Study On Robot Sorting System Base On Visual Servoing

In the context of rapid development of industrial automation,more and more robotics are being used in industrial operations,this includes robotic sorting techniquees.Traditional sorting techniques often use sports teaching methods,the location requirements for the target must be fixed,and has a low efficiency.Combine machine vision with robotic arm,so that the robot can independently identify the target object,increased operating speed,reduced labor costs,has a significant practical significance.Robot vision technology allows the robot to have a visual perception.Now is the focus of today's industrial automation research,is currently widely used in industry for defect detection,logistics,palletizing,welding and many other areas.This article by reference to the relevant literature,based on the current research status,taking puma560 manipulator as the research object,Research on location based visual servo control method,designed a visual sorting system,And the target workpiece image processing,the target workpiece matching recognition,coordinate system transformation,design of the visual controller to complete the servo control as the main line throughout the text,and finally realized the robot arm for the target of the static positioning and movement tracking,the main contents of this paper are as follows:The design and research of the target image processing algorithm are carried out,the best image processing scheme was selected by experiment comparison,the noise of the target binary image is removed,get a satisfactory target binary map.Otsu algorithm is one of the most commonly used methods in image segmentation,For the traditional Otsu algorithm to traverse each gray value of the interclass variance calculation,a fast Otsu improved algorithm is proposed,in the introduction of the image complexity and its related properties to reduce the gray scale of the search range,at the same time in the search range using a fast calculation method,which is secondary to the traditional Otsu algorithm,the experimental results prove,the algorithm improves the computational speed compared with the traditional Otsu algorithm,and the image segmentation results of the two algorithms are the same.Harris Corner Feature Extraction for Sorting Targets,the corners of the template and the original image are obtained,to the template and the original image of the corner feature image,and uses the normalized cross correlation(NCC)strategy to roughly match the original and the template,get the matching result that contains the mismatch,eventually by using the RANSAC policy to remove the mismatch,get the correct match result after purification.The calibration of the visual system was carried out,The theoretical and calibration experiments were carried out fo r camera calibration and hand-eye calibration,the camera's internal and external parameters are calculated by ensuring sufficient calibration accuracy,and gives the general method of hand-eye matrix solution.For puma560 for the kinematics of the solution,the control method of positionbased visual servo is studied,In the simulink under the visual servo control model structures,and the simulation of the stationary target location and the tracking of the moving target are realized respectively.Through the above research work,the visual system proposed in this paper has high accuracy for the target workpiece recognition and good tracking effect for the linear motion target,which proves that the system has some practical significance.