Research On The Robot Arm Sorting System Based On Machine Vision

With the concept of "industry 4", the fourth industrial revolution, which is the goal of intelligent manufacturing, has swept the globe. In the automatic production system based on industrial robot, the intelligent degree of the production system has become a research hotspot in the present research. In this paper, the sorting system of the work-piece in modern industry is studied, and the disadvantages of the traditional sorting system are studied. The development direction and research focus of the current automatic sorting system are analyzed. What's more, the application of machine vision in sorting system is the key research object, and the visual servo, image processing, robot arm kinematics and other key technologies need to be further study. At last, based on the laboratory teaching manipulator SCORBOT-Er 4u, a machine vision based robotic arm sorting system is proposed.First of all, learning the application of machine vision technology development at present, and summarize the technology. On the basis of the above, the visual servo control structure of this paper is established, which is based on the position of the servo control system. Then, study the camera calibration technology principle, compare and analysis the different camera calibration methods, and use the Zhang Zhengyou calibration method to calibrate the sorting system.Image processing technology is the core of machine vision technology. In this paper, we study the image preprocessing method, and compare the Gauss filter, mean filter, median filter and adaptive median filter. On the basis of studying the classical image edge extraction operator, improve the Canny operator and apply it to the work-piece edge extraction; Analysis of threshold segmentation and clustering segmentation, and proposes a fast FCM clustering segmentation method based on improved particle swarm optimization algorithm. Studying the algorithms of image feature extraction and recognition, and using the geometric features of graphics, SUSAN corner detection algorithm and Hough algorithm to complete the work-piece identification, classification, at the same time, obtain the work-piece related pose parameters.Learning robot arm kinematics knowledge and using D-H modeling method to model SCORBOT-Er 4u manipulator arm. Using Open GL image database to build it's simulation, and then carrying out motion simulation. Finally, using the laboratory resources, to build a visual robot arm sorting system, then experiment of the sorting system, which is for sorting work-piece models, is carried out.After many experiments and debugging, in this paper, a primary mechanical arm sorting system based on machine vision is completed. The feasibility of the sorting system is proved by experiment. This provides a useful reference for the design of the robot arm picking system based on vision in the actual production.