Research On Industrial Robot Sorting System Based On Machine Vision

Workpiece sorting is an important link of industrial production, which is completed manually on the traditional production line, with low efficiency, poor product quality and health risks to workers. Consequently, it is necessary to realize the automation of workpiece sorting. As a result, the industrial robot, as a typical representative of mechanical automation, has been widely used in industrial production due to its labor saving, high efficiency and improvement of products’ market competitiveness.For the purpose of increasing the adaptability, improving the efficiency and accuracy of sorting system, an industrial robot workpiece sorting system based on machine vision was presented in this paper, including modeling and calibration of vision system, image distortion correction, image pre-processing, image recognition and path optimization, etc. Firstly, a non-linear model based on radial distortion of the visual system was established after analyzing its linear model, according which a calibration method of the visual system based on two dimensional targets was designed to establish the corresponding relationship between image pixels and three-dimensional world scene points. And the image distortion was corrected through bilinear interpolation method to increase the recognition accuracy. Then the workpiece image was preprocessed, including gray-scale transformation, binary transformation, morphological and multi-objective blocked processing to eliminate the irrelevant information and retain the effective information. In order to obtain the shape, pose and color information of circular, triangle and star shaped workpiece, a method of equivalent edge curve was used to recognize the shape of workpieces, then geometric centers of workpieces were obtained by fast calculation method of multi-objective centroid, and then corners of workpieces were detected by Harris corner detection algorithm and corners on the external rectangle were screened to calculate the rotation angle of workpieces. Robot sorting path was optimized by branch and bound algorithm based on the principle of drawer to achieve the fastest sorting. Finally, the experimental platform of sorting system based on machine vision and three-axis parallel Tripod robot was established and a control interface of automatic sorting system was designed to improve operability and usability of the system.It is concluded from the Experimental results that the industrial robot workpiece sorting system designed in this paper can realize the auto-sorting of circular, triangular and star shaped workpiece effectively and accurately, improving the adaptability and efficiency of sorting system. Therefore the auto-sorting system may be applied in industrial production.