Three-dimensional sleeve method for laser processing curved surface workpiece

Abstract

The present invention discloses a three-dimensional nesting method for laser-processed curved workpieces and proposes a solution based on robot end-point three-dimensional camera scanning and LSCM mapping. This solution not only solves the problems of manual dependence and efficiency in traditional methods, but also ensures the accuracy of complex surface processing through an innovative nesting algorithm. It has significant technical advantages and application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of laser processing, in particular to a three-dimensional nesting method for curved workpieces in laser processing. BACKGROUND

[0002] In the scene of laser processing robots, in order to realize laser processing operations such as etching and cleaning on curved workpieces, the workpiece model needs to be obtained first in the processing system, and the processing area is determined and the processing path is generated through the model.

[0003] Therefore, the operator needs to paste multiple marker points on the workpiece surface, and then performs three-dimensional reconstruction on the workpiece by manually using a handheld three-dimensional scanner. Finally, the three-dimensional model containing the curved surface normal vector information is imported into the processing system. In this process, the three-dimensional model of the workpiece needs to be unified with the robot base coordinate system by means of a calibration system. After the coordinate system is unified, the user usually needs to use a marker pen or marker points to mark the processing area on the workpiece surface, which cannot be conveniently planned, and the processing effect cannot be previewed. At the same time, the existing three-dimensional nesting scheme often uses projection method or unfolding method, which cannot effectively process complex curved surfaces, resulting in serious deformation of the processing pattern and difficulty in ensuring the precision, and therefore needs to be improved. SUMMARY

[0004] In view of the deficiencies in the prior art, the present application aims to provide a three-dimensional nesting method for curved workpieces in laser processing, which can realize laser processing preview and obtain high-precision processing patterns to adapt to complex curved surface high-precision processing.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0006] A three-dimensional nesting method for curved workpieces in laser processing, comprising the following steps:

[0007] S1, connecting a three-dimensional depth camera with a robot, and moving the three-dimensional depth camera through the robot to complete point cloud acquisition of the curved workpiece;

[0008] S2, preprocessing the point cloud of the curved workpiece;

[0009] S3, calculating the normal vector of each point in the point cloud and performing Poisson reconstruction to obtain a triangular facet mesh;

[0010] S4, performing plane parameterization on the triangular facet mesh, mapping the triangular facet mesh to a plane parameter domain using LSCM to obtain plane parameter domain vertices;

[0011] S5, obtaining a parameter plane mapping texture effect map and a triangular mesh curved surface texture mapping effect map according to the plane parameter domain vertices;

[0012] S6, obtaining the graphic processing coordinates according to the plane parameter domain vertexes, forming a processing path, and making the robot complete the processing action according to the processing path.

[0013] As a further improvement of the present application, the point cloud collection corresponding to step S1 comprises the following steps:

[0014] S11, controlling the robot to move to different poses, collecting the calibration plate images at each pose, and recording the robot end pose matrix;

[0015] S12, calculating the transformation matrix between the camera coordinate system and the robot coordinate system;

[0016] S13, converting the point cloud in the camera coordinate system observed by the three-dimensional depth camera to the robot end coordinate system through the transformation matrix;

[0017] S14, controlling the robot to move to each sampling position according to the preset path, recording the robot end pose matrix of each position, and collecting the curved workpiece point cloud data at each position through the three-dimensional depth camera;

[0018] S15, using the transformation matrix and the robot pose information to calculate the complete coordinate transformation matrix, and performing coordinate unified transformation on all point cloud data to convert all point clouds to the robot base coordinate system.

[0019] As a further improvement of the present application, the preprocessing in step S2 comprises the following steps:

[0020] S21, clipping the point cloud;

[0021] S22, down-sampling the point cloud.

[0022] As a further improvement of the present application, the point cloud clipping in step S21 is a quadrilateral.

