Research On Binocular Vision System For Industrial Robots

In this paper,the binocular vision sensing system of industrial robots is studied,in order to solve the problems that most industrial robots have a single target for identification and positioning,and many types of work pieces,the recognition and positioning and grabbing are not well realized,this paper uses binocular stereo vision as the visual servo system to identify and locate the circular,hexagonal and square work pieces respectively,and control the industrial robot to grab work pieces successfully.(1)By studying the imaging model of the monocular camera,Zhang Zhengyou camera calibration method is used in the article,the binocular camera calibration and hand-eye calibration of the industrial robot binocular vision system were carried out by using HALCON machine vision software,and the calibration results were verified by experiments to meet the accuracy requirements.(2)Realize research on object recognition and localization algorithms with HALCON machine vision software.Firstly,the image is pre-processed to improve the quality of the image useful information,then Gaussian filtering is performed on the image,and finally the Canny operator is used to extract the edge contour features of the target workpiece to achieve more accurate positioning.By analyzing the principle of three basic template matching methods,propose an improved template matching method that uses the gradientbased shape template matching method to identify and match the target work piece,and the positioning and grasping of the work piece is completed.The method can be applied to the matching and identification of multiple target objects,and the calculation speed is fast and flexible.(3)The experimental platform of the industrial robot binocular vision sensing system is built.Combined with the results of the previous binocular camera calibration and handeye calibration,the feasibility of the improved template matching method in the actual capture process is verified.In the experiment,the industrial robot can successfully capture the target work piece,and the data analysis of the experimental results proves that the template matching algorithm in this paper has high positioning accuracy.