Research On Delta Robot Workpiece Sorting System Based On Machine Vision

Workpiece sorting is an important part of industrial manufacturing.In the traditional sorting method,manual sorting is greatly affected by the on-site environment and subjective factors.Long-term work will cause visual fatigue and result in lower industrial sorting efficiency.Delta robot has the advantages of fast speed,small motion inertia and high positioning accuracy.Combining machine vision with Delta robot to realize automatic sorting of industrials is the trend of intelligent industrial pipeline.The use of machine vision algorithms to detect industrial defects and achieve fast sorting operations has important theoretical and practical significance.Aiming at the problem of poor robustness of workpiece contour damage detection and the difficulty of surface defect segmentation affected by workpiece texture,the flange workpiece is used as sorting object to carry out theoretical research and experimental verification.Firstly,the overall design of the sorting system is introduced in detail.According to the workpiece size and expected accuracy,the hardware selection of the visual system is completed.By establishing a monocular vision measurement model,the transformation relationship between the workpiece world coordinate system,the camera coordinate system,the image physical coordinate system and the image pixel coordinate system is analyzed.The calibration of the internal and external parameters of the camera is completed by the high precision point calibration plate calibration plate,and the measurement and error analysis of the dimensions of each part of the industrial are realized.Then,aiming at the problem that the shape and size of the damaged area of the workpiece contour affect its segmentation and recognition,a method of shape def'ect detection based on edge distance of workpiece is proposed.The sub-pixel edge information of workpiece contour is extracted,and the distance between fitting edge and workpiece contour is calculated.By comparing whether the distance is larger than the given threshold,the damage of workpiece contour is judged.The detection experiments of workpiece with different degree of damage are carried out.The recognition accuracy of workpiece with broken contour reaches 100/%,which verifies the robustness of the algorithm.Aiming at the problem that the complex texture of workpiece surface affects the segmentation of scratches and rusts on workpiece surface,a PixelNet convolution neural network based on pixel stratified sampling is used to segment the surface defects.The experimental results show that the proposed method can effectively segment the workpiece surface defects.The mean intersection over union(MloU)of the segmentation results reaches 92.3%.Finally,the calibration of the vision system and the motion system is completed to determine the optimal gripping position.The motion system completes the sorting experiment of the industrial according to the industrial category and the gripping position information provided by the vision system,and achieve the sorting of scratches,rust and damaged workpiece.