| In recent years, the application of machine vision in industry has become a hot research direction. The vision system can detect the information of workpieces in real time. Then the intelligent level of the robot can be improved by using of the artificial intelligence theory. So the real-time visual measurement and control system largely expands the scope of robot applications.By reading a large number of literatures, the domestic and foreign research situation of visual servo system is analyzed. A moving workpiece sorting system with position-based visual servo is designed in this thesis. The main work includes the research of image segmentation, target recognition, vision system calibration, construction of the sorting system, programming of the PC software and so on. The main work and achievements are as follows:Firstly, target extraction methods based on threshold and background difference are introduced. These target extraction methods are compared, and the experimental results are given.Secondly, a target recognition method based on the minimum mean square delta rule neural networks is proposed, and the effectiveness of the algorithm is verified through experiments.Thirdly, vision system calibration algorithms are researched, including Faugeras camera calibration method, monocular vision system calibration and camera calibration based on Matlab Calibration Toolbox, and related experiments are conducted.Fourthly, how to use the Kalman filter algorithm to predict the speed unknown motion target is researched, and predicting the varying accelerated motion target or obscured target is also be researched, the experimental results are given.Fifthly, robot servo control methods of the sorting system are researched, the trajectory planning method with online trajectory update is proposed, and the simulation and experimental results are given.Finally, the work of this thesis is summarized, and some contents need to be improved are put forward. |
PDF Full Text Request