[0023] As a further improvement of the present application, the normal vector calculation in step S3 comprises the following steps:

[0024] S311, searching for K1 nearest neighbor points of each point in the point cloud;

[0025] S312, performing eigenvalue decomposition on the covariance matrix by calculating the covariance matrix, the covariance matrix being:

[0026]

[0027] wherein n is the number of points in the neighborhood, p i is the coordinate vector (x, y, z) of the i-th point in the neighborhood, and μ is the centroid of the neighborhood point set;

[0028] The eigenvector corresponding to the minimum eigenvalue is the normal vector.

[0029] As a further improvement of the present application, the normal vector obtained in S3 is smoothed.

[0030] As a further improvement of the present application, the smoothing process comprises the following steps:

[0031] S321, for each point in the point cloud, find its K2 nearest neighbors;

[0032] S322, calculate the average normal vector of the nearest neighbors, which is the smoothed normal vector for this point.

[0033] As a further improvement of the present application, the plane parameterization in step S4 comprises the following steps:

[0034] S41, extract the sequence of boundary vertices of the mesh model, and form a closed loop by connecting the boundary vertices in order;

[0035] S42, let the total number of boundary vertices be n, and associate each boundary vertex to a parameter value on the interval [0, 1] by normalizing the parameter t = i / (n-1) (i is the vertex number, i∈[0,n-1]);

[0036] S43, map the boundary vertices to the four edges of the unit square based on the value of the parameter t.

[0037] As a further improvement of the present application, the mapping rule corresponding to step S43 is:

[0038] When t∈[0,0.25), the vertex is mapped to the lower edge of the square with coordinates (0,4t);

[0039] When t∈[0.25,0.5), the vertex is mapped to the right edge of the square with coordinates (4(t-0.25),1);

[0040] When t∈[0.5,0.75), the vertex is mapped to the upper edge of the square with coordinates (1,1-4(t-0.5));

[0041] When t∈[0.75,1], the vertex is mapped to the left edge of the square with coordinates (1-4(t-0.75),0).

[0042] The beneficial effects of the present application are:

[0043] 1. A solution for robot end three-dimensional camera scanning and LSCM mapping, which realizes high-precision machining of curved surface machining patterns in cooperation with a nesting algorithm;

[0044] 2. The three-dimensional depth camera is connected with the robot, the workpiece surface point cloud can be obtained automatically and quickly, so that the mark points are not needed to be pasted on the workpiece surface, the laser processing area can be planned, the processing pattern is mapped on the three-dimensional curved surface, and the user can preview the laser processing effect in advance. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a whole flowchart of the application;

[0046] Figure 2 It is a point cloud collection flowchart of the application;

[0047] Figure 3 It is a schematic diagram of the application that the robot cooperates with the three-dimensional depth camera to obtain the workpiece point cloud;

[0048] Figure 4 It is a schematic diagram of the point cloud before cutting of the application;

[0049] Figure 5 It is a schematic diagram of the point cloud after cutting of the application;

[0050] Figure 6 It is a Poisson reconstruction grid model diagram of the application;

[0051] Figure 7 It is a grid plane parameterization schematic diagram of the application;

[0052] Figure 8 It is a parameter plane mapping texture effect diagram of the application;

[0053] Figure 9 It is a triangular mesh curved surface texture mapping effect diagram of the application;

[0054] Figure 10 It is a graphic nesting rendering effect schematic diagram of the application based on LSCM. DETAILED DESCRIPTION

[0055] The application will be further described in detail below in combination with the drawings and embodiments. Same parts are denoted by same reference numerals.

[0056] As shown in the drawings, Figures 1-10 A three-dimensional nesting method for laser processing curved workpiece, which is used for high-precision processing of the curved workpiece pattern surface, comprises the following steps.

[0057] S1, connect the three-dimensional depth camera with the robot, and complete the curved workpiece point cloud collection by moving the three-dimensional depth camera through the robot.

[0058] The point cloud collection corresponding to step S1 comprises the following steps.

[0059] S11, control the robot to move to different poses, capture the calibration board image at each pose, and record the robot end pose matrix, specifically, capture the calibration board image at each pose, preferably 20 pairs.

