Spoke polishing track generation method based on point cloud and theoretical model fusion

Abstract

This invention provides a method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model, relating to the field of wheel spoke grinding, and includes the following steps: scanning the actual wheel hub workpiece to obtain the actual contour point cloud P of the wheel spokes. s The theoretical profile point cloud P is obtained based on the theoretical model of the wheel hub. t The actual contour point cloud and the theoretical contour point cloud are registered using ICP, and the theoretical model is transformed into the measurement coordinate system of the actual wheel hub workpiece to obtain the transformed theoretical contour point cloud. The actual contour point cloud is then compared with the theoretical model. s Compare with theoretical contour point clouds to eliminate actual contour point cloud P s Anomalies are identified in the data to obtain a filtered actual contour point cloud. The CPD algorithm is then used to perform non-rigid deformation registration between the actual and theoretical contour point clouds, resulting in a new contour point cloud. A grinding trajectory is then generated based on this contour point cloud. This invention effectively improves grinding accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of wheel spoke grinding, and in particular to a method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model. Background Technology

[0002] For wheel spoke grinding, traditional robotic grinding mainly relies on the theoretical CAD model of the part, directly generating the machining path through CAM software. However, in actual production, due to manufacturing or clamping errors, the geometry of the part deviates from the theoretical model. Directly using the theoretical path for grinding can easily lead to over-cutting or incomplete grinding, especially in boundary areas requiring high-precision contour control. Summary of the Invention

[0003] The main objective of this invention is to propose a method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model, which can effectively improve grinding accuracy and stability.

[0004] This invention is achieved through the following technical solution:

[0005] The method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model includes the following steps:

[0006] Step S1: Scan the actual wheel hub workpiece to obtain the set of wheel spoke contour points, and then obtain the actual contour point cloud P of the wheel spokes. s Based on the theoretical model of the wheel hub, multiple spoke edges are identified, and the contours of each spoke segment are uniformly discretized to obtain the theoretical contour point cloud P. t ;

[0007] Step S2: Perform ICP registration between the actual contour point cloud and the theoretical contour point cloud, transforming the theoretical model to the measurement coordinate system of the actual wheel hub workpiece, to obtain the transformed theoretical contour point cloud. The actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. ;

[0008] Step S3: Implement the actual contour point cloud using the CPD algorithm. With theoretical outline point cloud Non-rigid deformation registration between them yields the contour point cloud. Based on this contour point cloud Generate polishing trajectory.

[0009] Furthermore, in step S1, a binocular surface scanning 3D camera is used to scan the wheel hub to generate three-dimensional point cloud data of the wheel hub. .

[0010] Furthermore, in step S1, the actual contour point cloud P is obtained through the following steps. s :

[0011] Step S11: Apply statistical filtering to the 3D point cloud data. Processing is performed to eliminate isolated noise points introduced during data acquisition;

[0012] Step S12: Use voxel grid filtering to uniformly downsample the statistically filtered point cloud data to obtain point cloud data. ;

[0013] Step S13: Based on point cloud data The relationship between the normal directions of each point and its neighboring points is used to determine whether the point lies on the spoke contour boundary, thus obtaining the actual contour point cloud P. s .

[0014] Furthermore, in step S1, the theoretical contour point cloud P is obtained through the following steps. t :

[0015] Step S14: Read multiple triangular facets from the STL file storing the CAD theoretical model of the wheel hub;

[0016] Step S15: For edges E with shared edges AB Two adjacent triangular facets T A and T B According to the formula Calculate T A and T B Angle between normal vectors ,in, and T respectively A and T B The normal vector;

[0017] Step S16, when the included angle When the angle is less than the set threshold, the shared edge E AB Let P be a segment of the wheel spoke profile. For each profile, it is uniformly discretized according to a pre-defined discrete interval to obtain the theoretical profile point cloud P. t .

[0018] Furthermore, step S2 specifically includes the following steps:

[0019] Step S21: For the theoretical contour point cloud P t Each point in the actual contour point cloud P s Search for the point with the closest Euclidean distance in the middle and establish a point-to-point correspondence;

[0020] Step S22: Based on the established point-to-point correspondence, construct an optimization function with mean square error as the objective, and solve for the optimal rigid body transformation parameters that minimize the mean square error. The optimal rigid body transformation parameters include the rotation matrix and translation vector.

