A welding seam milling guide method based on base material edge features

Abstract

The application provides a welding seam milling guide method based on base material edge characteristics, the base materials connected by the welding seam are recorded as base material I and base material II; edge line I and edge line II are selected on the base material I / base material II; a welding piece point cloud is acquired, and local point clouds of the edge line I and the edge line II are segmented from the point cloud; a spatial straight line I / spatial straight line II corresponding to each local point cloud is acquired; all intersection points between the spatial straight line I and the spatial straight line II are calculated and stored in a measured feature point set; a measured normal vector P of the current welding piece is acquired; a deviation matrix is calculated by using the measured feature point set and the measured normal vector P; and the robot corrects a milling track according to the deviation matrix; the method uses the base material edge point cloud information to accurately position the welding seam, calculates the deviation amount of the current welding piece compared with a teaching welding piece, guides the robot to accurately and efficiently mill the welding seam based on the deviation amount, avoids excessive milling of the base material, and is suitable for automatic milling processes of various types of welding seams.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, in particular to a welding seam milling guide method based on the edge features of base material. BACKGROUND

[0002] Welding is a process of combining two or more metal or non-metal components by heating and pressing to achieve atomic bonding, thereby forming an inseparable whole. It is one of the most important processing methods in manufacturing industry, widely used in construction, shipbuilding, automobile, aviation, aerospace and many other fields. Welding seam milling is a post-welding processing technology used to improve the appearance and dimensional accuracy of the weld, reduce stress concentration, or prepare for subsequent processing or detection. It is a high-precision and high-efficiency process.

[0003] With the development of automation technology, the current mainstream welding seam milling process is to mount a milling cutter at the end of a robot, and to program and teach the robot to obtain a trajectory that can mill the welding seam at the end of the milling cutter. Each time a new workpiece is positioned, the robot repeats this trajectory to mill the welding seam on the surface of the workpiece. This method can effectively reduce labor costs and improve milling efficiency. However, due to manufacturing errors, positioning errors and fixture clamping errors of the workpiece, the welding seam state of the workpiece surface after each milling cannot be guaranteed to be consistent, and problems such as excessive residual welding seam on the surface or excessive milling of the base material often occur. SUMMARY

[0004] To solve the above technical problems, the present application provides a welding seam milling guide method based on the edge features of the base material. This method is based on the point cloud information of the base material surface of the workpiece to describe the positioning of the current welding piece and its own welding state. The point cloud information of the edge of the base material is used to accurately position the welding seam. Based on the positioning coordinate information, the offset of the current welding piece compared to the teaching welding piece is calculated. Based on the offset, the robot accurately and efficiently mills the welding seam. This method can improve the precision of robot welding seam milling, effectively avoid the problem of excessive milling of the base material, and is suitable for automatic milling of various types of welding seams.

[0005] The technical solution is as follows:

[0006] A welding seam milling guide method based on the edge features of the base material, the welding seam milling process is as follows: a milling cutter is installed at the end of a robot, and a batch of welding pieces are sequentially positioned at a specified location. The robot mills the welding seam on the welding piece according to the taught milling trajectory.

[0007] The base material on both sides of the welding seam is respectively marked as base material I and base material II. The edge line of the welding seam on the base material I / base material II is selected and marked as edge line I. At least two edge lines are selected on the other base material and marked as edge line II.

[0008] The robot is guided to complete milling in the following manner:

[0009] ①Obtain the surface point cloud of the current welded piece, and sequentially segment the edge line I and the local point cloud corresponding to each edge line II from the point cloud;

[0010] ②Obtain the spatial straight line corresponding to each local point cloud, respectively;

[0011] ③According to the corresponding relationship between the local point cloud and the edge line I and the edge line II, mark the spatial straight line as a spatial straight line I / spatial straight line II;

[0012] Obtain all intersection points between the spatial straight line I and the spatial straight line II;

[0013] ④Store each intersection point in the measured feature point set;

[0014] Based on the surface point cloud of the current welded piece, obtain the plane normal vector corresponding to the base material I / base material II of the current welded piece, denoted as the measured normal vector P; retrieve the standard feature point set and the standard normal vector Q of the welded piece pre-stored by the teaching robot milling trajectory; use the corresponding relationship between each point in the standard feature point set and the measured feature point set, and the corresponding relationship between the standard normal vector Q and the measured normal vector P, to obtain the offset matrix;

[0015] ⑤The robot corrects the milling trajectory according to the offset matrix, and mills the weld on the current welded piece based on the corrected trajectory.

