Research On Automatic Welding Trajectory Planning Of Steel Mesh Based On 3D Vision

With the development of society and the extensive development of infrastructure construction,the demand for steel mesh is increasing.The steel mesh is a non-standard workpiece with many and discontinuous welds.Therefore,the welding of steel mesh is still mainly manual.However,manual welding has the disadvantages of harsh working environment,low efficiency and poor welding consistency.In order to overcome the shortcomings of manual welding,with the development of automation technology and robot welding technology,more and more welding scenarios begin to use robot welding instead of manual welding.At present,the programming methods of welding robots are mainly based on manual teaching programming and offline programming.Manual teaching programming requires frequent repeated programming to ensure the consistency of machining accuracy;offline programming is highly dependent on the 3D model of the workpiece.Therefore,neither of the two programming methods of robotic welding can be well applied to the welding of steel mesh.In order to overcome the above shortcomings and realize the robot automatic welding of steel mesh without programming and teaching,with the gradual deepening of the application of vision technology in the welding field,this thesis studies the vision-guided robot automatic welding technology,and proposes a 3D vision-based automatic welding technology.Welding trajectory planning method for steel mesh.The main research contents of this thesis are as follows:(1)A welding experiment system based on 3D vision is built,which is divided into two parts:a three-dimensional vision system and a welding execution system.Because the steel mesh has the characteristics of large size,poor consistency and many welding points,this thesis proposes the general idea of welding by 3D camera multi-point shooting.(2)Point cloud preprocessing.Due to the initial point cloud captured by the camera,the background is cluttered,dense and noisy.Use Voxel grids down-sampling filter to reduce point cloud density to improve processing efficiency;use pass-through filter design software to interactively extract point cloud ROI and remove redundant point cloud;use points based on Gaussian distribution Cloud filter removes point cloud noise;uses random consistency(RANSAC)based method to remove plane background of workpiece;finally proposes an iterative extraction method based on RANSAC to extract straight line point cloud for subsequent weld feature extraction.(3)Weld feature extraction and welding trajectory planning.Weld features in this thesis include weld gap width and weld feature points.First,use the line fitting method based on singular value decomposition(SVD)to perform line fitting on the point cloud of each steel bar;then,solve the common perpendiculars of the fitting line segments of the two intersecting steel bars and obtain the weld characteristics;finally,Plan the welding trajectory according to the welding process requirements.(4)Welding system software design.Combined with the overall process of welding and the analysis of specific requirements,the software development is realized on the Visual Studio+ Qt platform,and the software effects are displayed.(5)Experimental analysis.Three groups of experiments were designed in this thesis.The first set of experiments is to compare the weld gap width calculated by the method in this thesis with the actual measured weld gap width,which verifies the reliability of the method.The second set of experiments is to compare the welding trajectory planned by the method in this thesis with the manually taught welding trajectory,which verifies the accuracy of the method.The third group of experiments is to shoot workpieces at different points for welding trajectory planning,which verifies the robustness of the method.Finally,it is shown that the method proposed in this study can realize the robot automatic welding of steel mesh without programming and teaching.