Arc-initiating and extinguishing plate material feeding and positioning welding system and method

通过设计引熄弧板上料及定位焊接系统,利用3D相机和机器人实现自动化上料和焊接,解决了人工焊接效率低和劳动强度大的问题,提升了焊接自动化和智能化水平。

CN118720362BActive Publication Date: 2024-11-15TAIYUAN UNIVERSITY OF SCIENCE AND TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411230971.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2024-11-15
Estimated Expiration
2044-09-04

AI Technical Summary

Technical Problem

In the prior art, the installation and welding process of the evacuation arc plate relies on manual operation, resulting in low efficiency, high labor intensity and low degree of automation.

Method used

A feeding and positioning welding system for evacuation arc plates is designed, including a top machine, loading module and welding module. The structured light 3D camera and line laser 3D camera are used to cooperate with the robot to realize the automatic loading and positioning welding of evacuation arc plates. , coordinate the workflow of each module through collaborative motion control module and data interaction module to ensure the smooth completion of the task.

Benefits of technology

The automation and intelligence level of quenching arc plate welding has been improved, the problems of low efficiency and high labor intensity of traditional manual welding have been solved, and automated loading and welding operations in non-standard environments have been realized.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118720362B_ABST
    Figure CN118720362B_ABST
Patent Text Reader

Abstract

The present invention discloses a system and method for loading and positioning arc-initiating and extinguishing plates, comprising: a host computer, a loading module and a welding module; the host computer comprises: a coordinated motion control module, an arc-initiating and extinguishing plate posture estimation module, a weld identification and positioning module and a data interaction module between the modules; the host computer is used to control the loading module and the welding module; the loading module is used to absorb the arc-initiating and extinguishing plate and place it at a designated welding position; the welding module is used to identify and locate the weld position and perform welding operations. The present invention can solve the problems of high labor intensity and low efficiency of traditional manual arc-initiating and extinguishing plate welding, and realize the intelligent upgrade of loading and welding of arc-initiating and extinguishing plates.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of automatic feeding and welding of arc-initiating and extinguishing plates, and in particular to a system and method for feeding and positioning welding of arc-initiating and extinguishing plates. Background Art

[0002] In submerged arc welding processes such as steel plate butt welding, straight weld steel pipe welding, and I-beam welding, arc starter plates and arc extinguishing plates are essential components to ensure welding quality. The arc starter plates are removed after the weldment is welded normally. The function of the arc starter plate is to provide a starting point for the welding arc to prevent the air around the arc from being ionized and spattering, thereby ensuring the quality of the welded joint. The function of the arc extinguishing plate is to extend the end of the welded joint so that the slag and spatter at the end of the weld can be easily removed after welding is completed, and defects such as arc pits and incomplete penetration can be prevented from occurring in the welded joint. At present, the installation and welding process of the arc starter plate are all performed manually by handheld arc welding, which has low production efficiency. Summary of the invention

[0003] In view of the problems of low efficiency, high labor intensity and low automation in the current manual welding of arc-initiating and extinguishing plates, the present invention proposes a system and method for loading and positioning arc-initiating and extinguishing plates to realize automatic loading and positioning welding of arc-initiating and extinguishing plates.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] An arc-initiating and extinguishing plate loading and positioning welding system comprises: a host computer, a loading module and a welding module;

[0006] The host computer includes: a coordinated motion control module, an arc-ignition plate posture estimation module, a weld identification and positioning module, and a data interaction module between the modules;

[0007] The arc striking and extinguishing plate position and posture estimation module is used to obtain the position and posture information of the arc striking plate;

[0008] The weld identification and positioning module is used to obtain the weld end face profile and position information, and the weld position information between the arc starter plate and the weld;

[0009] The collaborative motion control module is used to assign tasks and execution order to the feeding module and the welding module according to the information obtained by the arc-initiating and extinguishing plate posture estimation module and the weld identification and positioning module, so as to ensure that the feeding module and the welding module complete the tasks according to the workflow without interference;

[0010] The data interaction module between the modules is used to coordinate the data interaction between the modules of the host computer;

[0011] The loading module is used to absorb the arc striking and extinguishing plate and place it at a designated welding position;

[0012] The welding module is used to identify and locate the weld position and perform welding operations.

[0013] Preferably, the loading module comprises: a loading robot, a structured light 3D camera and an electromagnet; wherein the electromagnet and the structured light 3D camera are installed at the end of the robot arm;

[0014] The structured light 3D camera is used to collect the scattered and stacked arc-initiating and arc-extinguishing plate scene point cloud, and the host computer performs segmentation and classification processing on the scene point cloud data to obtain the position and posture of the arc-initiating and arc-extinguishing plate;

[0015] The loading robot places the arc striking and extinguishing plate at the welding position;

[0016] The electromagnet is used to absorb the arc striking and extinguishing plate, and serves as the execution end of the loading robot to realize the arc striking and extinguishing plate grabbing operation.