[0016] Further, in step ②, the manner of obtaining the spatial straight line corresponding to each local point cloud is as follows:

[0017] For a single local point cloud, obtain the main direction of the local point cloud, and then construct a spatial straight line based on the main direction and a three-dimensional point S, where the three-dimensional point S is one of the projection points of the local point cloud after projection along the main direction, or a three-dimensional point in the local point cloud.

[0018] Further, the manner of obtaining the main direction of the local point cloud is as follows:

[0019] Use the principal component analysis method (PCA) to obtain the main direction, or obtain the cross section of the current local point cloud, and denote the normal vector n of the cross section as the main direction.

[0020] Further, the manner of obtaining the cross section of the current local point cloud is as follows:

[0021] Pre-set an initial normal vector to construct an initial plane;

[0022] Project each three-dimensional point in the local point cloud onto the plane, and then on the current plane, count the area occupied by the projection points;

[0023] Iterative update of the normal vector, and construct a plane based on the updated normal vector, re-statistics of the projected point area;

[0024] When the area value is less than the threshold value or the area value tends to be stable, stop iteration, and record the current normal vector as the normal vector n and the current plane as the cross section.

[0025] Further, the three-dimensional point S is one of the projected points of the local point cloud along the main direction, and the three-dimensional point S is obtained in the following manner:

[0026] Step S1, any coordinate axis of the camera coordinate system is recorded as the A axis;

[0027] Calculate the angle θ between the normal vector n and the A axis, and then use the cross product of the normal vector n and the A axis to obtain the vector C. Rotate the normal vector n by the angle θ and the A axis to be parallel, and obtain the rotation matrix R through the Rodrigues transformation;

[0028] Step S2, rotate each projected point in the cross section using the rotation matrix R to obtain the rotated projected point;

[0029] Extract the other two coordinates except the A axis coordinate from the coordinates of each rotated projected point and store them in the two-dimensional point set;

[0030] Step S3, fit a curve using the two-dimensional point set, select a feature point on the curve, multiply the coordinates of the rotated projected point corresponding to the feature point by the inverse matrix of the rotation matrix R to obtain the pre-rotated projected point, and record the pre-rotated projected point as the three-dimensional point S;

[0031] Preferably, the feature point is the point with the maximum curvature change on the curve, the point with a slope of 0, the point with the maximum longitudinal coordinate value, or the preset distance point. The longitudinal coordinate is the coordinate component perpendicular to the parent material I / parent material II direction. The preset distance point is: draw a tangent line at the point with a slope of 0, translate the tangent line by a preset distance to intersect the curve, and record the intersection point as the preset distance point.

[0032] Further, the edge line I intersects the edge line II or is in a different plane;

[0033] In step ③, the intersection point includes a real intersection point and a virtual intersection point (intersection point of a space straight line in a different plane);

[0034] Preferably, the virtual intersection point is: point I, point II, or the midpoint of point I and point II; the point I is the point on the space straight line I closest to the space straight line II, and the point II is the point on the space straight line II closest to the space straight line I.

[0035] Further, the way to obtain the virtual intersection point includes the following two ways:

[0036] Method one:

[0037] The direction vector b1 of the spatial straight line I and the direction vector b2 of the spatial straight line II are cross-multiplied to obtain a vector b3;

[0038] A first plane is constructed by using the vectors b1 and b3, a second plane is constructed by using b2 and b3, and a straight line intersecting the first plane and the second plane is solved and is denoted as spatial straight line III;

[0039] An intersection point between the spatial straight line I and the spatial straight line III is solved and is denoted as point I;

[0040] An intersection point between the spatial straight line II and the spatial straight line III is solved and is denoted as point II;

[0041] The point I, the point II, or a midpoint of the point I and the point II is denoted as a virtual intersection point.

