Research And Design Of Vision Guide System For Casting Polishing Robot

With the industrial upgrading and upgrading of the manufacturing industry and the improvement of the level of industrial automation,more and more enterprises are undergoing transformation and upgrading from low-end manufacturing to high-end intelligent manufacturing.The identification and positioning technology of the surface defects of the castings to be polished is an urgent problem that needs to be solved in the current automatic grinding station system for many varieties and small batches of castings.More and more companies are beginning to introduce machine vision technology to improve the automation of the robot grinding system.The topic of the thesis comes from the actual project of the existing casting grinding production line.The existing casting grinding station uses robot teaching method to guide the robot to polish the casting defects.In order to further improve the level of automation and intelligence of the casting grinding system,the author designed a set of machine visionbased automatic identification of casting defects and machine vision guided grinding system.The main work content is as follows:(1)In view of the interference of dust and vibration and other unfavorable factors in the casting cleaning and grinding station,this article performs gray-scale processing,threshold processing,filtering processing,morphological feature processing and other image pre-processing operations on the collected casting partial images to facilitate subsequent image processing The algorithm extracts the image morphological feature information of the defective area.(2)In order to reduce the wear rate of the grinding head at the end of the manipulator,the grinding system requires the robot to have certain machine learning capabilities to perceive the type of flaw to be polished,and then select the targeted grinding head for cleaning.The machine vision system can collect the digital image of the area to be polished,and then extract the morphological feature information of the defect area in the batch of sample images.Establish a machine learning model to classify and identify the feature information of the defect area,and select the corresponding type of grinding head in the process requirements for the identified defect type to clean and polish.(3)Use Daheng Imaging Company's MER?201?25GM industrial digital camera and supporting light source to build a digital image acquisition system,collect different types of defective digital images in the casting grinding station,and actually implement the above theoretical assumptions to verify the identification of this article based on the imaging characteristics of defective areas Classification method.(4)Based on the camera calibration,obtain the data of the internal and external parameters of the camera,install the LMI Gocator 2140 line laser camera at the end of the six-axis manipulator to complete the hand-eye calibration,and obtain the spatial conversion relationship between the camera coordinate system and the manipulator end coordinate system.On the basis of hand-eye calibration results,verify the accuracy of the vision guidance system.The specific scheme is to select the discretely distributed circular area in the calibration plate to simulate the defect area on the casting surface.Then,calculate the coordinate values of ten points of the center of the circle in the image coordinate system of the digital image of the calibration board and convert them to the physical coordinates of the center of the circle in the robot base coordinate system.The difference between the physical coordinate value of the center space calculated above and the physical coordinate value of the actual space center is obtained,and the error between the two is used as the basis for verifying the accuracy of visual guidance.