Point Sets Optimization And Positioning Research Of Automatic Spraying System

As key equipment of advanced manufacturing,industrial robot has been promoted to a national strategy level by many highly developed industrial countries in the world. Our country has also made it a major project in “13th Five-Year” development plan. Spraying robot,as one of its kind,has many merits in working process, such as high painting quality,better efficiency,environment-friendly and safety, but at present what spraying robot can achieve is repeatedly perform the trajectory that has been manually planed before, which causes long adjustment procedure,low flexibility and economic effectiveness.In this dissertation, we propose a new spraying system to improve the intelligence level of spraying robot, avoiding manual teaching process. The system acquires 3D point sets model of spraying object by 3D scanner which provides the basic data for trajectory planning algorithm to calculate execution route in point sets coordinate automatically and combining vision matching module to translate the execute route to the robot coordinate, realizing automatic spraying process. The system vests robot the ability to recognize spraying object, plan spraying trajectory and finally execute spraying process on its own.We carry out the following research:(1) Denoising 3D point sets model of spraying object and normal optimization. Weobtain 3D point sets model of spraying object by scanner, utilizing PCA approach to calculate initial normal of the model and fairing it by bilateral fairing algorithm. Based on the normal difference, we establish point sets neighborhood where we use the distance between noise points and the weighed geometry center of the point sets neighborhood as an noise index to wipe out those noise points.(2) Camera calibration and calibration process optimization. We adopt Matlabcalibration tool to perform on-plane camera calibration, using priori knowledge of the calibration process to improve the size selection of corner detector window, enhancing the detecting accuracy under different image scale space, so that the purpose of simplifying calibration process and reducing manual error is achieved.(3) Image and point sets matching research. Using the method in image matching for reference to match the point sets edge and image edge, we project point sets to image by grids, proposed image curvature chain matching method. The camera shoot the image of spraying object and unfold the image edge to match with the projection of point sets, achieving the trajectory translation from point sets coordinate to robot coordinate.