Research On Weld Seam Recognition And Path Planning Of Welding Robot

The wide application of welding robot technology has significantly improved the overall production efficiency and quality of the welding manufacturing industry.With the transformation of the national economic development mode,in the context of the transformation and upgrading of the manufacturing industry,the traditional automated manufacturing is becoming more flexible and intelligent.The welding processing requirements for small batches and various types of workpieces are gradually increasing.The method of planning the welding path by online teaching has great limitations in single-piece and small-batch production.Besides,the offline path planning based on CAD model has high requirements on the accuracy of the workpiece itself and the processing position accuracy.So this method is difficult to ensure the welding quality.Therefore,the research on intelligent welding technology is of great significance.This paper focuses on the welding position identification and the independent planning of welding path in the field of intelligent welding technology.Based on the 3D point cloud data of the workpiece,a set of method for the identification of the right angle weld and the independent planning method of the welding path are proposed.A single arc welding robot is used to weld workpieces in different weld forms.This set of methods helps to achieve the welding of workpieces in different weld forms by a single arc welding robot.The main contents of this paper include:1.The task requirements and equipment conditions of the subject are analyzed.In addition,the structure of the robot welding system and the functions to be realized are introduced.Also,the principle methods of different point cloud data acquisition equipment are compared.According to the requirements of this research,the 3D scanner was used to measure the workpiece and the original point cloud data was obtained.2.The preprocessing scheme of point cloud data is studied,and the feature information of the workpiece to be welded is extracted effectively.The noise of the point cloud is removed,and the point cloud data is simplified while the feature of the workpiece is completely retained.3.A new method of weld seam identification is proposed.The concept of virtual ruler and the concept of boundary extension are proposed respectively.The recognition task is divided into two processes: local feature matching and continuous feature search.Firstly,the GA(Genetic Algorithm)is used to optimize the pose transformation parameters,and the point cloud is partially aligned with the workpiece point cloud to identify the pose of a certain weld in the workpiece,and then the ruler boundary is extended to search for the position of the workpiece weld with the same characteristics.Replace the workpiece feature with the ruler feature after completing the search.After the search is completed,the workpiece features arereplaced by the ruler features,and the position information and posture information of the entire weld seam of the workpiece are extracted.4.By using the position information and posture information of the weld position marked during the search of the ruler boundary extension,the corresponding welding path point data and welding gun attitude data are further calculated,and the data is processed to realize the independent planning of the welding path.5.By using the offline simulation software of the robot to build the simulation environment,the autonomous planning and simulation planning are compared.The results show that the path of the self-planning path and the simulation planning have a high degree of coincidence,and the rationality of the self-planning welding gun posture is verified.Finally,the existing path points are optimized to obtain a more ideal welding path and torch attitude.