[0017] Preferably, the welding module comprises: a welding robot, a line laser 3D camera and a welding machine; wherein the line laser 3D camera is installed at the end of the welding robot; the welding robot is equipped with a welding gun at the end;

[0018] The line laser 3D camera is used to collect the weld posture, track the weld posture in real time, and perform the welding operation by the welding robot;

[0019] The welding robot is used to issue arc initiation and arc extinguishing instructions to the welding machine, thereby controlling the welding process of the welding gun;

[0020] The welding machine is a controller of a welding gun and is used to set welding parameters.

[0021] Preferably, dual modules are used to work in collaboration, with the loading module and the welding module working together to complete the loading and positioning welding process of the arc-initiating and extinguishing plates.

[0022] The present invention also provides a method for arc-initiating and extinguishing plate loading and positioning welding, which is implemented by using the arc-initiating and extinguishing plate loading and positioning welding system, and includes the following steps:

[0023] S1: The loading robot carries a structured light 3D camera to the top of the loading area. The structured light 3D camera collects the scene point cloud of the loading area and sends it to the host computer;

[0024] S2: The host computer performs background removal and downsampling preprocessing on the scene point cloud, and uses supervoxel clustering and local convex connection algorithms to segment the preprocessed scene point cloud to obtain a single independent arc-igniting and arc-extinguishing plate point cloud subset in a scattered stacking state, and screens out candidate absorption targets according to the absorption scoring formula;

[0025] S3: Perform point cloud registration on the selected target point cloud data and the existing arc-initiating and extinguishing plate point cloud model to obtain the 6D pose of the target point cloud with the highest matching degree, and provide 6D pose data for the feeding robot to absorb;

[0026] S4: The loading robot picks up the arc-initiating and extinguishing plate and moves to the end face of the weldment. The structured light 3D camera collects the point cloud data of the weldment end face morphology, and uses the edge fitting algorithm to extract the weldment end face contour and position information to provide visual guidance and positioning for the installation and placement of the arc-initiating and extinguishing plate. The loading robot places the arc-initiating and extinguishing plate at the installation position, aligns it with the edge of the weldment, and keeps the loading robot posture unchanged;

[0027] S5: Start the welding robot, and the line laser 3D camera recognizes the weld between the arc-initiating and arc-extinguishing plate and the weldment. The host computer extracts the weld posture between the weldment and the arc-initiating and arc-extinguishing plate, plans the welding path, generates the welding trajectory offline, and the welding robot performs the welding operation.

[0028] S6: After welding is completed, the welding robot returns to the waiting point, and the loading robot continues to complete the loading process of the next arc-initiating and extinguishing plate.

[0029] Preferably, in S2, the method of performing point cloud segmentation on the preprocessed scene point cloud using supervoxel clustering and local convex connection algorithm includes:

[0030] Perform pixel-level segmentation on the preprocessed scene point cloud based on the supervoxel clustering algorithm;

[0031] Using the voxel cloud connectivity segmentation algorithm VCCS, the point cloud after pixel-level segmentation is clustered into several supervoxels, each of which is described by a 36-dimensional feature vector:

[0032]

[0033] Where x, y, z are the coordinates of the centroid of the super-body, FPFH 1...33 It is a 33-dimensional feature descriptor;

[0034] Establish the adjacency relationship between adjacent supervoxels, expressed by the centroid space distance D: ,in is the weight parameter, m is a constant, is the supervoxel radius;

[0035] After the supervoxel clustering is completed, the concave-convex relationship between different supervoxels is calculated for the over-segmented point cloud;

[0036] After marking the concave-convex relationship between adjacent supervoxels, the region growing algorithm is used to cluster small regions into larger objects. The region growing algorithm is limited by the concave-convexity of the small region, that is, it only allows the region to grow across the convex edge. At this point, the segmentation is completed.

[0037] Preferably, in S2, the method of screening out candidate absorption targets according to the absorption scoring formula includes:

[0038] Create a scoring formula:

[0039]

[0040] Where: c1 and c2 are the weight coefficients of the point cloud centroid position and the number of point clouds, Z represents the distance from the point cloud centroid to the origin of the structured light 3D camera coordinate system, N represents the number of points in the point cloud, and N0 is a constant;

[0041] According to the scoring formula, the target with the top layer and the most complete number of point clouds is selected.

