A symmetry workpiece weld positioning method based on single station structured light vision

Abstract

This application belongs to the field of welding automation technology, specifically disclosing a method for symmetrical workpiece weld seam positioning based on single-camera structured light vision. The method includes: classifying planar point cloud data using the original point cloud data of the welding workpiece to obtain multiple principal planar point cloud sets and a corresponding thickness surface point cloud set for each principal planar point cloud set; dividing the workpiece into plate structures using each principal planar point cloud set and its corresponding thickness surface point cloud set to determine multiple independent plate structure models; modeling the workpiece geometry based on the geometric topological constraints between the various plate structure models to obtain a welding workpiece structure model; and determining the spatial position of each weld seam in the welding workpiece based on the connection relationships between adjacent plates in the welding workpiece structure model. This application enables accurate and stable weld seam positioning even when the point cloud acquired by a single-camera structured light vision system is incomplete.

Description

Technical Field

[0001] This application belongs to the field of welding automation technology, specifically to the field of computer vision technology, and more specifically, to a method for locating weld seams in symmetrical workpieces based on single-camera structured light vision. Background Technology

[0002] In the field of welding automation, weld seam positioning is a key prerequisite for achieving robotic automated welding. Symmetrical welded workpieces are typically composed of multiple symmetrically distributed plates, with the weld seam generally located at the interface between adjacent plates.

[0003] In practical industrial applications, due to limitations in production space, the installation location of structured light vision sensors, and economic considerations, scanning of such welded workpieces is typically limited to a single viewpoint. This results in point cloud data with issues such as partial occlusion and missing data. To address the incomplete point cloud data, existing technologies often employ model matching methods, which match the actual collected point cloud data with a pre-established workpiece template to locate the weld. However, this method usually requires the pre-establishment of an accurate model of the welded workpiece, making the model construction and matching process complex. Furthermore, when the workpiece model changes and no corresponding template is available, effective weld location becomes impossible, leading to poor stability and applicability of the method.

[0004] Therefore, under single-camera structured light vision acquisition conditions, how to understand the structure of symmetrical welded workpieces when point cloud data is incomplete, and on this basis achieve stable and accurate weld positioning, remains a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this application is to better realize the weld seam positioning of symmetrical welded workpieces, and to solve the problems of poor stability and applicability of the existing weld seam positioning methods.

[0006] To achieve the above objectives, in a first aspect, this application provides a method for locating weld seams in symmetrical workpieces based on single-camera structured light vision, comprising: The original point cloud data of the welded workpiece is used to classify the planar point cloud data, resulting in multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set; The workpiece plate structure is divided using each main plane point cloud set and its corresponding thickness surface point cloud set to determine multiple independent plate structure models. Based on the geometric and topological constraints between the various plate structure models, the workpiece geometry is modeled to obtain the welded workpiece structure model. Based on the connection relationship between adjacent plates in the structural model of the welded workpiece, the spatial position of each weld in the welded workpiece is determined.

[0007] Optionally, the step of classifying planar point cloud data using the original point cloud data of the welded workpiece to obtain multiple principal planar point cloud sets and a thickness surface point cloud set corresponding to each principal planar point cloud set includes: The original point cloud data is used to construct and analyze bounding boxes to determine the voxel size benchmark, thereby determining the voxel downsampling size, neighborhood search size, clustering space scale, and target analysis scale. The original point cloud data is downsampled according to the voxel downsampling size to obtain downsampled point cloud data, and the local surface normal vector corresponding to each point in the downsampled point cloud data is determined based on the neighborhood search size. Classify the points based on the coordinate axis components of the local surface normal vector corresponding to each point, determine multiple point cloud subsets with the same normal vector direction, and perform cluster analysis on each point cloud subset according to the clustering space scale to determine multiple principal plane point cloud sets and multiple non-principal plane point cloud sets corresponding to each point cloud subset. Spatial geometric analysis is performed on each principal plane point cloud set and its corresponding non-principal plane point cloud set using the target analysis scale, and the thickness surface point cloud set corresponding to each principal plane point cloud set is determined from the corresponding non-principal plane point cloud set.

[0008] Optionally, the principal plane point cloud set includes points whose principal normal direction is non-linear. z The first principal plane point cloud set and the principal direction of the normal vector are: z The second principal plane point cloud set in the direction; the process of dividing the workpiece plate structure using each principal plane point cloud set and its corresponding thickness surface point cloud set to determine multiple independent plate structure models includes: Three-dimensional plate structure modeling is performed using each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models. Using each set of point clouds of the second principal plane, a three-dimensional plate structure model is performed to obtain multiple independent second plate structure models; Based on multiple first plate structure models and multiple second plate structure models, multiple independent plate structure models are obtained.

[0009] Optionally, the step of using each first principal plane point cloud set and its corresponding thickness surface point cloud set to perform three-dimensional plate structure modeling to obtain multiple independent first plate structure models includes: For any one of the first principal plane point cloud sets in each of the first principal plane point cloud sets, the normal vector direction of the principal plane formed by the first principal plane point cloud sets is taken as the first principal direction of the orthogonal coordinate system, and the normal vector direction of the thickness surface formed by the corresponding thickness surface point cloud sets is taken as the second principal direction of the orthogonal coordinate system. The third principal direction of the orthogonal coordinate system is determined based on the first principal direction and the second principal direction; Based on the maximum and minimum values ​​of the point cloud coordinates in the first principal direction, the second principal direction, and the third principal direction, three-dimensional modeling is performed on each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models.

[0010] Optionally, the step of using each second principal plane point cloud set to perform three-dimensional plate structure modeling to obtain multiple independent second plate structure models includes: Project the point cloud data in each of the second principal plane point cloud sets to xy The plane is used to obtain the two-dimensional projection point set corresponding to each point cloud set of the second principal plane; Extract the outer contour of each of the two-dimensional projection point sets to obtain the outer contour graphics of each plate; Extract the line edge information of the outer contour graphics of each plate, and map the line edge information back to the three-dimensional space and the corresponding thickness surface point cloud set to perform three-dimensional plate structure modeling, thereby obtaining multiple independent second plate structure models.

