Research Of Sorting System On Delta Robot Based On Machine Vision

With the advent of intelligent manufacturing,industrial robots are widely used in various types of automated production lines.The parallel robot has many advantages,such as compact structure,fast response speed and so on.Therefore,it has been extensively applied in the field of high precision and speed or large workspace.Combined with machine vision,a kind of Delta robot sorting system for automatic recognition and dynamic capture of online objects is proposed to improve food packing efficiency.Principles for selecting machine vision system components are also determined according to system operating environment.To achieve target recognition in sorting system,firstly,the paper calibrates the internal and external parameters of camera.Secondly,the improved genetic algorithm combined with Ostu image segmentation method is used to find the optimal threshold.Then,a target recognition and location method based on SIFT algorithm is proposed to determine the target centroid location by affine transformation.Finally,with the assistance of shape-based template matching method,the paper realizes an universal image recognition algorithm.In order to realize dynamic crawling,the robot control system is researched from three aspects.The synchronous tracking strategy between robot and conveyor belt is studied which is verified through the development of robot control system.For the occurrence of same product in continuous images,the paper gives a judgment method for multiple repeating targets.A trajectory planning method based on S curve is presented to overcome large shock impact and less beat numbers in sorting process.According to the actual requirements of sorting system,a visual software based on HALCON algorithm library is developed to provide location guidance for the robot.The paper also summarizes the human-computer interaction demands and designs a HMI teaching software with QT,in order to debug the sorting system conveniently.The experimental results show that the sorting system can effectively identify the type of geometric product,and accurately complete sorting action,which greatly improves the sorting efficiency,and has high application value.