[0042] Method two:

[0043] The direction vector b1 of the spatial straight line I and the direction vector b2 of the spatial straight line II are cross-multiplied to obtain a vector b3;

[0044] One of the spatial straight lines is translated along the direction of the vector b3 to be coplanar with the other spatial straight line, and an intersection point between the two spatial straight lines at this time is solved and is denoted as a virtual intersection point.

[0045] Further, in step ④, the method for obtaining the offset matrix is:

[0046] Any intersection point in the measured feature point set is denoted as a measured intersection point H, and an intersection point opposite to the position of the point H in the standard feature point set is denoted as a standard intersection point K;

[0047] The measured intersection point H is translated to the origin of the camera coordinate system to obtain a translation matrix T1, other intersection point coordinates in the measured feature point set are translated by using the translation matrix T1, and the translated intersection point coordinates and the translated measured intersection point H are stored in a measured translation point set;

[0048] The standard intersection point K is translated to the origin of the camera coordinate system to obtain a translation matrix T2, other intersection point coordinates in the standard feature point set are translated by using the translation matrix T2, and the translated intersection point coordinates and the translated standard intersection point K are stored in a standard translation point set;

[0049] 3) Taking the standard intersection point K as a rotation center, the measured translation point set and the measured normal vector P are rotated by using a rotation matrix, and an optimal rotation matrix R' corresponding to a time when a difference between the measured translation point set and the standard translation point set, the measured normal vector P and the standard normal vector Q is smallest is obtained by using an optimization method;

[0050] 4) Then, the standard intersection K is moved from the original point of the coordinate system to the coordinate position of the original measured intersection H, and a translation matrix T3 is obtained;

[0051] Offset matrix = T3 x R' x T2.

[0052] Preferably, the standard feature point set of the welding part and the standard normal vector Q are obtained as follows:

[0053] Before teaching the robot to mill the trajectory, an optional welding part from the batch of welding parts is placed at a designated position, and is recorded as the teaching welding part;

[0054] Based on the teaching welding part, the robot is taught to mill the trajectory so that it can drive the milling cutter to accurately mill the weld on the teaching welding part;

[0055] The steps ①-③ are performed on the teaching welding part, and each intersection is stored in the standard feature point set;

[0056] Based on the surface point cloud of the teaching welding part, a plane normal vector corresponding to the base material I / base material II of the teaching welding part is obtained, and is recorded as the standard normal vector Q.

[0057] Preferably, in step ①, the local point cloud corresponding to the edge line I and the edge line II is segmented from the point cloud as follows:

[0058] In the base material I and the base material II of the teaching welding part, one or more edge lines are selected, and are recorded as edge line I and edge line II;

[0059] In the surface point cloud of the teaching welding part, the local point cloud corresponding to each edge line I and each edge line II is manually framed, and the position and size of each frame are recorded;

[0060] In step ①, the local point cloud corresponding to each edge line I and each edge line II is framed from the point cloud according to the recorded position and size of each frame.

[0061] The method solves the problem of excessive dependence on the repeat positioning accuracy and manufacturing accuracy of the workpiece by relying on the robot to mill the welding part; the workpiece base material edge features are positioned through point cloud data, the edge features represent the state of the two lapped base materials, the intersection coordinates are calculated, the workpiece feature changes of the current welding part (current state) and the teaching welding part (teaching state) are calculated based on the coordinates, the robot is guided to mill, which can greatly improve the accuracy of weld milling while ensuring efficiency, and avoid excessive milling of the base material. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 A schematic view of selecting an edge line on a welding part;

[0063] Figure 2 Effect diagram of single piece local point cloud and projection point on cross section;

[0064] Figure 3 Effect diagram of curve fitted by two-dimensional point set in step S3. DETAILED DESCRIPTION

[0065] The technical solutions of the present application will be described in detail below in combination with the drawings and specific embodiments.

[0066] A welding seam milling guide method based on base material edge characteristics, the welding seam milling process is: a milling cutter is installed at the end of a robot, batch welding parts reach designated positions in turn, at the designated positions, the robot mills the welding seam on the welding part according to the taught milling trajectory;

[0067] As shown in the figure, the base materials connected on both sides of the welding seam are respectively marked as base material I and base material II; an edge line where the welding seam is located on the base material I / base material II is selected and marked as edge line I, at least two edge lines on the other base material are selected and respectively marked as edge line II; the edge line I intersects with the edge line II or is in different planes; Figure 1 In detail, the edge line can be understood as including the contour line of the outer convexity / inner concavity of the base material surface and the edge line of the base material itself.