[0060] S12, calculate the transformation matrix between the camera coordinate system and the robot coordinate system, which is the hand-eye calibration matrix.

[0061] S13, convert the point cloud in the camera coordinate system observed by the three-dimensional depth camera to the robot end coordinate system through the transformation matrix.

[0062] S14, control the robot to move to each sampling position according to the preset path, record the robot end pose matrix at each position, and capture the curved surface workpiece point cloud data at each position through the three-dimensional depth camera.

[0063] S15, use the transformation matrix and robot pose information to calculate the complete coordinate transformation matrix, and perform coordinate unified transformation on all point cloud data to convert all point clouds to the robot base coordinate system.

[0064] S2, pre-process the curved surface workpiece point cloud.

[0065] In step S2, the pre-processing includes the following steps.

[0066] S21, crop the point cloud, specifically crop the large-format point cloud to the workpiece area.

[0067] Preferably, the cropped working area is a quadrilateral, so as to obtain two pairs of mutually perpendicular boundaries.

[0068] S22, down-sample the point cloud.

[0069] Down-sampling can reduce the amount of data and retain effective information, and improve the efficiency of later calculation. In this embodiment, voxel grid down-sampling is used, which includes the following steps:

[0070] S221, divide the 3D space into a fixed-size cubic grid.

[0071] S222, traverse all points and calculate which voxel each point belongs to.

[0072] S223, use a spatial hash table to store voxels and points contained therein, for each non-empty voxel, calculate the centroid of all points therein, and replace all points in the voxel with the centroid.

[0073] S3, calculate the normal vector of each point in the point cloud and perform Poisson reconstruction to obtain a triangular facet mesh.

[0074] The calculation of the normal vector includes the following steps.

[0075] S311, for each point in the point cloud, search its K1 nearest neighbors, in this embodiment, the K1 is preferably 20.

[0076] S312, by calculating the covariance matrix, perform eigenvalue decomposition on it, the covariance matrix is

[0077]

[0078] Wherein, n is the number of points in the neighborhood, p i is the coordinate vector (x, y, z) of the i-th point in the neighborhood, μ is the centroid of the neighborhood point set.

[0079] S313, the eigenvector corresponding to the minimum eigenvalue is the sought normal vector.

[0080] And, after obtaining the normal vector, smooth it, and the smoothing process specifically includes the following steps.

[0081] S321, for each point in the point cloud, find its K2 nearest neighbors, in this embodiment, the K2 is 10.

[0082] S322, calculate the average normal vector of the nearest neighbors, which is the smoothed normal vector of this point.

[0083] The smoothing process can reduce the influence of noise and make the normal vector change more smoothly, so as to improve the reconstruction quality.

[0084] S4, plane parameterization is performed on the triangular patch mesh, and LSCM is used to map the triangular patch mesh to the plane parameter domain, the vertices of the plane parameter domain formed are one-to-one corresponding to the vertices of the triangular mesh surface, so as to obtain the graphic processing coordinates in the robot base coordinate system. The method of LSCM (Least Squares Conformal Maps) can realize the mapping of 3D grid to 2D plane by minimizing the deformation energy. At the same time, it does not need to fix the boundary points, so it can get a more natural unfolding result, effectively reducing the distortion of the edge area.

[0085] The LSCM mapping is as follows.

[0086] The conformal mapping from plane X to surface U requires that the tangent vector X(u, v) after mapping not only maintains orthogonality, but also maintains equal length, satisfying the basic equation of Cauchy-Riemann equation of formula (1-1).

[0087] (1-1)

[0088] To construct the parametric linear system, the complex form is used to express the u, v coordinates. For each patch in the triangular mesh surface, a local orthogonal coordinate system is defined, in which the local coordinates of the three vertices are denoted as (x1, y1), (x2, y2) and (x3, y3) respectively. The process of mapping the surface point U(x, y) to the parameter plane (u, v) can be represented by the discretized Cauchy-Riemann equations (1-2).

[0089] (1-2)

[0090] With the complex form of X and U, that is, X = x + iy, U = u + iv, equation (1-2) can be converted to equation (1-3).

