| Robot visual servo control is a fundamental research subject with very important theoretical research sense and industry application prospect.Firstly, A method for object recognition in a complex noisy environment based on GA pattern matching is presented in this thesis, combining with distance transformation. Simulations show that this method can quickly and accurately recognize the object in input image and has a few strong robustness against noise in the input image.Secondly, The simulations for two kinds of control algorithms based on PD, namely, direct PD control and PD control with feedforward compensation, have been completed in this thesis. Simulations show that both control algorithms have good control performance in tracking a known trajectory and the effect of dynamic compensation of PD control with feedforward compensation is obvious. But direct PD control has better robustness against modeling error and uncertain disturbance compared with PD control with feedforward compensation.Finally, the forward kinematics model of MOTOMAN robot is set up and its inverse kinematics solution is obtained by using algebraic method and iterative method respectively; A MOTOMAN industry robot, a CCD camera and an image grabber card, along with PChost computer, formed eye-in-hand coordinated visual servo control system, which constructed a hardware platform of the theoretical research and simulation experiment for robot real-time visual servo control. The 3D vision locating of object is done using "hand-eye" stereovision locating algorithms based on this experiment system. Experiment results show that the algorithm is effective as well as settlement. Some measures for improving the locating precision are presented also. A robot vision locating experiment is completed based on dynamic position-based look-and-move visual servo control structure and experiment results show that the robot vision system has high locating precision.
|
PDF Full Text Request