[0042] Preferably, in S3, the method of performing point cloud registration on the selected target point cloud data and the existing arc striking and extinguishing plate point cloud model to obtain the target point cloud 6D pose with the highest matching degree includes:

[0043] The sampling consistency algorithm SAC-IA is used for coarse registration to obtain the initial transformation matrix (R0, t0);

[0044] Based on the particle swarm optimization algorithm and iterative closest point algorithm (PSO-ICP) precise alignment, the optimal transformation matrix (R, t) is obtained, thereby calculating the precise 6D pose of the target.

[0045] Preferably, in S4, the method of extracting the weldment end face profile and position information using an edge fitting algorithm includes:

[0046] S41: performing discrete point removal and filtering downsampling on the collected point cloud;

[0047] S42: Use kd-tree to implement k-nearest neighbor data search, build a tree-like topological relationship of point cloud data, fit the k-nearest neighbor point plane through least squares, and calculate the plane normal vector of each point;

[0048] S43: Solve the normal curvature according to the plane normal vector of each point; Specifically: Assume that the m nearest neighboring points of point p and its surrounding neighborhood are , let N be the normal vector of the point, Q i For point q i The normal vector of point p is relative to q i The normal curvature of is estimated as follows:

[0049]

[0050] In the formula, The vector N and the vector The angle between vector N and vector Q. iThe angle between For p and q i The Euclidean distance between

[0051] S44: According to the normal curvature and Euler's formula, the principal curvature problem is solved, that is, the relationship between the normal line and the principal curvature is converted into a least squares problem shown in the following formula. The principal curvatures k1 and k2 can be obtained by solving this least squares problem;

[0052]

[0053] In the formula, θ i +θ is the principal direction passing through point p through q i The angle between the tangent lines of the normal section;

[0054] S45: After obtaining the principal curvature, a threshold is set for the maximum principal curvature, and points greater than the threshold, i.e., contour feature points, are saved to complete the extraction of boundary contour features;

[0055] S46: Calculate the coordinates of the center point of the contour according to the calculated boundary contour features, thereby calculating the coordinates and posture of the material placement point of the loading robot.

[0056] Preferably, in S5, the method of fitting and extracting the weld posture between the weldment and the arc striking and extinguishing plate, and planning the welding path includes:

[0057] Use a line laser 3D camera to collect the weld image between the arc-starting and arc-extinguishing plate and the weldment;

[0058] The laser center line is extracted based on the grayscale centroid method, and the characteristic points of the weld are obtained using the slope analysis method and the straight line fitting algorithm. The coordinates of the characteristic points are converted from the camera coordinate system to the robot base coordinate system through the hand-eye matrix and the robot DH parameters.

[0059] The coordinates of all extracted feature points are fitted with straight lines to plan the welding path.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] The present invention provides an arc-initiating and extinguishing plate loading and positioning welding system and method, which provides a set of automatic loading and welding solutions in non-standard environments for the arc-initiating and extinguishing plate welding process, solves the problems of low efficiency and high labor intensity of traditional manual welding of arc-initiating and extinguishing plates, and improves the automation and intelligence level of arc-initiating and extinguishing plate welding. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0063] Figure 1 It is a schematic diagram of a control system for an arc-initiating and arc-extinguishing plate disorderly feeding and positioning welding system according to an embodiment of the present invention;

[0064] Figure 2 A schematic diagram of the structural layout of a system for disorderly feeding and positioning welding of arc-initiating and arc-extinguishing plates according to an embodiment of the present invention;

[0065] Figure 3 This is a schematic diagram of the arc-initiating and arc-extinguishing plate of the weldment after welding according to an embodiment of the present invention;

[0066] Figure 4 A schematic flow chart of a method for disorderly feeding and positioning welding of arc-initiating and arc-extinguishing plates according to an embodiment of the present invention;

[0067] Description of the drawings: 1- loading robot; 2- welding robot. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0070] Embodiment 1

[0071] like Figure 1-3 As shown, the present invention provides an arc-initiating and extinguishing plate loading and positioning welding system, comprising: a host computer, a loading module and a welding module;

[0072] The host computer includes: a collaborative motion control module, an arc-ignition plate posture estimation module, a weld identification and positioning module, and a data interaction module between the modules;

[0073] The arc striking and extinguishing plate posture estimation module is used to obtain the position and posture information of the arc striking plate;

[0074] The weld identification and positioning module is used to obtain the weld end face profile and position information, and the weld position information between the arc starter plate and the weld;

[0075] The collaborative motion control module is used to assign tasks and execution order to the feeding module and welding module according to the information obtained by the arc-initiating and extinguishing plate posture estimation module and the weld identification and positioning module, ensuring that the feeding module and the welding module complete their tasks according to the workflow without interference;

[0076] The data interaction module between modules is used to coordinate the data interaction between modules of the host computer;

[0077] The loading module is used to absorb the arc-starting and extinguishing plates and place them at the designated welding position;

[0078] The welding module is used to identify and locate the weld position and perform welding operations.

