An Application Of Machine Vision On Sorting Clutter Workpiece By Robot

With the rapid development of industrial automation, robot sorting technology is widelyused in various production lines. The traditional robot sorting is mostly used in method ofteaching. This method has a series of very strict requirements on the gesture of the initial ofthe robot and the termination. Because of this, it brings amounts of problems. That is not onlythe low speed of sorting efficiency is low, but also results in the failure of robot grasping oncethe position of the workpiece changes which will surely influences the production efficiency.Therefore, people turn to pay their attention to the field of machine vision. The visiontechnology combined with robot and enabled the robot to have the recognition function whichis very similar to the function of human vision. So that it makes the robot sorting more flexible,high accuracy, intelligent and then leads to the reduction of the cost of the labor. It also has avery significant meaning to ensure the quality of the product as well as the improvement of theproduction efficiency. Nowadays, the technology of machine vision has been widely used innondestructive testing, food packaging, pharmaceutical production, logistics sorting, PCBmopping and many other fields. Then the robot technology which based on the machine visionwill become the main trend of the development in the future.The thesis puts forward a kind of workpiece visual recognition algorithm based ontemplate matching which concluded from the production line workpiece sorting out cluttertarget workpiece pile. In addition, the thesis taking RH6-A SIASUN6axis industrial robot asthe foundation, built the sorting system of industrial robot based on machine vision, hasformulated the detailed visual sorting scheme. This thesis firstly analyses the kinematics ofindustrial robot, the application of D-H coordinate transformation method to calculate thekinematics forward solution and inverse transformation method to calculate the inversesolution, and developed robot simulation system based on OpenGL according to the algorithm.Secondly, it analyzed the camera model OpenCV vision algorithms library provides.Furthermore it establishes the robot hand-eye relationship, solves the camera intrinsic, externalparameters and transforms matrix under the use of Zhang Zhengyou calibration method forcamera calibration. And then, collected the image of workpiece by the industrial camera, madethe image prepossessing by smoothing and denoising in order to eliminate the noise, as well as detected the edge of the workpiece using Sobel edge operator. So that the workpiece edgeinformation is obtained. After that, it gets the image binarization. Finally it makes thesimilarity measure of the template image and the image to be matched, and completes thetemplate matching.The conclusion of the experiment shows that the method successfully identified a targetworkpiece among the workpiece pile based the edge template matching algorithm. After that,it has calculated the coordinates of the center of the workpiece using the central moment of theimage. Then it has given back the key feature information to the robot controller so as toachieve the purpose of the robot sorting.