[0068] The robot is guided to complete milling in the following way:

[0069] ①Obtain the surface point cloud of the current welding part, and sequentially segment the edge line I and the local point cloud corresponding to each edge line II from the point cloud; that is, M pieces of local point clouds are obtained, M represents the total number of the edge line I and the edge line II;

[0070] ②Obtain the spatial straight line corresponding to each local point cloud respectively;

[0071] ③According to the corresponding relationship between the local point cloud and the edge line I and the edge line II, the spatial straight line is marked as spatial straight line I / spatial straight line II correspondingly;

[0072] Obtain all intersection points between the spatial straight line I and the spatial straight line II; the intersection points include real intersection points and virtual intersection points; wherein, the virtual intersection points are: point I, point II or the midpoint of point I and point II; point I is the point on the spatial straight line I closest to the spatial straight line II, and point II is the point on the spatial straight line II closest to the spatial straight line I;

[0073] As a preferred embodiment, in order to prevent excessive milling of the base material, the coordinate component values of point I and point II in the direction perpendicular to the base material I / base material II are compared, and the point with the larger coordinate component value is marked as the virtual intersection point.④Store each intersection point to the measured feature point set;

[0074] As a preferred embodiment, in order to prevent excessive milling of the base material, the coordinate component values of point I and point II in the direction perpendicular to the base material I / base material II are compared, and the point with the larger coordinate component value is marked as the virtual intersection point.④Store each intersection point to the measured feature point set;

[0075] Based on the surface point cloud of the current welding piece, a plane normal vector corresponding to the base metal I / base metal II of the current welding piece is obtained, denoted as a measured normal vector P; when the teaching robot milling trajectory is called, the pre-stored standard feature point set and standard normal vector Q of the welding piece are retrieved; the offset matrix is obtained by using the correspondence between the standard feature point set and each point in the measured feature point set and the correspondence between the standard normal vector Q and the measured normal vector P;

[0076] As a preferred embodiment, in order to prevent excessive milling of the base metal, the point cloud corresponding to the base metal I / base metal II can be compared, and the base metal with a higher average depth of the point cloud is denoted as the reference base metal, and the normal vector of the reference base metal is the normal vector of the high plane;

[0077] ⑤ The robot corrects the milling trajectory according to the offset matrix, and mills the weld on the current welding piece based on the corrected trajectory.

[0078] In detail, in step ②, the manner of obtaining the spatial straight line corresponding to each local point cloud is as follows:

[0079] For a single local point cloud, the main direction of the local point cloud is obtained, and then a spatial straight line is constructed based on the main direction and the three-dimensional point S, wherein the three-dimensional point S is one of the projection points of the local point cloud after projection along the main direction, or is one of the three-dimensional points in the local point cloud.

[0080] The manner of obtaining the main direction of the local point cloud is as follows:

[0081] The principal component analysis method (PCA) is used to obtain the cross section of the current local point cloud, and the normal vector n of the cross section is denoted as the main direction.

[0082] Specifically, the manner of obtaining the cross section of the current local point cloud is as follows:

[0083] An initial normal vector is preset to construct an initial plane;

[0084] For example, Figure 2 Each three-dimensional point in the local point cloud is projected onto the plane, and the area occupied by the projection points on the current plane is counted;

[0085] The normal vector is iteratively updated, and the plane is constructed based on the updated normal vector, and the area of the projection points is counted again;

[0086] When the area value is less than a threshold value or the area value tends to be stable, the iteration is stopped, and the current normal vector is denoted as the normal vector n, and the current plane is denoted as the cross section.

[0087] In specific implementation, the normal vector is iteratively updated based on an optimization method, and the objective function can be set as:

[0088]

[0089] wherein, p1, p2...p n is the n three-dimensional point coordinates in the local point cloud, p is any point coordinate on the plane, and n represents the normal vector that needs to be iterated.