[0011] Optionally, the geometric topological constraints include plate thickness information, height constraint information, length constraint information, and geometric symmetry information; the step of modeling the workpiece geometry based on the geometric topological constraints between the various plate structure models to obtain the welded workpiece structure model includes: Based on the plate thickness information, the structural models of each second plate are stretched along the thickness direction to obtain the normal vector directions of each plane. z The first complete plate model of the direction; The distance between each of the first complete plate models is used as the height constraint information, and the line edge information of each of the first complete plate models is used as the length constraint information. Based on the height constraint information and the length constraint information, the first plate structure model is stretched to obtain each plane normal vector direction is non-linear. z The second complete plate model in the direction; Spatial fusion is performed on each of the first complete plate models and each of the second complete plate models to obtain the initial workpiece geometric structure model; The geometric symmetry information is used to perform plate completion processing on the initial workpiece geometric structure model to obtain the welded workpiece structure model.

[0012] Optionally, determining the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the welded workpiece structural model includes: Based on the connection relationship between adjacent plates in the welded workpiece structure model, extract the boundary lines of each plate in the welded workpiece structure model; Based on the coordinate information of the boundary lines of each plate, the spatial location of the weld corresponding to the boundary lines of each plate is determined.

[0013] Secondly, this application provides a symmetrical workpiece weld positioning device based on single-camera structured light vision, comprising: The classification module is used to classify planar point cloud data using the original point cloud data of the welded workpiece, and obtain multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set; The partitioning module is used to partition the workpiece plate structure using each main plane point cloud set and its corresponding thickness surface point cloud set, and to determine multiple independent plate structure models. The modeling module is used to model the workpiece geometry based on the geometric topological constraints between the various plate structure models, and to obtain the welded workpiece structure model. The positioning module is used to determine the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the structural model of the welded workpiece.

[0014] Thirdly, this application provides an electronic device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute the method described in the first aspect or any possible implementation thereof.

[0015] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0016] Fifthly, this application provides a computer program product that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0017] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0018] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art: This application provides a method for locating weld seams in symmetrical workpieces based on single-camera structured light vision. By introducing a planar point cloud data classification method, the method performs plate structure analysis and three-dimensional structural modeling on symmetrical welded workpieces. Utilizing the geometric topological constraints and symmetrical structural features between the plates, the method infers the weld seam position in the workpiece from the constructed overall plate model. This method can achieve accurate and stable weld seam positioning even when the point cloud acquired by a single-camera structured light vision system is incomplete, improving the versatility and industrial applicability of the workpiece weld seam positioning method. Furthermore, it eliminates the need for pre-establishing workpiece templates, significantly reducing costs. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the symmetrical workpiece weld positioning method based on single-camera structured light vision provided in this application embodiment; Figure 2 This is a schematic diagram of the modeling effect of symmetrical workpiece weld positioning provided in the embodiments of this application; Figure 3 This is a schematic diagram of the symmetrical workpiece weld positioning device based on single-camera structured light vision provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0021] The terms "first" and "second," etc., used in the specification and claims of this application are used to distinguish different objects, not to describe a specific order of objects. For example, "first principal plane point cloud set" and "second principal plane point cloud set," etc., are used to distinguish different principal plane point cloud sets, not to describe a specific order of principal plane point cloud sets.

[0022] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0023] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.

[0024] The embodiments of this application are described below with reference to the accompanying drawings.

[0025] Figure 1 This is a flowchart illustrating the symmetrical workpiece weld positioning method based on single-camera structured light vision provided in this application embodiment, as shown below. Figure 1 As shown, it includes: Step S100: Use the original point cloud data of the welded workpiece to classify the planar point cloud data, and obtain multiple main planar point cloud sets and the thickness surface point cloud set corresponding to each main planar point cloud set; Step S200: Use the point cloud set of each main plane and its corresponding thickness surface point cloud set to divide the workpiece into plate structures and determine multiple independent plate structure models. Step S300: Model the workpiece geometry based on the geometric topological constraints between the various plate structure models to obtain the welded workpiece structure model; Step S400: Determine the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the structural model of the welded workpiece.

[0026] Specifically, the main plane point cloud set described in the embodiments of this application refers to the point cloud set composed of point cloud clusters that have a large proportion of points and continuous spatial distribution in the workpiece point cloud data.

[0027] The thickness surface point cloud set described in this application refers to the set of point clouds in the workpiece point cloud data that are adjacent to the main plane point cloud and whose normal vector direction is basically perpendicular to the main plane point cloud.

[0028] The geometric topological constraint relationship described in the embodiments of this application refers to the geometric topological structure and positional constraint relationship of each plate structure model used to constitute the welded workpiece in space. Specifically, it may include plate thickness information, height constraint information, length constraint information, and geometric symmetry information.

[0029] In an embodiment of this application, in step S100, a structured light vision sensor is first used to scan a welded workpiece with a symmetrical structure to obtain raw point cloud data for weld positioning. More specifically, this includes the following steps: Step 1: Fix the position of the structured light vision sensor, installing it diagonally above the workpiece to ensure complete visibility of the top surface of the workpiece. Construct a robot positioning system based on structured light vision, which can consist of two parts: a vision perception system and a robot welding system. The vision perception system acquires 3D point cloud data of the workpiece, while the robot welding system executes the welding task based on the information acquired by the vision perception system.

[0030] Step two involves fixing the calibration target to the robot's end flange and attaching the structured light vision sensor to the outside of the robot, forming a hand-eye calibration structure with the eye outside the hand. By controlling the robot to move in multiple different spatial poses, the calibration target is kept within the effective field of view of the structured light vision sensor. The robot pose must encompass both translational and rotational movements. Under different robot poses, the calibration target information captured by the structured light vision sensor is collected, and the corresponding robot flange pose data is recorded. Based on multiple sets of calibration target information and robot pose data, the extrinsic parameters of the structured light vision sensor relative to the robot's base coordinate system are calculated and saved to ensure coordinate consistency in subsequent point cloud data.