[0091] (1-3)

[0092] For each triangular patch, the energy function shown in equation (1-4) is defined, and the least square conformal parameterization is to find the minimum value of the energy function.

[0093] (1-4)

[0094] wherein A T represents the area of the current triangular patch T. Then the energy function C(U) in the target mesh region is the sum of the energy functions of all single triangular patches, as shown in equation (1-5).

[0095] (1-5)

[0096] Therefore, the energy function equation of all triangular patches is shown in equation (1-6).

[0097] (1-6)

[0098] wherein d T is twice the area of the triangular patch T, U j1 , U j2 , U j3 represent the three vertices corresponding to the jth triangular patch, and W j1 , W j2 , W j3 are defined as:

[0099] (1-7)

[0100] Since the energy function C(U) is a quadratic form of complex U1, …, Un, it can also be represented by a matrix, as shown in equation (1-8).

[0101] (1-8)

[0102] where is an n x n Hermitian Gram matrix, U* represents the Hermitian conjugate matrix of U, which can be shown as (1-9).

[0103] (1-9)

[0104] Suppose the number of triangular patches contained in the triangular mesh surface is m, and the number of vertices is n. In order to solve the solution of the optimization problem, two or more points are added as fixed points. Suppose p vertices are determined as fixed points, then where Therefore, the matrix is divided into two parts, as shown in equation (1-10).

[0105] (1-10)

[0106] where, M f is an m x (n-p) matrix, M p is an m x p matrix. Therefore, equation (1-8) can be rewritten as shown in equation (1-11).

[0107] (1-11)

[0108] Finally, the linear system is shown in equation (1-12).

[0109] (1-12)

[0110] where.

[0111] (1-13)

[0112] By solving the least square numerical solution of the equation, the UV coordinates of all vertices of the target triangular mesh surface parameterized to the plane domain based on the LSCM method can be calculated. In order to solve the equation, the matrix A must be full rank, and must have two or more constraint fixed points, that is, p is greater than or equal to 2.

[0113] Since the cropped point cloud has a pair of parallel boundaries in the embodiment, the embodiment adopts a scheme of mapping the boundary points to a square parameter plane.

[0114] The plane parameterization specifically includes the following steps.

[0115] S41, extract the boundary vertex sequence of the mesh model, and form a closed loop according to the connection order of the boundary vertices.

[0116] S42, the total number of boundary vertices is n, then each boundary vertex can be associated with a parameter value on the interval [0, 1] by normalizing the parameter t = i / (n-1) (i is the vertex number, i∈[0,n-1]).

[0117] S43, the boundary vertices are mapped to the four edges of the unit square based on the value of the parameter t.

[0118] When t∈[0,0.25), the vertex is mapped to the lower edge of the square, with coordinates (0,4t).

[0119] When t∈[0.25,0.5), the vertex is mapped to the right edge of the square, with coordinates (4(t-0.25),1).

[0120] When t∈[0.5,0.75), the vertex is mapped to the upper edge of the square, with coordinates (1,1-4(t-0.5)).

[0121] When t∈[0.75,1], the vertex is mapped to the left edge of the square, with coordinates (1-4(t-0.75),0).

[0122] The above mapping scheme uniformly distributes the boundaries of the original mesh on the target square boundary, ensuring the continuity and uniformity of the boundaries. By fixing the positions of these boundary points, stable boundary constraints are provided for LSCM parameterization, which is conducive to obtaining more controllable parameterization results.

[0123] S5, according to the plane parameter domain vertex, to obtain a parameter plane mapping texture effect diagram and a triangular mesh surface texture mapping effect diagram, so as to obtain a preview effect according to the plane mapping texture effect diagram and the triangular mesh surface texture mapping effect diagram.

[0124] S6, according to the plane parameter domain vertex to obtain a graphic processing coordinate, and form a processing path, the robot completes the processing action according to the processing path, and through the set of algorithms, the mapping scheme of LSCM can obtain the need of ultra-high precision processing.