[0090] The method for calculating the area occupied by the projection points is the grid method. Specifically, the length a is defined, and the grid is filled on the current plane, and a single grid is a square with a side length of a; the total number of grids occupied by the projection points is counted, and the area occupied by the projection points is obtained by multiplying the total number by the area of a single grid.

[0091] In more detail, the three-dimensional point S is one of the projection points of the local point cloud along the main direction, and the three-dimensional point S is obtained in the following manner:

[0092] Step S1, any coordinate axis (X axis, Y axis or Z axis) of the camera coordinate system is denoted as A axis;

[0093] The angle θ between the normal vector n and the A axis is calculated, and the vector C is obtained by the cross product of the normal vector n and the A axis. The normal vector n is rotated by the angle θ and parallel to the A axis, and the rotation matrix R is obtained by the Rodrigues transformation;

[0094] Step S2, the rotation matrix R is used to rotate each projection point in the cross section to obtain the rotated projection point;

[0095] The other two coordinates except the A axis coordinate are extracted from the coordinates of each rotated projection point and stored in the two-dimensional point set;

[0096] Step S3, a curve (such as Figure 3 ) is fitted using the two-dimensional point set, a feature point is selected on the curve, the coordinates of the rotated projection point corresponding to the feature point are multiplied by the inverse matrix of the rotation matrix R to obtain the projection point before rotation, and the projection point before rotation is denoted as the three-dimensional point S;

[0097] wherein, the feature point can be the point with the maximum curvature change on the curve, the point with a slope of 0, the point with the maximum vertical coordinate value, or the preset distance point, the vertical coordinate is the coordinate component perpendicular to the parent material I / parent material II direction, and the preset distance point is: a tangent line is drawn at the point with a slope of 0, the tangent line is translated by a preset distance to intersect with the curve, and the intersection point is denoted as the preset distance point.

[0098] In this scheme, the three-dimensional point S is a point with stability and consistency in the point cloud characteristics of the edge line, and in specific implementation, one three-dimensional point in the local point cloud can also be directly extracted, and the specific steps are as follows: the local point cloud is denoised, and the mean value of the X axis coordinate of all three-dimensional points in the local point cloud is calculated the mean value of the Y axis coordinate the mean value of the Z axis coordinate In the local point cloud, the point closest to The point is recorded as a three-dimensional point S.

[0099] In implementation, in step ①, the local point cloud corresponding to the edge line I and the edge line II is segmented from the point cloud in the following manner:

[0100] In the base material I and the base material II of the teaching welding piece, one or more edge lines are selected, and are marked as edge line I and edge line II.

[0101] In the surface point cloud of the teaching welding piece, the local point cloud corresponding to each edge line I and each edge line II is manually framed, and the position and size of each frame are recorded.

[0102] In step ①, the local point cloud corresponding to each edge line I and each edge line II is framed from the point cloud according to the recorded position and size of each frame.

[0103] Further, in step ③, the manner of obtaining the virtual intersection point between each space straight line I and each space straight line II is provided in the following two ways in the embodiment:

[0104] Method one:

[0105] The direction vector b1 of the space straight line I and the direction vector b2 of the space straight line II are cross-multiplied to obtain a vector b3;

[0106] A first plane is constructed using the vectors b1 and b3, a second plane is constructed using b2 and b3, and a straight line intersecting the first plane and the second plane is solved, which is recorded as a space straight line III;

[0107] The intersection point between the space straight line I and the space straight line III is solved, which is recorded as a point I;

[0108] The intersection point between the space straight line II and the space straight line III is solved, which is recorded as a point II;

[0109] The point I, the point II, or the midpoint of the point I and the point II is recorded as the virtual intersection point.

[0110] Method two:

[0111] The direction vector b1 of the space straight line I and the direction vector b2 of the space straight line II are cross-multiplied to obtain a vector b3;

[0112] One of the space straight lines is translated along the direction of the vector b3 to be coplanar with the other space straight line, and the intersection point between the two space straight lines at this time is solved, which is recorded as the virtual intersection point.