[0031] Step 3: After completing the hand-eye calibration, the welding workpiece is scanned by a structured light vision sensor to obtain the point cloud data of the welding workpiece within the visible range of the structured light vision sensor, and then converted to the robot base coordinate system to obtain the original point cloud data of the welding workpiece.

[0032] The method in this application embodiment only requires a single-position structured light vision sensor to achieve weld seam positioning, reducing hardware costs. Furthermore, since it is an eye-out-of-hand position determination mode, it reduces robot interference problems.

[0033] Furthermore, in the embodiments of this application, the acquired raw point cloud data is processed at multiple scales to achieve planar point cloud data classification. The raw point cloud data can be classified into multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set.

[0034] In the embodiments of this application, in step S200, the point cloud sets of each principal plane are classified according to the principal direction of the point cloud normal vector, and three-dimensional plate structure modeling is performed on the point cloud sets of each type of principal plane and their corresponding thickness surface point cloud sets to realize the plate structure division of the welded workpiece, and multiple independent plate structure models can be determined.

[0035] Here, it can be understood that the plate structure model is a three-dimensional spatial structure model.

[0036] Furthermore, in the embodiments of this application, in step S300, the geometric topological constraint relationship between each plate structure model is used, including workpiece plate thickness information, height constraint information, length constraint information and geometric symmetry information, to perform workpiece geometric structure modeling on the entire welded workpiece structure, including plate structure model stretching in different dimensional directions, missing plate completion and other processing, and finally a complete welded workpiece structure model can be obtained.

[0037] Furthermore, in the embodiments of this application, in step S400, weld seam identification and position analysis can be performed based on the connection relationship between adjacent plates in the welded workpiece structure model to locate the spatial position of each weld seam in the welded workpiece.

[0038] Based on the above embodiments, as an optional embodiment, step S400, determining the spatial position of each weld in the welded workpiece according to the connection relationship between adjacent plates in the welded workpiece structural model, includes: Based on the connection relationship between adjacent plates in the structural model of the welded workpiece, extract the boundary lines of each plate in the structural model of the welded workpiece. Based on the coordinate information of the boundary lines of each plate, the spatial location of the weld corresponding to the boundary lines of each plate is determined.

[0039] Specifically, in the embodiments of this application, the connection relationships between all adjacent plates in the corrected complete welded workpiece structure model can be analyzed by algorithms, and the geometric edges of the intersection of adjacent plates can be calculated and extracted. These edges are the preliminary theoretical weld position lines, thereby extracting the boundary lines of each plate in the welded workpiece structure model. Further, the three-dimensional spatial coordinate information of each plate boundary line is obtained, the spatial position of the weld corresponding to each plate boundary line is determined, and finally these boundary lines can be converted into continuous motion trajectories that can be executed by welding robots or automatic welding machines to realize automated robotic welding.

[0040] The method of this application embodiment uses a precise three-dimensional model to extract the boundary line between adjacent plates to locate the weld position, eliminating visual and human errors caused by manual teaching or measurement, ensuring the absolute accuracy and repeatability of the welding trajectory, which is beneficial to improving welding quality; at the same time, through the geometric spatial relationship between plates, it can also effectively avoid the problem of mislocating edges as welds in traditional methods.

[0041] In the embodiments of this application, multi-view point cloud stitching is not required, and weld seam positioning of symmetrical welded workpieces can be achieved under single-camera structured light vision conditions, which can reduce the system's requirements for equipment space layout.

[0042] The symmetrical workpiece weld seam positioning method based on single-camera structured light vision in this application introduces a planar point cloud data classification method to perform plate structure analysis and three-dimensional structure modeling of the welded workpiece. By utilizing the geometric topological constraints and symmetrical structural features between the plates, the weld seam position in the workpiece is inferred from the constructed overall plate model. This method can achieve accurate and stable weld seam positioning even when the point cloud acquired by a single-camera structured light vision system is incomplete. It does not require the pre-establishment of a workpiece template and can be applied to symmetrical welded workpieces, thus improving the versatility and industrial applicability of the workpiece weld seam positioning method.

[0043] Based on the above embodiments, as an optional embodiment, step S100 involves classifying the planar point cloud data using the original point cloud data of the welded workpiece to obtain multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set, including: Using raw point cloud data, bounding boxes are constructed and analyzed to determine the voxel size benchmark, thereby determining the voxel downsampling size, neighborhood search size, clustering space scale, and target analysis scale. The original point cloud data is downsampled according to the voxel downsampling size to obtain downsampled point cloud data. Based on the neighborhood search size, the local surface normal vector corresponding to each point in the downsampled point cloud data is determined. Classify the points based on the coordinate axis components of the local surface normal vector corresponding to each point, determine multiple point cloud subsets with the same normal vector direction, and perform cluster analysis on each point cloud subset according to the clustering space scale to determine multiple principal plane point cloud sets and multiple non-principal plane point cloud sets corresponding to each point cloud subset. Spatial geometric analysis is performed on each principal plane point cloud set and its corresponding non-principal plane point cloud set using the target analysis scale. The thickness surface point cloud set corresponding to each principal plane point cloud set is determined from the corresponding non-principal plane point cloud set.

[0044] Specifically, in the embodiments of this application, in the specific implementation of bounding box construction and analysis using raw point cloud data, it is first necessary to preprocess the acquired raw point cloud data to remove noise points and reduce the size of the point cloud data, so as to improve the stability and computational efficiency of subsequent analysis.

[0045] Specifically, by analyzing the distance distribution characteristics between each point and its neighboring points in the original point cloud data, outliers that are significantly inconsistent with the overall point cloud distribution can be identified. Points with a neighborhood distance greater than a set threshold are identified as noise points and removed from the point cloud.

[0046] More specifically, the acquired 3D raw point cloud dataset can be represented as: For each point Construct a set of neighborhood points, which can be represented as ,in, Indicates the preset neighborhood search radius. This represents the Euclidean distance. It calculates the average distance from this point to its neighborhood points. This process can be represented as: ; By analyzing the distance distribution characteristics of all points, the overall average distance is calculated: ; Set anomaly detection threshold When satisfied When the point is identified as a noise point, it is removed from the original point cloud data to obtain the denoised point cloud data.

