms can satisfy the requirements of real-time and accuracy.Using uncalibrated visual servo control of micro assembly system, this paper sets up the double closed loop control structure. Control system adopts the method of Kalman filter to solve the image jacobian matrix, the matrix of real-time identification. Through the manipulator trajectory tracking test experiment, the algorithm in this paper is feasible.