Research On Intelligent Robot Visual Servo System

Intelligent robot visual servo is the fusion of results from many elemental areas including real-time image processing, robot kinematics, robot dynamics, control theory, computer technology and real-time computation, and is a main subject in the research field of computer vision.Several aspects of intelligent robot visual servo system are concerned with in this thesis, including camera calibration method, image recognition and image-based robot visual servo system.Firstly the camera model and camera imaging principle are introduced and the traditional camera calibration methods are studied. Further, a novel camera calibration method based on RBF (Radial Basis Function) neural network is proposed. This method improves the accuracy and the robustness in comparison with the traditional calibration methods. Moreover, it doesn't need some physical parameters of camera in the calibration, so it has more usability.The image features of object and ambiance in the real robot visual servo system are analyzed. The method of object recognition based on image segmentation and cornerdetection is presented. An image is segmented into several regions by image segmentation. Only one object exists in each region. The regions in which the object has the same corner quantity with that of pattern can be obtained by corner detection. An pattern matching method based on genetic algorithm with fuzzy adaptive genetic operator is applied to seek the optimal matching. The experiments indicate that this method can reduce the seeking range and accelerate the seeking velocity obviously.Image-based robot visual servo control scheme is reviewed. According to Motoman-SV3 robot, a robot simulation model is constructed. The simulation experiments of plane position and plane tracking based on the model prove the validity of this scheme.