[0047] Furthermore, the 3D bounding boxes of the denoised point cloud data are obtained, and their spatial diagonal lengths are calculated. Based on preset partitioning parameters, the spatial diagonal length is divided by the partitioning parameters to obtain the voxel size reference, which serves as the size reference for subsequent processing. Here, the partitioning parameters can be modified according to specific computer performance and desired processing speed.

[0048] Specifically, the extreme values ​​of the denoised point cloud data in three spatial coordinate directions are obtained respectively. This leads to the construction of a 3D bounding box, whose diagonal length... L It can be represented as: ; Based on the preset division parameters, the diagonal length of this space is determined. L The voxel size reference can be calculated by dividing by the division parameter.

[0049] Furthermore, the voxel downsampling size is obtained by multiplying the voxel size reference obtained above by the first processing coefficient. The point cloud space is divided into a regular three-dimensional cubic voxel grid, and the geometric center of each point within a voxel represents that voxel, thereby effectively reducing the amount of point cloud data while maintaining the overall geometric shape.

[0050] The final voxel downsampling size is obtained by multiplying the voxel size reference obtained above by the first processing coefficient. For a set of points falling within the same voxel, it can be represented as: ; Calculate the geometric center of all points within the voxel. As a representative point, it can be represented as: By replacing all points within the voxel with this geometric center, the original point cloud data is downsampled to obtain the downsampled point cloud data.

[0051] Further, in the embodiments of this application, the neighborhood search size for normal vector estimation is obtained by multiplying the aforementioned voxel size reference by a second processing coefficient. Based on the downsampled point cloud data, the local surface normal vector of each point is calculated. Specifically, local plane fitting is performed on the neighborhood points within the size according to the neighborhood search size, and the covariance matrix of the neighborhood points is calculated. This process can be expressed as: ; In the formula It represents the geometric center of the neighborhood points.

[0052] Then, eigenvalue decomposition is performed on the covariance matrix, and the eigenvector corresponding to the smallest eigenvalue is selected as the local surface normal vector at that point. By using the above processing method, the local surface normal vector corresponding to each point in the downsampled point cloud data can be obtained.

[0053] Here, it can be understood that the local surface normal vector refers to the direction of the normal to a small region near a certain point on the surface of a three-dimensional object, which reflects the local geometric features of the object's surface at that point.

[0054] Further, in the embodiments of this application, the point cloud subsets with consistent normal vector directions are classified according to the coordinate axis components of the local surface normal vector corresponding to each point. Then, cluster analysis is performed on each point cloud subset based on the clustering space scale to determine multiple principal plane point cloud sets and multiple non-principal plane point cloud sets corresponding to each point cloud subset. Specifically, the local surface normal vector corresponding to each point is calculated in... x , y , z The absolute component values ​​of the coordinate axes are used to perform cluster analysis based on the coordinate axis direction corresponding to the maximum absolute component value, dividing the downsampled point cloud data into multiple point cloud subsets with consistent directions.

[0055] Specifically, the absolute component values ​​of each local surface normal vector along the three coordinate axes are calculated. This is used to distinguish structural surfaces with different spatial orientations.

[0056] Determine the largest absolute component: Then, based on the coordinate axis direction corresponding to the maximum absolute component value, the point cloud is divided into different subsets. When The corresponding points are assigned to the main direction of the normal vector. x A subset of point clouds in terms of direction; when The corresponding points are assigned to the main direction of the normal vector. y A subset of point clouds in terms of direction; when The corresponding points are assigned to the main direction of the normal vector. z A subset of point clouds representing directions.

[0057] Through the above processing, multiple subsets of point clouds with consistent normal vector directions are formed, thereby achieving preliminary separation of structural surfaces with different spatial orientations, laying the foundation for subsequent principal plane recognition and weld feature extraction.

[0058] Next, the voxel size benchmark obtained above is multiplied by the third processing coefficient to obtain the clustering spatial scale. Based on this clustering spatial scale, clustering analysis is performed on the various point cloud subsets obtained above to obtain multiple point cloud clusters. Then, point cloud regions with larger areas and continuous distribution can be selected from each point cloud cluster as principal planes, and the remaining point clouds are designated as non-principal plane point clouds, and indexes are established for each.

[0059] Within each subset of point clouds corresponding to each direction, clustering can be performed based on spatial adjacency, grouping points whose spatial distance is less than the clustering spatial scale parameter into the same cluster, thus obtaining multiple point cloud clusters, which can be represented as follows: , This indicates the number of point cloud clusters.

[0060] For each point cloud cluster, the proportion of points within it and its spatial continuity are calculated. Point cloud clusters with a larger proportion and continuous spatial distribution are defined as the principal plane point cloud set, while the remaining point cloud clusters are defined as non-principal plane point cloud sets. Furthermore, index structures are established for each cluster to facilitate subsequent edge region analysis and rapid localization. This allows for the identification of multiple principal plane point cloud sets and multiple non-principal plane point cloud sets corresponding to each point cloud subset.

[0061] Furthermore, in the embodiments of this application, spatial geometric analysis is performed on each principal plane point cloud set and its corresponding non-principal plane point cloud set using a target analysis scale. The thickness surface point cloud set corresponding to each principal plane point cloud set is determined from the corresponding non-principal plane point cloud set. Specifically, the target analysis scale can be obtained by multiplying the aforementioned voxel size reference by a fourth processing coefficient. Then, the edge region of the principal plane formed by the principal plane point cloud sets can be locally subdivided according to the target analysis scale. Through spatial position constraints and size constraints, the thickness surface point cloud adjacent to the principal plane and whose normal vector direction is substantially perpendicular to the principal plane is identified.

[0062] Here, it can be understood that the thickness surface point cloud refers to the point cloud data in the thickness direction of the workpiece plate.