[0021] Step S23: Apply the rotation matrix and translation vector to the theoretical contour point cloud P. t Update the spatial position of each point, proceed to step S21 for the next iteration, until the mean square error change is lower than a preset threshold or the maximum number of iterations is reached, then terminate the iteration, obtaining the final rotation matrix R and translation vector T. The transformed theoretical contour point cloud is represented as follows: ;

[0022] Step S24: Transfer the actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. ;

[0023] Furthermore, step S24 specifically involves: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] s Each point in Calculate theoretical contour point cloud Neutral nearest point Calculation points With point The difference between the Euclidean distance Dis and the Z coordinate. If the Euclidean distance Dis is less than the set distance threshold and the difference is... If the deviation is less than the threshold along the Z-axis, then the point If a point is considered valid, it is considered invalid. Filtering out all invalid points yields the actual contour point cloud. .

[0024] Furthermore, in step S3, the contour point cloud is obtained through the following steps. :

[0025] Step S31: Using the Gaussian mixture theory model as the core, convert the actual contour point cloud... Viewed as a point cloud of theoretical outline Generated by sampling from a Gaussian mixture distribution; motion consistency constraints are introduced to refine the theoretical contour point cloud. A displacement field is applied to each point in the cloud to gradually approximate the actual contour point cloud while maintaining the consistency of the topological structure. Location;

[0026] Step S32: Calculate the actual contour point cloud Each point in the middle corresponds to a theoretical contour point cloud. The posterior probabilities of each Gaussian component are obtained. Based on these posterior probabilities, the optimal displacement vector is solved by minimizing the negative log-likelihood function and applying motion smoothness constraints. The theoretical contour point cloud is then updated based on this optimal displacement vector. The positions of each point are determined, and the process proceeds to step S32 for the next iteration. After the iteration is completed, the contour point cloud is obtained. .

[0027] Furthermore, in step S3, based on the contour point cloud... The polishing trajectory is obtained through the following steps:

[0028] Step S33: Calculate the contour point cloud geometric centroid Calculate the contour point cloud respectively various points in the middle With geometric centroid The connecting vectors between them are determined by the angle between the connecting vectors and the preset reference direction. Size, for contour point cloud The points are arranged in ascending order to form a continuous and non-intersecting ordered list of points along the circumferential contour of the wheel spokes. ;

[0029] Step S34: Arrange the ordered points As the control vertex, a non-uniform rational B-spline curve is fitted, and parameters are sampled along the B-spline curve according to a preset fixed step size to obtain a series of discrete trajectory point sets with equal arc length spacing. The grinding trajectory is the discrete trajectory point set.

[0030] As can be seen from the above description of the present invention, compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. This invention first scans the actual wheel hub workpiece to obtain the set of wheel spoke contour points, and then obtains the actual contour point cloud P of the wheel spokes. s Based on the theoretical model of the wheel hub, multiple spoke edges are identified, and the contours of each spoke segment are uniformly discretized to obtain the theoretical contour point cloud P. t Then, the actual contour point cloud and the theoretical contour point cloud are registered using ICP, transforming the theoretical model to the measurement coordinate system of the actual wheel hub workpiece, thus obtaining the transformed theoretical contour point cloud. The actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. Finally, the CPD algorithm was used to realize the actual contour point cloud. With theoretical outline point cloud Non-rigid deformation registration between them yields the contour point cloud. Based on this contour point cloud The generation of grinding trajectories integrates the actual spoke point cloud, which contains deformations and gaps, with the theoretical model. This not only compensates for geometric deviations caused by machining deformation but also effectively repairs discontinuous areas in the actual point cloud using theoretical model data. This provides a complete and accurate contour foundation for subsequent trajectory generation, avoiding over-cutting or incomplete grinding, thereby improving grinding accuracy and stability. From point cloud acquisition to trajectory generation, the entire process is highly automated, significantly reducing manual intervention and improving the efficiency and consistency of spoke grinding operations, providing reliable technical support for intelligent manufacturing. By constructing a multi-level registration system of ICP rigid registration and CPD non-rigid registration, the rigid positioning error and non-rigid deformation error of the actual workpiece are eliminated sequentially, achieving comprehensive and accurate matching between the actual contour and the theoretical model. This effectively overcomes the limitations of traditional single registration methods in dealing with complex deformations.

[0032] 2. This invention employs a point cloud preprocessing workflow that combines statistical filtering, voxel mesh downsampling, and boundary contour extraction, which significantly improves the quality and processing efficiency of point cloud data. In particular, by introducing a contour deviation filtering mechanism, based on the dual criteria of geometric features and spatial location, it effectively identifies and removes abnormal data points caused by factors such as burr residue and surface reflection, ensuring the reliability of contour data.

[0033] 3. This invention ensures smooth motion and consistent speed during the grinding process by using equal arc length discretization resampling, effectively avoiding mechanical vibration and speed abrupt changes caused by traditional piecewise linear trajectories. Attached Figure Description

[0034] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] Figure 1 This is a flowchart of the present invention.

