Weld Seam Recognition And Grinding Tool Orientation Optimization Based On Point Cloud

With the rapid development of industry,a new industrial upgrading is going on in the manufacturing field of China.The conventional manual teaching accounts for majority of obtaining machining location and orientation in robot machining now,which is tedious.It is an alternative and effective way to obtain complicated machining location and orientation by using visual system.Thus,this paper solves the weld seam recognition and grinding tool orientation optimization in the manufacturing field based on point cloud measured by robot vision system directly.The task of the weld seam recognition is to identify the weld seams of the workpieces from point cloud,and the goal of grinding tool orientation optimization is to realize the smooth transition between adjacent tool orientations by planning and optimizing tool orientations for grinding wheel hubs of automobile.In the process of robot weld seam recognition,a simple and easy way is used to obtain 3D data information of workpieces by top scanning,which can preserve the information of weld seams to the utmost.The processing algorithm includes two sub-steps,i.e.,preprocess and feature recognition.For preprocess,a series of algorithm is designed such as simplifying,filtering,segmenting and so on.For feature recognition,the information of weld seam can be obtained efficiently and accurately through the operations of segmenting,layering,recognizing of edge information and restoration of topology information.For this purpose,in this paper,we propose successively a new layering algorithm,improve the classical RANSAC algorithm,and design a topology information restoration algorithm to realize the depth information restoration.Finally,the weld seam can be obtained directly from the original point cloud data without model reconstruction by performing these algorithm.In the process of optimizing the grinding tool orientations,the measured contour points are first interpolated using B-spline curve.Then,the grinding tool orientation is planned initially according to the tangent information of each point described by B-spline curve.Next,the point is characterized by the corresponding curvature,which is calculated from the interpolated B-spline curve,and then the measured points are divided into several crucial regions.Finally,the crucial regions are optimized aiming to smoothen the orientations by spherical linear interpolation and control z-fluctuation by the linear interpolation,which can make the whole grinding movement smoother and the wearing of grinding tool evener.At last,the validity and practicability of the proposed algorithms for weld seam identification and grinding tool orientation optimization are verified by numerical analysis and experiments.The proposed methods can be applied in the process of robot machining on the basis of the measured 3D point cloud of workpieces.The weld seam can be extracted effectively and the orientation of grinding tool can be smoothed,which improve the efficiency of production greatly.