[0063] Specifically, spatial geometric analysis is performed on each principal plane point cloud set and its corresponding non-principal plane point cloud set using the target analysis scale. First, the principal plane point cloud regions are identified to determine the principal plane point cloud. For each point in the non-principal plane point cloud, the angle between its normal vector and the normal vector of a point in the principal plane point cloud is calculated. .

[0064] A point cloud can be classified as a thickness surface point cloud when the following conditions are met: the spatial distance from the point cloud on the principal plane is less than the target analysis scale, and the direction of the normal vector is basically perpendicular to the normal vector of the point cloud on the principal plane, i.e., the angle tolerance parameter is met. .

[0065] Using the above processing method, the thickness surface point cloud set corresponding to each principal plane point cloud set can be determined from the corresponding non-principal plane point cloud set.

[0066] It should be noted that the first processing coefficient, the second processing coefficient, the third processing coefficient and the fourth processing coefficient mentioned in the embodiments of this application are all empirical parameters preset by the welded workpiece. They can be set according to different point cloud data, workpiece and its plate thickness. Manual setting is supported. By setting different processing coefficients, point cloud information processing at multiple scales can be achieved.

[0067] The method of this application embodiment, under the condition that the point cloud obtained by single-camera structured light vision is incomplete, determines the main plane point cloud set and its corresponding thickness surface point cloud set by using multi-scale point cloud information processing on the basis of establishing a voxel size reference. The main plane and thickness surface of the workpiece are identified for subsequent assembly of plate units. The complete welding workpiece structure is completed through symmetry characteristics, which is conducive to realizing the three-dimensional structural understanding from local point cloud to overall structure, and breaks through the high dependence of existing multi-plane fitting methods on the integrity of point cloud.

[0068] Based on the above embodiments, as an optional embodiment, the principal plane point cloud set includes normal vectors whose principal directions are non-normal. z The first principal plane point cloud set and the principal direction of the normal vector are: z The second principal plane point cloud set in the direction; Step S200, using each principal plane point cloud set and its corresponding thickness surface point cloud set to divide the workpiece plate structure, determining multiple independent plate structure models, including: Three-dimensional plate structure modeling is performed using each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models. Using each set of point clouds on the second principal plane, a 3D plate structure model is created, resulting in multiple independent second plate structure models. Based on multiple first-plate structure models and multiple second-plate structure models, multiple independent plate structure models are obtained.

[0069] Specifically, in the embodiments of this application, the principal plane point cloud set includes points whose principal direction of normal vector is non-normal. z The first principal plane point cloud set and the principal direction of the normal vector are: z The set of point clouds in the second principal plane of the direction, here, is described zThe direction is the same as in the aforementioned point cloud subset partitioning implementation method. z They are in the same direction.

[0070] Furthermore, in the embodiments of this application, each normal vector's principal direction is non- z By modeling the 3D plate structure using the point cloud set of the first principal plane and its corresponding thickness surface point cloud set, multiple independent first plate structure models can be obtained. Specifically, through spatial adjacency relationship and normal vector direction analysis, the point cloud is divided into multiple independent plate structure models. Each plate structure includes a principal plane and its associated thickness surface.

[0071] Based on the above embodiments, as an optional embodiment, three-dimensional plate structure modeling is performed using each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models, including: For any set of first principal plane point clouds in each set of first principal plane point clouds, the normal vector direction of the principal plane formed by any set of first principal plane point clouds is taken as the first principal direction of the orthogonal coordinate system, and the normal vector direction of the thickness surface formed by the corresponding set of thickness surface point clouds is taken as the second principal direction of the orthogonal coordinate system. The third principal direction of the orthogonal coordinate system is determined based on the first and second principal directions; Based on the maximum and minimum values ​​of the point cloud coordinates in the first principal direction, the second principal direction, and the third principal direction, three-dimensional modeling is performed on each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models.

[0072] Specifically, in the embodiments of this application, a cubic structure is fitted for the principal plane point cloud set whose principal normal direction is not in the z-direction, i.e., the first principal plane point cloud set. The normal direction of the principal plane formed by the first principal plane point cloud set is taken as the first principal direction of the orthogonal coordinate system, and the normal vector of the corresponding thickness surface is taken as the second principal direction. A third principal direction is determined using the first and second principal directions according to the left-hand rule. In this local coordinate system, the maximum and minimum values ​​of all points in each of the first principal plane point cloud sets and their corresponding thickness surface point cloud sets in the three directions are recorded respectively. This constructs a three-dimensional enclosing structure for the corresponding plate, resulting in multiple independent first plate structure models.

[0073] More specifically, assume the normal vector of the principal plane of the plate is First main direction ; Corresponding thickness surface normal vector Second main direction Determine the third principal direction based on the left-hand rule. This allows us to establish a local orthogonal coordinate system for the plate, which can be represented as follows: .

[0074] Furthermore, all point clouds within the plate, including the point clouds in each first principal plane point cloud set and its corresponding thickness surface point cloud set, are projected onto this local coordinate system. This process can be represented as: ; In the formula, This represents the three-dimensional coordinates of the point cloud within the plate in the global coordinate system. This represents the coordinate components in the local coordinate system.

[0075] The maximum and minimum values ​​of point cloud coordinates in the first, second, and third principal directions are recorded respectively. A 3D model is then created for each first principal plane point cloud set and its corresponding thickness surface point cloud set, establishing a cube-shaped enclosure structure to represent the spatial geometric extent of the corresponding plate. This allows for 3D modeling of point cloud data, resulting in multiple independent first plate structure models.

[0076] The method in this embodiment is applicable when the principal direction of the normal vector is non-linear. z The principal plane point cloud set of the direction, by using the plane normal vector formed by the principal plane point cloud set and its corresponding thickness surface point cloud set to construct an orthogonal coordinate system, can more accurately describe the geometric features of the welded workpiece surface and improve the accuracy of the 3D modeling of the workpiece plate.

[0077] Furthermore, in the embodiments of this application, the principal direction of each normal vector is used as... z By using the point cloud set of the second principal plane in the direction to model the three-dimensional plate structure, multiple independent second plate structure models can be obtained.