[0079] In this embodiment, the loading module includes: a loading robot, a structured light 3D camera and an electromagnet; wherein the electromagnet and the structured light 3D camera are installed at the end of the robot arm;

[0080] The structured light 3D camera is used to collect the scattered and stacked arc-initiating and arc-extinguishing plate scene point cloud, and the host computer segments and classifies the scene point cloud data to obtain the position and posture of the arc-initiating and arc-extinguishing plates;

[0081] The loading robot places the arc striking and extinguishing plate into the welding position;

[0082] The electromagnet is used to absorb the arc-initiating and extinguishing plate, and serves as the execution end of the loading robot to realize the arc-initiating and extinguishing plate grabbing operation.

[0083] In this embodiment, the welding module includes: a welding robot, a line laser 3D camera and a welding machine; wherein the line laser 3D camera is installed at the end of the welding robot; the welding robot is equipped with a welding gun at the end;

[0084] The line laser 3D camera is used to collect the weld posture, track the weld posture in real time, and the welding robot performs the welding operation;

[0085] The welding robot is used to give arc initiation and arc extinguishing instructions to the welding machine, thereby controlling the welding process of the welding gun;

[0086] The welding machine is the controller of the welding gun, which is used to set welding parameters, such as welding current and voltage.

[0087] In this embodiment, dual modules are used to work in collaboration, and the loading module and the welding module cooperate to complete the loading and positioning welding process of the arc-initiating and extinguishing plates.

[0088] Embodiment 2

[0089] like Figure 4As shown, the present invention also provides a method for loading and positioning arc-extinguishing plates, which is implemented by using the above-mentioned loading and positioning arc-extinguishing plates system, and is specifically applied to the automatic loading and welding process of arc-extinguishing plates in H-beam welding. The specific implementation plan is as follows:

[0090] S1: The loading robot carries a structured light 3D camera to the top of the loading area. The structured light 3D camera collects the scene point cloud of the loading area and sends it to the host computer;

[0091] S2: The host computer performs background removal and downsampling preprocessing on the scene point cloud, and uses supervoxel clustering and local convex connection algorithms to segment the preprocessed scene point cloud to obtain a single independent arc-igniting and arc-extinguishing plate point cloud subset in a scattered stacking state, and screens out candidate absorption targets according to the absorption scoring formula;

[0092] S3: Perform point cloud registration on the selected target point cloud data and the existing arc striking and extinguishing plate point cloud model (the existing arc striking and extinguishing plate point cloud model refers to a complete arc striking and extinguishing plate 3D point cloud model made by CAD software, and the camera collects the arc striking plate point cloud in the real scene), and obtain the 6D pose of the target point cloud with the highest matching degree, so as to provide 6D pose data for the feeding robot to absorb;

[0093] S4: The loading robot picks up the arc-initiating and extinguishing plate and moves to the end face of the weldment. The structured light 3D camera collects the point cloud data of the weldment end face morphology, and uses the edge fitting algorithm to extract the weldment end face contour and position information to provide visual guidance and positioning for the installation and placement of the arc-initiating and extinguishing plate. The loading robot places the arc-initiating and extinguishing plate at the installation position, aligns it with the edge of the weldment, and keeps the loading robot posture unchanged;

[0094] S5: Start the welding robot, and the line laser 3D camera recognizes the weld between the arc-initiating and arc-extinguishing plate and the weldment. The host computer extracts the weld posture between the weldment and the arc-initiating and arc-extinguishing plate, plans the welding path, generates the welding trajectory offline, and the welding robot performs the welding operation.

[0095] S6: After welding is completed, the welding robot returns to the waiting point, and the loading robot continues to complete the loading process of the next arc-initiating and extinguishing plate.

[0096] In this embodiment, since the two robot ends included in the arc-strike-and-extinguishing plate loading and positioning welding system are equipped with industrial cameras, it is necessary to obtain the conversion relationship between the camera coordinate system and the robot coordinate system before the system is applied, that is, to solve the hand-eye matrix. The arc-strike-and-extinguishing plate absorption posture and weld posture obtained by the above-mentioned structured light 3D camera and line laser 3D camera need to be converted to the robot coordinate system through the hand-eye matrix.

[0097] Set the hand-eye matrix to X, and the coordinates of the target object in the camera coordinate system to [Xc, Yc, Zc, 1] T , the coordinates of the target object in the robot base coordinate system are [Xb, Yb, Zb, 1] T .

