Research And Design Of Robot Sorting System Based On Machine Vision

With the continuous development of industrial robot technology,more and more fields have begun to use robots instead of workers to complete production operations.Sorting robots are one of many industrial robots,which can replace workers to sort the materials on the production line.The traditional sorting robot completes the sorting work by means of teaching and programming.Although the sorting operation is automated,the workpiece size and placement position are often required.Once the position is deviated,the position is deviated,The workpiece cannot be accurately captured.Combining machine vision technology with industrial robots,and acquiring the position information of the workpiece through the vision system to guide the robot to grasp the workpiece,not only can solve the problems of low flexibility and weak anti-interference ability of the traditional robot sorting system,but also accelerate the pace of industrial production intelligence,which has great significance.This topic is aimed at the problem that the industrial robot sorting system has high requirements on the position of the workpiece,and the robot automatic sorting system based on machine vision is designed.This topic designs a machine vision based robot sorting system for the problem that the industrial robot sorting system requires high position of the workpiece.Aiming at the characteristic of the vision-based robot sorting system with long working distance and large field of view,Nachi's MZ04 light six-axis robot,Basler's Aca-1300-gm industrial camera and Fuji's HF16SA-1 lens are selected,the robotic sorting platform based on machine vision was builded.Aiming at the problems of large amount of calculation and many operation steps in the traditional hand-eye calibration algorithm,a fast hand-eye calibration algorithm is proposed.Three Mark points are used for calibration of the sorting system,and an affine transformation matrix between the robot coordinate system and the camera coordinate system is established.The algorithm has fast calculation speed,convenient operation and wide application range.Aiming at the problem that the acquired image is not high in contrast and difficult to identify the workpiece,the image of the workpiece is extracted by using gray stretch,threshold-based image segmentation and morphological closure algorithm,and the Canny algorithm is used to detect the edge of the image.The edge-based template matching method completes the classification of different workpieces,extracts the coordinates of the workpiece centroid,and achieves the accurate positioning of the workpiece.Aiming at the problem of Software development problem for robot sorting system the host computer program was developed by C# and Halcon,the functions of system calibration,workpiece classification and positioning are realized.The robot communication program is developed by robot language,the communication between the robot and the host computer is established,and data transmission and reception was realized.The robot motion program is written in robot language.The motion control function of the robot was realized,and the three methods is combined in a layered communication manner to form a robot vision sorting system software.Finally,in the working range of the machine vision-based robot sorting platform,a variety of workpieces are placed arbitrarily,and the machine vision-based robot sorting system software is started to complete the vision-based multi-objective automatic sorting experiment.the laser tracker was used to measure the positioning error of the system,the experimental results show that the sorting system can sort the workpieces of different shapes placed at any position within the working range,with small system error and good sorting effect.