[0078] Based on the above embodiments, as an optional embodiment, three-dimensional plate structure modeling is performed using each second principal plane point cloud set to obtain multiple independent second plate structure models, including: Project the point cloud data in each second principal plane point cloud set to xy The plane is used to obtain the two-dimensional projection point set corresponding to each second principal plane point cloud set; Extract the outer contour of each plate from each two-dimensional projection point set to obtain the outer contour graphics of each plate; Extract the line edge information of the outer contour graphics of each plate, and map the line edge information back to the three-dimensional space and the corresponding thickness surface point cloud set to perform three-dimensional plate structure modeling, resulting in multiple independent second plate structure models.

[0079] Specifically, in the embodiments of this application, for the point cloud set of the second principal plane with the main direction of the normal vector in the z direction, a two-dimensional projection method is used to identify and model the plate structure.

[0080] More specifically, the point cloud data in each second principal plane point cloud set is projected onto... xy The planar data is used to obtain the two-dimensional projection point set corresponding to each second principal plane point cloud set. The outer contour of each plate is extracted from each two-dimensional projection point set, resulting in the outer contour graphics of each plate. Image processing methods are used to extract the edge contours of the plates from these two-dimensional projection images, including but not limited to edge detection and connected component analysis. Specifically, a polar coordinate system is established with the geometric center of each plate's outer contour graphics as the origin. By sorting edge points according to angles and smoothing the radius sequence, stable polyline edge information can be obtained, thereby reducing the impact of discrete noise on edge morphology.

[0081] Furthermore, these line edge information are mapped back to three-dimensional space according to the z coordinate, and spatially fused with the thickness surface point cloud set corresponding to the second principal plane point cloud set to improve the three-dimensional structure of the z-direction plate, resulting in multiple independent second plate structure models.

[0082] The method in this application embodiment, for the point cloud set of the second principal plane with the main direction of the normal vector in the z direction, takes into account the spatial distribution geometric characteristics of the corresponding plate and uses the image processing method of three-dimensional spatial point cloud data projection to perform data dimensionality reduction processing and plate three-dimensional structure modeling. This can improve the accuracy of three-dimensional modeling of workpiece plates, while reducing computational complexity and improving modeling efficiency.

[0083] Furthermore, in the embodiments of this application, multiple independent plate structure models are constituted by multiple first plate structure models and multiple second plate structure models.

[0084] The method of this application embodiment utilizes the geometric information and symmetrical structural features of the plates in the welded workpiece to perform targeted processing and plate structure modeling on the principal plane point cloud set with different principal directions of normal vectors. This enables a structured understanding of the welded workpiece plates, which is beneficial to improving the stability of weld positioning. Since it does not depend on the model data of a specific workpiece, it can provide an efficient and economical implementation method for welding robot welding tasks.

[0085] Based on the above embodiments, as an optional embodiment, the geometric topological constraint relationship includes plate thickness information, height constraint information, length constraint information, and geometric symmetry information; step S300, based on the geometric topological constraint relationship between each plate structure model, the workpiece geometric structure is modeled to obtain the welded workpiece structure model, including: Based on the plate thickness information, the structural models of each second plate are stretched along the thickness direction to obtain the directions of the normal vectors of each plane. z The first complete plate model of the direction; The distance between each first complete plate model is used as the height constraint information, and the line edge information of each first complete plate model is used as the length constraint information. Based on the height and length constraints, the structural models of each first plate are stretched to obtain the non-linear normal vectors of each plane. z The second complete plate model in the direction; Spatial fusion of each first complete plate model and each second complete plate model yields the initial workpiece geometric structure model. By using geometric symmetry information to perform plate-by-plate completion processing on the initial workpiece geometric structure model, a welded workpiece structure model is obtained.

[0086] Specifically, in the embodiments of this application, an overall geometric structure model of the welded workpiece is constructed based on the spatial positional relationship and geometric constraint relationship between the workpiece plate structures.

[0087] More specifically, for a plate whose planar normal vector direction is the z-direction, using the closed polyline edge information obtained in the aforementioned embodiment, and based on the plate thickness information, each second plate structure model can be stretched along the thickness direction to form the corresponding complete plate, thereby obtaining the plane normal vector direction of each plate. z The first complete plate model of the direction.

[0088] In the embodiments of this application, for a plate whose plane normal vector direction is not the z-direction, the first complete plate model whose plane normal vector direction is the z-direction is used as a spatial geometric constraint. Specifically, the distance between each first complete plate model can be used as height constraint information, and the line edge information of each first complete plate model can be used as length constraint information to use the z-direction plate as a spatial geometric constraint.

[0089] Among them, the height constraint is as follows: Let the distance between the two first complete plate models in space be H, and let this distance be used as the plane normal vector direction, which is the non-z direction (including the z-direction). x direction or y (Direction) Plate height constraint information.

[0090] Length constraint: Using the line edge contour information of the first complete plate model, determine the projection range of the non-z-direction plate in the plane direction as the plane normal vector direction.

[0091] Furthermore, based on the height and length constraints determined above, the structural models of each first plate can be stretched to obtain the plane normal vectors with non-linear directions. z The second complete plate model in the direction.

[0092] Furthermore, in the embodiments of this application, the first complete plate models and the second complete plate models are spatially fused to obtain an initial workpiece geometric structure model. Specifically, according to x , y , z The spatial overlap between the complete plate models in the direction divides the overall model into multiple spatial sub-regions. Through geometric symmetry analysis, the smallest repeating unit in the structure is extracted and finally combined to obtain the initial workpiece geometric structure model.

[0093] Furthermore, in the embodiments of this application, geometric symmetry information is used to perform overall segment completion processing on the initial workpiece geometric structure model. Specifically, after the initial workpiece geometric structure model is constructed, based on the spatial constraint relationship between the segments, the missing point cloud areas caused by single-camera acquisition of structured light vision are structurally completed, thereby obtaining a complete welded workpiece structure model.

