Research On Workpiece Positioning And Path Extraction Technology Of Dispensing Machine Based On Three-dimensional Vision

With the development of science and technology,the miniaturization and high concentration of electronic products are becoming more and more obvious.This puts forward higher requirements for dispensing and packaging in the production process of electronic products.At present,the workpiece positioning and dispensing path extraction of automatic dispensing machine are based on two-dimensional vision and assisted by manual teaching.This method is limited by the technology of twodimensional vision and depends on manual experience,which makes the acquisition accuracy of dispensing path not high and the process cumbersome,especially for curved surface and special-shaped workpiece.Therefore,aiming at this problem,this paper introduces the 3D vision with high precision to study the automatic acquisition technology of dispensing path of dispensing machine.The main research contents are as follows.Firstly,aiming at the problems of low positioning accuracy and poor robustness in the workpiece pose recognition of dispensing machine based on 2D,and it is difficult to meet the positioning requirements of workpieces with complex surface types due to the lack of 3D information,a workpiece pose information recognition method based on 3D vision is proposed in this paper.In this method,the pose of the workpiece is recognized in two steps.Firstly,the approximate pose information of the workpiece is recognized by the sampling consistency algorithm based on feature points（sac-ia）point cloud coarse registration algorithm,which provides a good initial pose for the fine registration stage,and then the accurate pose information of the workpiece is recognized by improving the ICP point cloud fine registration algorithm.Finally,experiments show that the algorithm has better pose recognition speed and accuracy.Secondly,aiming at the problems of cumbersome and time-consuming process,highly dependent on labor and difficult to adapt to the dispensing path acquisition of complex curved surface workpiece in the current dispensing path acquisition method based on 2D vision assisted manual teaching,a dispensing path acquisition method based on 3D vision contour offset is proposed.Firstly,the preprocessed 3D point cloud of the workpiece is projected to the xoy plane.Secondly,the edge contour point set is obtained from the projected point cloud through the convex hull algorithm.Then,the edge contour points are offset by a fixed distance along the direction of the normal vector through the offset algorithm.Finally,the offset contour is back projected to the3 D point cloud to obtain the 3D dispensing path.In addition,the dispensing path acquisition method based on 3D vision can also obtain the normal vector on the dispensing path.Using this normal vector to dynamically adjust the direction of the dispensing head can obtain better dispensing effect.Experiments show that this method can correctly extract the dispensing path and has good accuracy.