[0125] The above only describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solutions falling within the scope of the present application are within the protection scope of the present application. It should be noted that for ordinary technical personnel in the technical field, some improvements and decorations without departing from the principles of the present application are also considered as the protection scope of the present application.

Claims

1. A three-dimensional nesting method for curved workpieces processed by laser, characterized in that: The following steps are involved: S1. Connect the 3D depth camera to the robot, and use the robot to move the 3D depth camera to complete the point cloud acquisition of the curved workpiece; S2, pre-processing the surface workpiece point cloud; S3, calculating the normal vector of each point in the point cloud and performing Poisson reconstruction to obtain a triangular face mesh; S4, performing plane parameterization on the triangular face mesh, mapping the triangular face mesh to a plane parameter domain using LSCM to obtain plane parameter domain vertices; S5. Obtaining a parameter plane mapping texture effect image and a triangular mesh surface texture mapping effect image according to the plane parameter domain vertices; S6. Obtain the graphic processing coordinates according to the plane parameter domain vertices and form a processing path. The robot completes the processing action according to the processing path. The pre-processing in step S2 includes the following steps: S21, cropping the point cloud; S22, downsampling the point cloud; In step S21, the point cloud is clipped into a quadrilateral; The plane parameterization in step S4 includes the following steps: S41, extracting a boundary vertex sequence of the mesh model, and forming a closed loop by connecting the boundary vertices in a sequence; S42. Let the total number of boundary vertices be n. Use the normalization parameter t = i / (n-1), where i is the vertex index, i∈[0,n-1], to associate each boundary vertex with a parameter value in the interval [0,1]. S43, mapping the boundary vertices to the four sides of the unit square based on the value of the parameter t; The mapping rule corresponding to step S43 is: When t∈[0,0.25), the vertex is mapped to the bottom edge of the square with coordinates (0,4t); When t∈[0.25,0.5), the vertex is mapped to the right side of the square with coordinates (4(t-0.25),1); When t∈[0.5,0.75), the vertex is mapped to the upper edge of the square with coordinates (1,1-4(t-0.5)); When t∈[0.75,1], the vertex is mapped to the left side of the square with coordinates (1-4(t-0.75),0).

2. A three-dimensional nesting method for laser-processed curved workpieces according to claim 1, characterized in that: The point cloud collection corresponding to step S1 includes the following steps: S11, control the robot to move to different postures, collect the calibration plate image at each posture, and record the robot end posture matrix; S12, calculating the transformation matrix between the camera coordinate system and the robot coordinate system; S13, transforming the point cloud of the camera coordinate system observed by the three-dimensional depth camera into the robot end coordinate system through the transformation matrix; S14, controlling the robot to move to each sampling position along a preset path, recording the robot end pose matrix at each position, and collecting point cloud data of the curved workpiece at each position through a three-dimensional depth camera; S15. Calculate the complete coordinate transformation matrix using the transformation matrix and the robot's posture information, perform a unified coordinate transformation on all point cloud data, and convert all point clouds into the robot's base coordinate system.

3. The method for three-dimensional nesting of curved workpieces processed by laser according to claim 1, characterized in that: The normal vector calculation in step S3 includes the following steps: S311. For each point in the point cloud, search for its K1 nearest neighbor points; S312, by calculating the covariance matrix, performing eigenvalue decomposition on it, the covariance matrix is: Where n is the number of points in the neighborhood, p i is the coordinate vector (x, y, z) of the i-th point in the neighborhood, and μ is the centroid of the neighborhood point set; S313. The eigenvector corresponding to the minimum eigenvalue is the desired normal vector.

4. A three-dimensional nesting method for laser-processed curved workpieces according to claim 1, characterized in that: After the normal vector is obtained in S3, it is smoothed.

5. A three-dimensional nesting method for laser-processed curved workpieces according to claim 4, characterized in that: The smoothing process consists of the following steps: S321. For each point in the point cloud, find its K2 nearest neighbor points; S322. Calculate the average normal vector of the nearest neighbor point, which is the smoothed normal vector of the point.

Images