Application Of The Computer Vision On The Industrial Robot System

After more than 40 years of development , since its first appearance till now, the robot has already been widely applied in every industrial fields, and it has become the important standard of industry modernization. However, most industrial robots that actually applied in factories work in the teaching-recurrence mode at present. Because the working mode is open-circuit, in which the robot know nothing about the ambience, such as the work pieces' displacement caused by the work pieces' deflection and distortion, or the barriers appearing in the robot's locus, so when working in this mode, the industrial robot can not adjust its moving locus, and lacks flexibility and adaptability. So we introduce the computer vision system into the industrial robot system, on basis of the primary Motoman UP6 industrial robot system, we exploit a set of computer vision system for it. The computer vision system is composed of a camera, an image-sampling card, an industry computer and the primary Motoman UP6 industrial robot system and relevant software. By this set of computer vision system we take images information of the work pieces and the ambience in the camera visual field, analyze these images and carry on a sequence of images pressing, achieve three-dimensional coordinates of the spatial points, by using the remote controlling software Motocom32, we can get the control over the robot through the industrial computer, and realize automatic positioning and tracking at last-In this dissertation we explain relevant knowledge of industrial robot at first, including Motoman UP6 industrial robot relevant knowledge, the positive and reverse problem about industrial robot kinematics, and by using the kinematics theory of the industrial robot, we get communication with the robot's controlling box through Motocom32 software, and upload this robot program, analyze the program, do planning and recombination to the robot program, and download it to the robot's controlling box, realize realized real-time planning of the robot's locus. Afterwards we expound the whole industrial robot software and hardware and the problem about the camera calibration. On basis of summarizing traditional methods of camera calibration, combined with the actual conditions, we put forward "a quick and augmented method of the camera calibration" . In this method we considerately consider the non-linear distortion of the scene and system error in the camera model, and introduce the non-linear distortion of the lens and system error into the direct-linear calibration algorithm, and quickly realize the camera calibration through a simple linear iterative method. And then we explain in detail the present ripe image -processing algorithm, on this basis, combined with the background of this system, we have proposed a practical images processing method, including images pretreatment, imagessegmentation and edge-measuring, the extraction of images' feature and the calculation of features, and can figure out the images' features information effectively. Finally we introduce three-dimensional imaging system, three-dimensional reconstruction method and model that we have adopted, and the problem about image matching in the three-dimensional vision. According to the images feature information that we have figured above, we match the a series of images that we take from difference view points, and utilize three-dimension reconstruction technique, combined with the relationship established by the camera calibration, we can create some equations about coordinates of the spatial points and their corresponding points projected on the two images and calculate out the three-dimensional coordinates of the spatial points. And then by using the robot remote controlling software Motocom32, we can control the robot and let it move to the spatial points and realize automatic positioning and tracking.