[0094] In some embodiments, specific implementation methods for block completion include: The first method outputs the plate thickness parameters corresponding to the identified thickness surfaces and uses this information to determine whether the plate thicknesses in the initial workpiece geometry model match the actual welded workpiece structure. If there is an error in the plate thickness parameters, the actual thickness parameters are input. Based on the modified thickness parameters, the thickness of the corresponding plate is updated, and the associated geometric relationships are adjusted synchronously, thereby updating the initial workpiece geometry model.

[0095] The second method involves selecting whether to perform symmetry operations on the initial workpiece geometry model. x If the orientation is symmetrical, then the initial workpiece geometric model will be distributed in... y The direction of the plate along x The directional blocks are symmetrically completed; if selected... y If the orientation is symmetrical, then the initial workpiece geometric model will be distributed in... x The direction of the plate along y The directional blocks are symmetrically completed; if selected... xy If the directions are symmetrical, then it will proceed first. x Perform symmetrical completion in the directional direction, and then perform symmetrical completion in the y-direction.

[0096] The method in this application transforms the multi-plane fitting intersection problem into a connection problem between plates, changing the traditional "point-to-surface-to-line" approach to a "point-to-plate-to-geometric structure" approach. This deepens the structural understanding of the entire symmetrical welded workpiece and can further improve the stability and accuracy of the workpiece weld positioning.

[0097] Figure 2This is a schematic diagram illustrating the modeling effect of symmetrical workpiece weld positioning provided in an embodiment of this application, such as... Figure 2 As shown in the embodiments of this application, different colors represent point clouds of workpiece plates at different locations, and their processed schematic diagrams. Specific implementation methods for symmetrical workpiece weld seam positioning include: Step S1, Point Cloud Acquisition. A structured light vision sensor is used to scan a welded workpiece with a symmetrical structure to acquire raw point cloud data for weld seam positioning.

[0098] Step S2, Point Cloud Preprocessing. The acquired raw point cloud data is preprocessed to remove noise points and reduce the data size, thereby improving the stability and computational efficiency of subsequent analysis. By setting different processing coefficients, point cloud information processing at multiple scales is achieved, resulting in multiple principal plane point cloud sets and a thickness surface point cloud set corresponding to each principal plane point cloud set.

[0099] Step S3, Plate Segmentation. Based on the classified point cloud data of the main plane and thickness surface, the plate structure of the welded workpiece is segmented to obtain multiple independent plate structure models.

[0100] Step S4: Geometric structure model construction and plate completion. Based on the spatial positional relationships and geometric topological constraints between the plate structures, an initial workpiece geometric structure model is constructed, and plate completion processing is performed to construct the overall plate geometric structure model of the welded workpiece, thus obtaining the welded workpiece structure model.

[0101] Step S5, weld seam positioning. Based on the connection relationship between adjacent plates in the welded workpiece structural model, determine the spatial position of each weld seam in the welded workpiece.

[0102] Here, the specific implementation methods of the above steps can be referred to the methods of the foregoing embodiments, and this application will not repeat them.

[0103] The following describes the symmetrical workpiece weld positioning device based on single-camera structured light vision provided in this application. The symmetrical workpiece weld positioning device based on single-camera structured light vision described below can be referred to in correspondence with the symmetrical workpiece weld positioning method based on single-camera structured light vision described above.

[0104] Figure 3 This is a schematic diagram of the symmetrical workpiece weld positioning device based on single-camera structured light vision provided in the embodiments of this application, as shown below. Figure 3 As shown, it includes: The classification module 10 is used to classify planar point cloud data using the original point cloud data of the welded workpiece, and obtain multiple main planar point cloud sets and the thickness surface point cloud set corresponding to each main planar point cloud set; The partitioning module 20 is used to partition the workpiece plate structure using each main plane point cloud set and its corresponding thickness surface point cloud set, and to determine multiple independent plate structure models. Modeling module 30 is used to model the workpiece geometry based on the geometric topological constraints between the various plate structure models, and obtain the welded workpiece structure model. The positioning module 40 is used to determine the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the structural model of the welded workpiece.

[0105] It is understood that the detailed functional implementation of each of the above units / modules can be found in the description in the aforementioned method embodiments, and will not be repeated here.

[0106] It should be understood that the above-described device is used to execute the methods in the above embodiments. The implementation principle and technical effect of the corresponding program modules in the device are similar to those described in the above methods. The working process of the device can be referred to the corresponding process in the above methods, and will not be repeated here.

[0107] The symmetrical workpiece weld seam positioning device based on single-camera structured light vision in this application embodiment performs plate structure analysis and three-dimensional structure modeling of symmetrical welded workpieces by introducing a planar point cloud data classification method. By utilizing the geometric topological constraints and symmetrical structural features between the plates, the weld seam position in the workpiece is inferred from the constructed overall plate model. This can achieve accurate and stable weld seam positioning even when the point cloud acquired by a single-camera structured light vision device is incomplete, improving the versatility and industrial applicability of the workpiece weld seam positioning method. Furthermore, it eliminates the need to pre-establish a workpiece template, greatly saving cost.

[0108] Based on the methods in the above embodiments, this application provides an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute the methods in the above embodiments.

[0109] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0110] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0111] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0112] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0113] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0114] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0115] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.

[0116] It should be understood that expressions such as “comprising” and “may include” used in this application indicate the existence of the disclosed functions, operations, or constituent elements, and do not limit one or more additional functions, operations, and constituent elements. In this application, terms such as “comprising” and / or “having” are to be interpreted as indicating a particular characteristic, number, operation, constituent element, component, or combination thereof, but not to exclude the existence or possibility of adding one or more other characteristics, numbers, operations, constituent elements, components, or combinations thereof.

[0117] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for locating weld seams in symmetrical workpieces based on single-camera structured light vision, characterized in that, include: The original point cloud data of the welded workpiece is used to classify the planar point cloud data, resulting in multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set; The workpiece plate structure is divided using each main plane point cloud set and its corresponding thickness surface point cloud set to determine multiple independent plate structure models. Based on the geometric and topological constraints between the various plate structure models, the workpiece geometry is modeled to obtain the welded workpiece structure model. Based on the connection relationship between adjacent plates in the structural model of the welded workpiece, the spatial position of each weld in the welded workpiece is determined.

