Research Of Scattered Parts Recognition And Grab System Based On Line Laser Camera

The positioning and grasping of scattered parts are widely used in automation industries such as warehousing and logistics,automobile assembly and electronic patch.With the transformation and upgrading of Chinese industrial production automation,a large number of robots have replaced manual labor to complete monotonous,heavy grasping and placing work.Traditional robotic grasping technology needs to determine the motion trajectory in advance,and cooperate with mechanical accurate positioning technology to complete the parts grasping.In this method,the position and attitude of the parts are fixed,and the flexibility of production is poor.In order to overcome the shortcomings of traditional methods in the application of automated production,this paper studies the key issues of 3D camera-based part recognition and grabbing systems.Including the research and design of parts point cloud acquisition systems,the research of line laser fringe center extraction methods,parts point cloud pre-processing research,parts positioning method and robot grabbing are several parts.The main research contents are as follows:(1)The existing 3D imaging technology and the accuracy requirements for capturing point clouds of parts is analyzed.And a 3D point cloud acquisition system based on line laser cameras is designed and developed.This system can adjust the working range of the camera by adjusting the position and angle of the camera and the line laser generator,and adjust the exposure of the camera and the brightness of the line laser generator according to environmental changes.The Zhang Zhengyou's camera calibration method commonly used in the monocular camera calibration method is proposed for line laser camera calibration,which overcomes the shortcomings of traditional calibration method and active calibration method that are difficult to meet the accuracy and speed requirements at the same time,and quickly realizes the line laser camera calibration.The line laser camera designed in this paper has a measuring range of 700-1000mm and a measuring field of view of 400-600mm.The measurement accuracy is within 0.5mm.(2)The laser stripe center extraction method of line laser cameras is studied to solve the problems of poor robustness and time-consuming of the existing line laser stripe center extraction methods.The laser fringe image is pre-processed according to the image characteristics and noise model of the collected line laser fringe image.The improved Steger algorithm is used to ensure the accuracy of the line laser fringe extraction center extraction while the single image processing time is less than 50ms.(3)Investigate the pre-processing method of point cloud acquisition by line laser camera.The statistical filtering method is used to reduce the influence of noise and environmental points in the part point cloud on the recognition and positioning effect.The downsampling process is used to reduce the part point cloud density.Experiments show that preserving the point cloud of the part to be identified through the point cloud preprocessing operation can effectively reduce the impact of the external point cloud and improve the part recognition rate and speed.(4)Investigate the method of part point cloud identification and localization based on two-step registration method.Aiming at the problem that the iterative closest point(ICP)algorithm has a high error matching rate when the initial pose of the point cloud is not ideal,the sampling consistency matching(SAC-IA)algorithm is used to initially match the part point cloud,and the parts are coarsely located Cloud location is optimized.Set up an experimental platform for robotic grab experiments on scattered parts.The two-step registration method is used to locate parts.The success rate of robotic grabbing experiments is 66%.