[0098] The calculation formula of the hand-eye matrix is: , .

[0099] In this embodiment, in S2, the method of performing point cloud segmentation on the preprocessed scene point cloud using supervoxel clustering and local convex connection algorithm includes:

[0100] First, the preprocessed scene point cloud is segmented at the pixel level based on the supervoxel clustering algorithm, and then the voxel cloud connectivity segmentation algorithm (VCCS) is used to cluster the pixel-level segmented point cloud into several supervoxels. Each supervoxel is described by a 36-dimensional feature vector:

[0101]

[0102] Where x, y, z are the coordinates of the centroid of the super-body, FPFH 1...33 It is a 33-dimensional feature descriptor.

[0103] Establish the adjacency relationship between adjacent supervoxels. FPFH is a position-invariant feature that uses the k nearest neighbor combination of a point to describe the local surface model properties of the point. Each voxel is assigned to the supervoxel according to the minimum normalized distance D between the voxel and the supervoxel centroid: ,in is the Euclidean distance in color space; is the spatial distance; is the distance in the fast point histogram space, and are the weights of color, spatial distance and normal vector respectively, m is a constant, is the supervoxel radius. The present invention does not consider color information, so the above formula is modified to After the supervoxel clustering is completed, the concave-convex relationship between different supervoxels needs to be calculated for the over-segmented point cloud. The concave-convex relationship is judged by the CC (extended convexity criterion), in which CC uses the angle between the centroid of two adjacent supervoxels and the normal vector to judge whether the relationship between the two supervoxels is concave or convex. If a1 is greater than a2, it is concave, otherwise it is convex. Considering factors such as measurement noise, a threshold needs to be introduced in actual use, that is, a1 needs to be larger than a2 by a certain amount to filter out smaller concave and convex misjudgments. In addition, in order to remove misjudgments caused by some small noise, third-party verification is also required. If a supervoxel intersects with two adjacent supervoxels, their concave and convex relationships must be the same. The CC criterion is finally as CCe:

[0104]

[0105] After marking the concave-convex relationship between adjacent supervoxels, the region growing algorithm is used to cluster small regions into larger objects. This region growing algorithm is limited by the concave-convexity of the small region, that is, it only allows the region to grow across the convex edge. At this point, the segmentation is completed.

[0106] In this embodiment, in S2, the method of screening out candidate absorption targets according to the absorption scoring formula includes:

[0107] In order to measure the absorption priority between the segmented targets and determine the optimal absorption target, a scoring formula is established:

[0108]

[0109] Where: c1 and c2 are the weight coefficients of the point cloud centroid position and the number of point clouds, Z represents the distance from the point cloud centroid to the origin of the structured light 3D camera coordinate system, N represents the number of points in the point cloud, and N0 is a constant;

[0110] According to the scoring formula, the target with the top layer and the most complete number of point clouds is selected.

[0111] In this embodiment, in S3, the method of performing point cloud registration on the selected target point cloud data and the existing arc-initiating and extinguishing plate point cloud model to obtain the target point cloud 6D pose with the highest matching degree includes:

[0112] For the point cloud registration problem in step S3, initial registration is performed based on the sampling consistency algorithm, and the optimal transformation matrix (R, t) is solved based on the gradient optimization algorithm. Finally, fine registration is performed based on the iterative closest point algorithm to solve the problem that the traditional point cloud registration method has low accuracy and is prone to falling into local optimality. In this way, a high-precision 6D absorption posture of the arc-initiating and extinguishing plate is obtained to ensure the assembly accuracy of the arc-initiating and extinguishing plate.

[0113] The basic mathematical model of point cloud registration is: when the corresponding points of two sets of point clouds are known, how to solve the transformation matrix, that is, the translation matrix t and the rotation matrix R. Define the source point cloud point set P and the target point cloud point set Q, let P = {p1, p2, ... p n} and Q={q1,q2,…q n} is R 3 The points corresponding to the two sets of point clouds in space, where pn is the coordinate of the nth point in the P point set (x n ,y n ,z n ), similarly: q n is the coordinate of the nth point in the Q point set (x n ,y n ,z n ). Our goal is to find an optimal transformation matrix that optimally aligns two point sets in the sense of least squares, that is, to find a rotation matrix R and a translation matrix t that satisfies:

[0114] ,

[0115] In the formula, is the weight of each point pair. The above formula is the classic objective function of the point cloud registration algorithm, and it is also the most essential objective function that can reflect the point cloud registration.

[0116] For solving the above objective function, the present invention proposes a new point cloud registration method by combining the particle swarm optimization algorithm (PSO) and the iterative closest point algorithm (ICP), thereby improving the registration accuracy and overcoming the shortcoming that the traditional ICP point cloud registration method is prone to fall into the local optimum.