2. The method for locating weld seams in symmetrical workpieces according to claim 1, characterized in that, The process of classifying planar point cloud data using the original point cloud data of the welded workpiece to obtain multiple principal planar point cloud sets and a thickness surface point cloud set corresponding to each principal planar point cloud set includes: The original point cloud data is used to construct and analyze bounding boxes to determine the voxel size benchmark, thereby determining the voxel downsampling size, neighborhood search size, clustering space scale, and target analysis scale. The original point cloud data is downsampled according to the voxel downsampling size to obtain downsampled point cloud data, and the local surface normal vector corresponding to each point in the downsampled point cloud data is determined based on the neighborhood search size. Classify the points based on the coordinate axis components of the local surface normal vector corresponding to each point, determine multiple point cloud subsets with the same normal vector direction, and perform cluster analysis on each point cloud subset according to the clustering space scale to determine multiple principal plane point cloud sets and multiple non-principal plane point cloud sets corresponding to each point cloud subset. Spatial geometric analysis is performed on each principal plane point cloud set and its corresponding non-principal plane point cloud set using the target analysis scale, and the thickness surface point cloud set corresponding to each principal plane point cloud set is determined from the corresponding non-principal plane point cloud set.

3. The method for locating weld seams in symmetrical workpieces according to claim 1, characterized in that, The principal plane point cloud set includes normal vectors whose principal directions are non-normal. z The first principal plane point cloud set and the principal direction of the normal vector are: z The set of point clouds in the second principal plane of the direction; The process of dividing the workpiece into sections using each principal plane point cloud set and its corresponding thickness surface point cloud set, thereby determining multiple independent section structure models, includes: Three-dimensional plate structure modeling is performed using each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models. Using each set of point clouds of the second principal plane, a three-dimensional plate structure model is performed to obtain multiple independent second plate structure models; Based on multiple first plate structure models and multiple second plate structure models, multiple independent plate structure models are obtained.

4. The method for positioning weld seams in symmetrical workpieces according to claim 3, characterized in that, The process of using each first principal plane point cloud set and its corresponding thickness surface point cloud set to perform 3D plate structure modeling results in multiple independent first plate structure models, including: For any one of the first principal plane point cloud sets in each of the first principal plane point cloud sets, the normal vector direction of the principal plane formed by the first principal plane point cloud sets is taken as the first principal direction of the orthogonal coordinate system, and the normal vector direction of the thickness surface formed by the corresponding thickness surface point cloud sets is taken as the second principal direction of the orthogonal coordinate system. The third principal direction of the orthogonal coordinate system is determined based on the first principal direction and the second principal direction; Based on the maximum and minimum values ​​of the point cloud coordinates in the first principal direction, the second principal direction, and the third principal direction, three-dimensional modeling is performed on each first principal plane point cloud set and its corresponding thickness surface point cloud set to obtain multiple independent first plate structure models.

5. The method for positioning weld seams in symmetrical workpieces according to claim 3, characterized in that, The process of using each second principal plane point cloud set to model a 3D plate structure yields multiple independent second plate structure models, including: Project the point cloud data in each of the second principal plane point cloud sets to xy The plane is used to obtain the two-dimensional projection point set corresponding to each point cloud set of the second principal plane; Extract the outer contour of each of the two-dimensional projection point sets to obtain the outer contour graphics of each plate; Extract the line edge information of the outer contour graphics of each plate, and map the line edge information back to the three-dimensional space and the corresponding thickness surface point cloud set to perform three-dimensional plate structure modeling, thereby obtaining multiple independent second plate structure models.

6. The method for positioning weld seams in symmetrical workpieces according to claim 3, characterized in that, The geometric topological constraints include plate thickness information, height constraint information, length constraint information, and geometric symmetry information; the process of modeling the workpiece geometry based on the geometric topological constraints between the various plate structure models to obtain the welded workpiece structure model includes: Based on the plate thickness information, the structural models of each second plate are stretched along the thickness direction to obtain the normal vector directions of each plane. z The first complete plate model of the direction; The distance between each of the first complete plate models is used as the height constraint information, and the line edge information of each of the first complete plate models is used as the length constraint information. Based on the height constraint information and the length constraint information, the first plate structure model is stretched to obtain each plane normal vector direction is non-linear. z The second complete plate model in the direction; Spatial fusion is performed on each of the first complete plate models and each of the second complete plate models to obtain the initial workpiece geometric structure model; The geometric symmetry information is used to perform plate completion processing on the initial workpiece geometric structure model to obtain the welded workpiece structure model.

7. The method for positioning weld seams in symmetrical workpieces according to any one of claims 1-6, characterized in that, The step of determining the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the structural model of the welded workpiece includes: Based on the connection relationship between adjacent plates in the welded workpiece structure model, extract the boundary lines of each plate in the welded workpiece structure model; Based on the coordinate information of the boundary lines of each plate, the spatial location of the weld corresponding to the boundary lines of each plate is determined.

8. A symmetrical workpiece weld positioning device based on single-camera structured light vision, characterized in that, include: The classification module is used to classify planar point cloud data using the original point cloud data of the welded workpiece, and obtain multiple main planar point cloud sets and a thickness surface point cloud set corresponding to each main planar point cloud set; The partitioning module is used to partition the workpiece plate structure using each main plane point cloud set and its corresponding thickness surface point cloud set, and to determine multiple independent plate structure models. The modeling module is used to model the workpiece geometry based on the geometric topological constraints between the various plate structure models, and to obtain the welded workpiece structure model. The positioning module is used to determine the spatial position of each weld in the welded workpiece based on the connection relationship between adjacent plates in the structural model of the welded workpiece.

9. An electronic device, characterized in that, Includes memory and one or more processors; The memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions; The one or more processors invoke the computer instructions to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-7.

Images