[0113] In step ④, the manner of obtaining the offset matrix is:

[0114] Record any intersection point in the measured feature point set as measured intersection point H, find the intersection point in the standard feature point set which is opposite to the position of point H, and record it as standard intersection point K;

[0115] Translate measured intersection point H to the origin of the camera coordinate system to obtain translation matrix T1, and translate the coordinates of other intersection points in the measured feature point set using translation matrix T1. Store the translated intersection point coordinates and the translated measured intersection point H in the measured translation point set.

[0116] Translate standard intersection point K to the origin of the camera coordinate system to obtain translation matrix T2; and translate the coordinates of other intersection points in the standard feature point set using translation matrix T2. Store the translated intersection point coordinates and the translated standard intersection point K in the standard translation point set.

[0117] 3) Take standard intersection point K as the center of rotation, rotate the measured translation point set and the measured normal vector P using the rotation matrix, and use the optimization method to obtain the rotation matrix R' corresponding to the minimum difference between the measured translation point set and the standard translation point set, the measured normal vector P and the standard normal vector Q.

[0118] 4) Translate standard intersection point K from the origin of the coordinate system to the coordinate position of the original measured intersection point H to obtain translation matrix T3.

[0119] Offset matrix = T3 x R' x T2.

[0120] More specifically, in step 3), a target function can be constructed, and the optimization method is used to obtain the rotation matrix R' corresponding to the minimum difference between the measured translation point set and the standard translation point set, the measured normal vector P and the standard normal vector Q.

[0121] For example, two expressions of the target function are given: see Target Function One and Target Function Two.

[0122] Among them, the target function one is:

[0123] Wherein, α, β are preset weight coefficients, N represents the number of intersection points, N≥2, a i represents the i-th intersection point coordinate in the measured translation point set, b i represents the i-th intersection point coordinate in the standard translation point set.

[0124] Using the optimization method to obtain the value of Euler angle θ(rx, ry, rz) corresponding to the minimum residual error between the measured translation point set and the standard translation point set, the measured normal vector P and the standard normal vector Q; and then based on the value of Euler angle θ(rx, ry, rz) to obtain the rotation matrix R' in the offset matrix.

[0125] Target function two:

[0126] wherein, α, β are preset weight coefficients, N represents the number of intersection points, N≥2, a i represents the i th intersection point coordinate in the measured translation point set, b i represents the i th intersection point coordinate in the standard translation point set;

[0127] The value of the Euler angle θ (rx, ry, rz) corresponding to the minimum residual between the measured translation point set and the standard translation point set and the minimum angle between the measured normal vector P and the standard normal vector Q is obtained by using an optimization method; and the rotation matrix R' in the offset matrix is obtained based on the value of the Euler angle θ (rx, ry, rz).

[0128] In specific implementation, the following processes are performed in advance: the stored standard feature point set of the welded part and the standard normal vector Q:

[0129] Before teaching the robot to mill the track, an optional welded part is selected from the batch of welded parts, denoted as a teaching welded part, and is placed at a designated position;

[0130] Based on the teaching welded part, the robot is taught to mill the track so as to accurately drive the milling cutter to mill the weld on the teaching welded part;

[0131] The teaching welded part is subjected to steps ①-③; and each intersection point is stored in the standard feature point set;

[0132] Based on the surface point cloud of the teaching welded part, a plane normal vector corresponding to the base metal I / base metal II of the teaching welded part is obtained, denoted as a standard normal vector Q.

[0133] The present scheme positions the workpiece base metal edge features through point cloud data, takes the edge features to represent the state of the two lapped base metals, obtains the intersection point coordinates, calculates the workpiece feature changes between the current welded part (current state) and the teaching welded part (teaching state) based on the coordinates, guides the robot to mill, can greatly improve the precision of weld milling while ensuring efficiency, and avoids excessive milling of the base metal.

[0134] The foregoing description of specific exemplary embodiments presenting the present application is for illustrative and descriptive purposes. The foregoing description is not intended to be all-inclusive or to be limited to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teachings. The specific exemplary embodiments are chosen and described in order to explain the particular principles of the present application and its practical application to thereby enable others skilled in the art to best utilize the present application and various embodiments thereof, as well as various alternatives and modifications thereof. It is intended that the scope of the present application be defined by the following claims and their equivalents.