[0117] First, the sampling consistency algorithm (SAC-IA) is used for rough registration to obtain the initial transformation matrix , as the initial value of the PSO-ICP algorithm.

[0118] Then, the PSO-ICP algorithm is used for precise registration to obtain the optimal transformation matrix. .

[0119] The algorithm flow is as follows:

[0120] (1) Initialize particle swarm

[0121] Initialization parameters: rotation parameter R0, translation parameter t0, number of iterations K.

[0122] (2) Fitness evaluation

[0123] The new and fitness, This is the result of this iteration. The fitness function calculation formula of the particle swarm is: ,in, is a point in the source point cloud, is the target point cloud and The nearest point, is the distance metric function.

[0124] (3) Individual extreme values ​​and global extreme values

[0125] Each particle determines the individual extreme value (pBest) according to its own fitness, and the particle group shares information to determine the global optimal solution (gBest):

[0126]

[0127] (4) Particle update

[0128] The particle's position and velocity are updated according to the following formula:

[0129]

[0130] in: It is a particle speed; is the inertia weight, which controls the decay of particle velocity; and is a learning factor that controls the influence of individual extreme values ​​and global extreme values; and is a random number in the range [0,1], simulating the random behavior of particles.

[0131] (5) Speed ​​and position limits

[0132] In order to prevent particles from flying out of the search space, the speed and position need to be restricted:

[0133]

[0134] in: and are the upper and lower bounds of the velocity, and are the upper and lower bounds of the position.

[0135] (6) Iteration termination condition

[0136] The iteration termination condition may be that the maximum number of iterations is reached or the accuracy of the solution meets a preset threshold.

[0137] (7) Output results

[0138] Output the transformation matrix (R, t) corresponding to the global optimal solution.

[0139] According to the final output results, the grasping posture of the feeding robot can be calculated in combination with the coordinates of the grasping target point.

[0140] The precise 6D pose of the target refers to the relative coordinates (X, Y, Z) of the arc-inducing and extinguishing plate to be sucked relative to the template point cloud in the robot base coordinate system and the rotation angle (R) relative to the known template point cloud. x ,R y ,R z), corresponding to the translation matrix t and the rotation matrix R. Through this parameter, combined with the coordinates of the grasping point, the robot's grasping posture (X, Y, Z, RX, RY, RZ) can be calculated. Converting the solved rotation matrix R into Euler angles is the robot's grasping posture (RX, RY, RZ). In general, the 6D posture is the translation coordinates of 3 dimensions and the rotation angles of 3 dimensions.

[0141] In this embodiment, dual robots S4 and S5 work together, and interference between the loading robot and the welding robot is avoided through the terminal structure design and robot posture setting.

[0142] In this embodiment, in S4, the method of extracting the weldment end face profile and position information using the edge fitting algorithm includes:

[0143] S41: performing discrete point removal and filtering downsampling on the collected point cloud;

[0144] S42: Use kd-tree to implement k-nearest neighbor data search, build a tree-like topological relationship of point cloud data, fit the k-nearest neighbor point plane through least squares, and calculate the plane normal vector of each point;

[0145] S43: Solve the normal curvature according to the plane normal vector of each point; Specifically: Assume that the m nearest neighboring points of point p and its surrounding neighborhood are , let N be the normal vector of the point, Q i For point q i The normal vector of point p is relative to q i The normal curvature of is estimated as follows:

[0146]

[0147] In the formula, The vector N and the vector The angle between vector N and vector Q. i The angle between For p and q i The Euclidean distance between

[0148] S44: According to the normal curvature and Euler's formula, solving the principal curvature problem, that is, the relationship between the normal line and the principal curvature, can be transformed into solving the least squares problem shown in the following formula. Solving this least squares problem can obtain the principal curvatures k1 and k2;

[0149]

[0150] In the formula, θ i +θ is the principal direction passing through point p through q i The angle between the tangent lines of the normal section;

[0151] S45: After obtaining the principal curvature, a threshold is set for the maximum principal curvature, and points greater than the threshold, i.e., contour feature points, are saved to complete the extraction of boundary contour features;

[0152] S46: Calculate the coordinates of the center point of the contour according to the calculated boundary contour features, thereby calculating the coordinates and posture of the material placement point of the loading robot.

[0153] In this embodiment, in S5, the method of fitting and extracting the weld posture between the weldment and the arc striking and extinguishing plate, and planning the welding path includes:

[0154] Use a line laser 3D camera to collect the weld image between the arc-starting and arc-extinguishing plate and the weldment;

[0155] The laser center line is extracted based on the grayscale centroid method, and the characteristic points of the weld are obtained using the slope analysis method and the straight line fitting algorithm. The coordinates of the characteristic points are converted from the camera coordinate system to the robot base coordinate system through the hand-eye matrix and the robot DH parameters.

