Design Of Workpiece Positioning System For Automatic Painting Robot Based On Stereo Vision

Presently in China,the spraying work is mostly done manually by workers,which leads to the consequence that the quality is influenced by the technical level of the workers.It not only leads to high labor costs,but also the spraying operation environment is very harmful for the workers.Moreover,many kinds of paint have a great impact on people's health,and long-term work will easily lead to various diseases even poisoning.Therefore,the automation of spraying operation is very important.It is difficult to say that the current automatic painting robot has been achieved,most of which are dependent on manual teaching,and then fix the workpiece through some device to do spraying work.However,it is difficult for a robot to apply the spraying to various workpieces.Therefore,a workpieces positioning system for spraying robot based on stereo vision with the low cost Kinect2 sensor is designed and implemented in this paper.Such system can make a good positioning for various workpieces,then the trajectory teached by workers can be used,which not only enhances the efficiency,but also improves the application scope of the painting robot.Three aspects of work have been done and listed as follows:(1)Research on 3D reconstruction method of workpiece surface based on Kinect2.First the measurement principle and error analysis of Kinect2 is introduced;according to the experiment of Kinect Fusion 3D reconstruction algorithm,we found that it is not applicable in this project.Therefore,a reconstruction method with fixed multi Kinect2 is designed,and introduced.(2)Research on localization method of point cloud based on hand eye calibration.The mathematical principle of Eye-to-Hand system calibration is analysised,considering the measuring noise sensitive defects of traditional hand eye calibration method,this paper used a calibration method based on Li Group that can calculate a least squares solution of more data in different positions.Such method can increase the accuracy of hand eye calibration process.Then this method is verified experimentally and the positioning accuracy of the point cloud is analyzed.(3)Research on paint trajectory matching method based on point cloud registration.After getting a point cloud and a corresponding trajectory in the robot coordinate system,a method of trajectory matching based on point cloud registration is designed and implemented.The traditional rigid registration method of point cloud is introduced,then a non-rigid point cloud registratioin method which is based on the deformation description and nonlinear least squares optimization is designed and implemented,the method is then analyzed and verified by the experimental results.