Claims

1. A method for guiding weld seam milling based on base material edge features, the weld seam milling process being: a robot end is equipped with a milling cutter, a batch of welded parts reach a specified position in turn, and the robot mills the weld seam on the welded parts according to a taught milling trajectory; characterized in that the base materials on both sides of the weld seam are respectively denoted as base material I and base material II; an edge line on which the weld seam is located is selected on the base material I / base material II and denoted as edge line I, and at least two edge lines are selected on the other base material and respectively denoted as edge line II; the robot is guided to complete milling in the following manner: ① the surface point cloud of the current welded part is obtained, and the edge line I and the local point cloud corresponding to each edge line II are sequentially segmented from the point cloud; ② the spatial straight line corresponding to each local point cloud is obtained respectively; ③ the spatial straight line is correspondingly marked as spatial straight line I / spatial straight line II according to the corresponding relationship between the local point cloud and the edge line I / edge line II; all intersection points between the spatial straight line I and the spatial straight line II are calculated; ④ each intersection point is stored in a measured feature point set; the plane normal vector corresponding to the base material I / base material II of the current welded part is obtained based on the surface point cloud of the current welded part and denoted as measured normal vector P; the standard feature point set of the welded part and the standard normal vector Q pre-stored by the teaching robot milling trajectory are called; the offset matrix is calculated by using the corresponding relationship between each point in the standard feature point set and the measured feature point set and the corresponding relationship between the standard normal vector Q and the measured normal vector P; ⑤ the robot corrects the milling trajectory according to the offset matrix, and mills the weld seam on the current welded part based on the corrected trajectory.

2. The parent material edge feature based weld beveling guidance method of claim 1, wherein: In step ②, the spatial straight line corresponding to each local point cloud is obtained in the following manner: for a single local point cloud, the main direction of the local point cloud is obtained, and then a spatial straight line is constructed based on the main direction and a three-dimensional point S, wherein the three-dimensional point S is one of the projection points of the local point cloud along the main direction, or is one of the three-dimensional points in the local point cloud.

3. The weld bevel based on parent material edge feature milling guidance method of claim 2, wherein: The main direction of the local point cloud is obtained in the following manner: the main direction is obtained by using a principal component analysis method, or a cross section of the current local point cloud is obtained, and the normal vector n of the cross section is denoted as the main direction.

4. The weld bevel based on parent material edge feature milling guidance method of claim 3, wherein: The cross section of the current local point cloud is obtained in the following manner: an initial normal vector is pre-set, and an initial plane is constructed; each three-dimensional point in the local point cloud is projected onto the plane, and the area occupied by the projection points on the current plane is counted; the normal vector is iteratively updated, and the plane is constructed based on the updated normal vector, and the area of the projection points is counted again; when the area value is less than a threshold value or the area value tends to be stable, the iteration is stopped, the current normal vector is denoted as the normal vector n, and the current plane is denoted as the cross section.

5. The weld bevel based on parent material edge features milling guidance method of claim 4, wherein: The three-dimensional point S is one of the projection points of the local point cloud along the main direction, and the three-dimensional point S is obtained in the following manner: In step S1, any coordinate axis of a camera coordinate system is denoted as an A axis; the included angle θ between the normal vector n and the A axis is calculated, and then the vector C is obtained by using the cross product of the normal vector n and the A axis, the normal vector n is rotated to be parallel to the A axis with the vector C as the rotation axis, and the rotation matrix R is obtained by using the Rodrigues transformation. Step S2, rotating each projection point in the cross section by using a rotation matrix R to obtain a rotated projection point; Extracting two coordinates other than the A-axis coordinate from the coordinates of each rotated projection point and storing them into a two-dimensional point set; Step S3, fitting a curve by using the two-dimensional point set, selecting a feature point on the curve, multiplying the coordinates of the rotated projection point corresponding to the feature point by an inverse matrix of the rotation matrix R to obtain a pre-rotated projection point, and recording the pre-rotated projection point as a three-dimensional point S; The feature point is a point with the largest curvature change on the curve, a point with a slope of 0, a point with the largest longitudinal coordinate value, or a preset distance point. The longitudinal coordinate is a coordinate component perpendicular to the parent material I / parent material II direction. The preset distance point is obtained by drawing a tangent line at the point with a slope of 0, translating the tangent line by a preset distance to intersect with the curve, and recording the intersection point as the preset distance point.