[0156] The coordinates of all extracted feature points are fitted with straight lines to plan the welding path.

[0157] In this embodiment, the present invention uses H-shaped steel as a weldment, as shown in the attached Figure 2 As shown, four arc-starting and arc-extinguishing plates need to be installed on both end faces of the H-beam. The working sequence is: first install the arc-starting plate at the front end of the H-beam, then transport the rear end face of the H-beam to the working area through a roller conveyor, and then install the arc-extinguishing plate at the rear end face of the H-beam.

[0158] In this embodiment, the loading robot keeps the position unchanged when bringing the arc-initiating and extinguishing plate to the installation position until the positioning welding is completed, and then the welding robot and the loading robot leave the working area successively to execute the next process.

[0159] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for arc-initiating and arc-extinguishing plate material preparation and positioning welding, characterized in that: The arc-initiating and extinguishing plate loading and positioning welding system is used to achieve the above, and the system includes: a host computer, a loading module and a welding module; The host computer includes: a coordinated motion control module, an arc-ignition plate posture estimation module, a weld identification and positioning module, and a data interaction module between the modules; The arc striking and extinguishing plate position and posture estimation module is used to obtain the position and posture information of the arc striking plate; The weld identification and positioning module is used to obtain the weld end face profile and position information, and the weld position information between the arc starter plate and the weld; The collaborative motion control module is used to assign tasks and execution order to the feeding module and the welding module according to the information obtained by the arc-initiating and extinguishing plate posture estimation module and the weld identification and positioning module, so as to ensure that the feeding module and the welding module complete the tasks according to the workflow without interference; The data interaction module between the modules is used to coordinate the data interaction between the modules of the host computer; The loading module is used to absorb the arc striking and extinguishing plate and place it at a designated welding position; The welding module is used to identify and locate the weld position and perform welding operations; The method comprises the following steps: S1: The loading robot carries a structured light 3D camera to the top of the loading area. The structured light 3D camera collects the scene point cloud of the loading area and sends it to the host computer; S2: The host computer performs background removal and downsampling preprocessing on the scene point cloud, and uses supervoxel clustering and local convex connection algorithms to segment the preprocessed scene point cloud to obtain a single independent arc-igniting and arc-extinguishing plate point cloud subset in a scattered stacking state, and screens out candidate absorption targets according to the absorption scoring formula; S3: Perform point cloud registration on the selected target point cloud data and the existing arc-initiating and extinguishing plate point cloud model to obtain the 6D pose of the target point cloud with the highest matching degree, and provide 6D pose data for the feeding robot to absorb; S4: The loading robot picks up the arc-initiating and extinguishing plate and moves to the end face of the weldment. The structured light 3D camera collects the point cloud data of the weldment end face morphology, and uses the edge fitting algorithm to extract the weldment end face contour and position information to provide visual guidance and positioning for the installation and placement of the arc-initiating and extinguishing plate. The loading robot places the arc-initiating and extinguishing plate at the installation position, aligns it with the edge of the weldment, and keeps the loading robot posture unchanged; S5: Start the welding robot, and the line laser 3D camera recognizes the weld between the arc-initiating and arc-extinguishing plate and the weldment. The host computer extracts the weld posture between the weldment and the arc-initiating and arc-extinguishing plate, plans the welding path, generates the welding trajectory offline, and the welding robot performs the welding operation. S6: After the welding is completed, the welding robot returns to the waiting point, and the loading robot continues to complete the loading process of the next arc-initiating and extinguishing plate; In S2, the method of performing point cloud segmentation on the preprocessed scene point cloud using supervoxel clustering and local convex connection algorithm includes: Perform pixel-level segmentation on the preprocessed scene point cloud based on the supervoxel clustering algorithm; Using the voxel cloud connectivity segmentation algorithm VCCS, the point cloud after pixel-level segmentation is clustered into several supervoxels, each of which is described by a 36-dimensional feature vector: Where x, y, z are the coordinates of the centroid of the super-body, FPFH 1...33 It is a 33-dimensional feature descriptor; Establish the adjacency relationship between adjacent supervoxels, expressed by the centroid space distance D: ,in is the weight parameter, m is a constant, is the supervoxel radius, is the Euclidean distance in color space, is the spatial distance, is the distance in the fast point histogram space, is the weight of the normal vector; After the supervoxel clustering is completed, the concave-convex relationship between different supervoxels is calculated for the over-segmented point cloud; After marking the concave-convex relationship between adjacent supervoxels, the region growing algorithm is used to cluster small regions into larger objects. The region growing algorithm is limited by the concave-convexity of the small region, that is, it only allows the region to grow across the convex edge. At this point, the segmentation is completed; In S2, according to the absorption scoring formula, the method of screening out candidate absorption targets includes: Create a scoring formula: Where: c1 and c2 are the weight coefficients of the point cloud centroid position and the number of point clouds, Z represents the distance from the point cloud centroid to the origin of the structured light 3D camera coordinate system, N represents the number of points in the point cloud, and N0 is a constant; According to the scoring formula, the target with the top layer and the most complete number of point clouds is selected.

2. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1 is characterized in that: The loading module includes: a loading robot, a structured light 3D camera and an electromagnet; wherein the electromagnet and the structured light 3D camera are installed at the end of the robot arm; The structured light 3D camera is used to collect the scattered and stacked arc-initiating and arc-extinguishing plate scene point cloud, and the host computer performs segmentation and classification processing on the scene point cloud data to obtain the position and posture of the arc-initiating and arc-extinguishing plate; The loading robot places the arc striking and extinguishing plate at the welding position; The electromagnet is used to absorb the arc striking and extinguishing plate, and serves as the execution end of the loading robot to realize the arc striking and extinguishing plate grabbing operation.

3. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1, characterized in that: The welding module comprises: a welding robot, a line laser 3D camera and a welding machine; wherein the line laser 3D camera is installed at the end of the welding robot; and the welding robot is equipped with a welding gun at the end; The line laser 3D camera is used to collect the weld posture, track the weld posture in real time, and perform the welding operation by the welding robot; The welding robot is used to issue arc initiation and arc extinguishing instructions to the welding machine, thereby controlling the welding process of the welding gun; The welding machine is a controller of a welding gun and is used to set welding parameters.

4. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1, characterized in that: The dual modules work together, with the loading module and the positioning module working together to complete the loading and positioning welding process of the arc-initiating and extinguishing plates.

5. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1, characterized in that: In S3, the method of performing point cloud registration on the selected target point cloud data and the existing arc striking and extinguishing plate point cloud model to obtain the target point cloud 6D pose with the highest matching degree includes: The sampling consistency algorithm SAC-IA is used for coarse registration to obtain the initial transformation matrix (R0, t0); Based on the particle swarm optimization algorithm and iterative closest point algorithm (PSO-ICP) precise alignment, the optimal transformation matrix (R, t) is obtained, thereby calculating the precise 6D pose of the target.

6. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1, characterized in that: In S4, the method of extracting the weldment end face profile and position information using the edge fitting algorithm includes: S41: performing discrete point removal and filtering downsampling on the collected point cloud; S42: Use kd-tree to implement k-nearest neighbor data search, build a tree-like topological relationship of point cloud data, fit the k-nearest neighbor point plane through least squares, and calculate the plane normal vector of each point; S43: Solve the normal curvature according to the plane normal vector of each point; Specifically: Assume that the m nearest neighboring points of point p and its surrounding neighborhood are , let N be the normal vector of the point, Q i For point q i The normal vector of point p is relative to q i The normal curvature of is estimated as follows: In the formula, The vector N and the vector The angle between vector N and vector Q. i The angle between For p and q i The Euclidean distance between S44: According to the normal curvature and Euler's formula, the principal curvature problem is solved, that is, the relationship between the normal line and the principal curvature is converted into a least squares problem shown in the following formula. The principal curvatures k1 and k2 can be obtained by solving this least squares problem; In the formula, θ i +θ is the principal direction passing through point p through q i The angle between the tangent lines of the normal section; S45: After obtaining the principal curvature, a threshold is set for the maximum principal curvature, and points greater than the threshold, i.e., contour feature points, are saved to complete the extraction of boundary contour features; S46: Calculate the coordinates of the center point of the contour according to the calculated boundary contour features, thereby calculating the coordinates and posture of the material placement point of the loading robot.

7. The arc-initiating and extinguishing plate material preparation and positioning welding method according to claim 1, characterized in that: In S5, the method of fitting and extracting the weld posture between the weldment and the arc-initiating and extinguishing plate, and planning the welding path includes: Use a line laser 3D camera to collect the weld image between the arc-starting and arc-extinguishing plate and the weldment; The laser center line is extracted based on the grayscale centroid method, and the characteristic points of the weld are obtained using the slope analysis method and the straight line fitting algorithm. The coordinates of the characteristic points are converted from the camera coordinate system to the robot base coordinate system through the hand-eye matrix and the robot DH parameters. The coordinates of all extracted feature points are fitted with straight lines to plan the welding path.