6. The parent material edge feature based weld beveling guidance method of claim 1, wherein: The edge line I intersects with the edge line II or is in a different plane; In step ③, the intersection point includes a real intersection point and a virtual intersection point; The virtual intersection point is a point I, a point II, or a midpoint of the point I and the point II. The point I is a point on the space straight line I closest to the space straight line II, and the point II is a point on the space straight line II closest to the space straight line I.

7. The weld bevel based on parent material edge feature milling guidance method of claim 6, wherein: The method for obtaining the virtual intersection point includes the following two methods: Method one: The direction vector b1 of the space straight line I and the direction vector b2 of the space straight line II are cross-multiplied to obtain a vector b3; A first plane is constructed by using the vectors b1 and b3, a second plane is constructed by using the vectors b2 and b3, a straight line intersecting the first plane and the second plane is solved and recorded as a space straight line III; An intersection point between the space straight line I and the space straight line III is solved and recorded as the point I; An intersection point between the space straight line II and the space straight line III is solved and recorded as the point II; The point I, the point II, or the midpoint of the point I and the point II is recorded as the virtual intersection point; Method two: The direction vector b1 of the space straight line I and the direction vector b2 of the space straight line II are cross-multiplied to obtain a vector b3; One of the space straight lines is translated in the direction of the vector b3 to be coplanar with the other space straight line, and an intersection point between the two space straight lines at this time is solved and recorded as the virtual intersection point.

8. The parent material edge feature based weld beveling guidance method of claim 1, wherein: In step ④, the method for obtaining the offset matrix is: Any intersection point in the measured feature point set is recorded as a measured intersection point H, and an intersection point opposite to the position of the point H in the standard feature point set is found and recorded as a standard intersection point K; The measured intersection point H is translated to the origin of the camera coordinate system to obtain a translation matrix T1, the coordinates of other intersection points in the measured feature point set are translated by using the translation matrix T1, and the translated intersection point coordinates and the translated measured intersection point H are stored in a measured translation point set; The standard intersection point K is translated to the origin of the camera coordinate system to obtain a translation matrix T2, the coordinates of other intersection points in the standard feature point set are translated by using the translation matrix T2, and the translated intersection point coordinates and the translated standard intersection point K are stored in a standard translation point set; 3) taking the standard intersection point K as the rotation center, rotating the measured translation point set and the measured normal vector P by using the rotation matrix, and using the optimization method to obtain the corresponding rotation matrix R' when the difference between the measured translation point set and the standard translation point set, the measured normal vector P and the standard normal vector Q is minimized; 4) then moving the standard intersection point K from the origin of the coordinate system to the coordinate position of the original measured intersection point H to obtain the translation matrix T3; Offset matrix = T3 x R' x T2.

9. The parent material edge feature based weld beveling guidance method of claim 1, wherein: The pre-stored standard feature point set of the welded part and the standard normal vector Q are obtained as follows: Before teaching the robot to mill the track, select a welded part from the batch of welded parts, denoted as the teaching welded part, and place it at a designated position; Based on the teaching welded part, teach the robot to mill the track so that it can accurately mill the weld on the teaching welded part with the milling cutter; Perform steps ①-③ on the teaching welded part, and store each intersection point in the standard feature point set; Based on the surface point cloud of the teaching welded part, obtain the plane normal vector corresponding to the base material I / base material II of the teaching welded part, denoted as the standard normal vector Q.

10. The weld bevel based on parent material edge feature milling guidance method of claim 9, wherein: In step ①, the local point cloud corresponding to the edge line I and the edge line II is segmented from the point cloud as follows: Select one or more edge lines from the base material I and the base material II of the teaching welded part, respectively, and mark them as edge line I and edge line II; In the surface point cloud of the teaching welded part, manually frame the local point cloud corresponding to each edge line I and each edge line II, and record the position and size of each frame; In step ①, frame the local point cloud corresponding to each edge line I and each edge line II from the point cloud according to the recorded position and size of each frame.

Images