[0036] Figure 2 The actual contour point cloud P of the present invention s picture.

[0037] Figure 3 The theoretical contour point cloud P of this invention t picture.

[0038] Figure 4 The theoretical contour point cloud after ICP registration in this invention and actual contour point cloud picture.

[0039] Figure 5 The actual contour point cloud after filtering according to the present invention. picture.

[0040] Figure 6 This is the result image after non-rigid registration CPD; orange represents the theoretical contour point cloud. Purple represents the outline of the cloud. Green represents the actual outline of the point cloud. .

[0041] Figure 7 This is the image after B-spline fitting. Detailed Implementation

[0042] The present invention will be further described below through specific embodiments.

[0043] like Figure 1 As shown, the method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model includes the following steps:

[0044] Step S1: Scan the actual wheel hub workpiece to obtain the set of wheel spoke contour points, and then obtain the actual contour point cloud P of the wheel spokes. s Based on the theoretical model of the wheel hub, multiple spoke edges are identified, and the contours of each spoke segment are uniformly discretized to obtain the theoretical contour point cloud P. t ;

[0045] Specifically, the steps include the following:

[0046] Step S11: Use a binocular surface scanning 3D camera to scan the wheel hub and generate 3D point cloud data of the wheel hub. Statistical filtering is used for 3D point cloud data Processing is performed to eliminate isolated noise points introduced during data acquisition;

[0047] More specifically, statistical filtering is applied to 3D point cloud data. Statistical analysis within the neighborhood of each point is used to identify and remove outliers that deviate too much from local geometric features, thereby improving the overall quality and reliability of point cloud data.

[0048] Step S12: Use voxel grid filtering to uniformly downsample the statistically filtered point cloud data to obtain point cloud data. ;

[0049] Specifically, the three-dimensional space is first divided into a series of voxel grids of equal size. Using the point set in each voxel grid, all points in the voxel grid are replaced with a representative point (such as the centroid) by means of the centroid substitution method or the average value method, thereby improving the execution efficiency of subsequent registration and trajectory planning.

[0050] Step S13: Based on point cloud data The relationship between the normal directions of each point and its neighboring points is used to determine whether the point lies on the spoke contour boundary, thus obtaining the actual contour point cloud P.s ;

[0051] The normal direction relationship between each point and its neighboring points is calculated using the BoundaryEstimation algorithm from the PCL open-source algorithm library. The boundary contour extraction performed in this step accurately identifies the actual boundary of the workpiece to be ground, such as... Figure 2 As shown.

[0052] Step S14: Read multiple triangular facets from the STL file of the CAD theoretical model of the storage hub, wherein the boundary of adjacent triangular facets is part of the outline of the theoretical model.

[0053] Step S15: For edges E with shared edges AB Two adjacent triangular facets T A and T B According to the formula Calculate T A and T B Angle between normal vectors ,in, and T respectively A and T B The normal vector;

[0054] Step S16, when the included angle When the angle is less than the set threshold It is believed that the shared edge E AB Located on the geometric edge of the spoke theory model, i.e., the shared edge E AB For each profile line, according to a pre-defined discrete interval, a segment of the wheel spokes is defined. It is uniformly discretized into a series of theoretical contour discrete points, and all discrete points form the theoretical contour point cloud P. t ,like Figure 3 As shown. This equidistant discretization process enables accurate geometric alignment and deviation analysis with the actual measured point cloud contour.

[0055] Step S2: Perform ICP registration between the actual contour point cloud and the theoretical contour point cloud, transforming the theoretical model to the measurement coordinate system of the actual wheel hub workpiece, to obtain the transformed theoretical contour point cloud. The actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. ;

[0056] ICP (Iterative Closest Point) registration can determine the precise pose of the actual workpiece in the measurement coordinate system, achieving accurate alignment between the theoretical model and the actual measurement data. The effect after registration is as follows: Figure 4 As shown.

[0057] Specifically, the steps include the following:

[0058] Step S21: Transfer the actual contour point cloud P s Set as the target point cloud, and use the theoretical contour point cloud P t Set as the source point cloud, in each iteration, for the theoretical contour point cloud P t Each point in the actual contour point cloud P s Search for the point with the closest Euclidean distance in the middle and establish a point-to-point correspondence;

[0059] Step S22: Based on the established point-to-point correspondence, construct an optimization function with mean square error as the objective. Solve for the optimal rigid body transformation parameters that minimize the mean square error using singular value decomposition or quaternion method. The optimal rigid body transformation parameters include rotation matrix and translation vector.

[0060] Step S23: Apply the rotation matrix and translation vector to the theoretical contour point cloud P. t The spatial positions of each point are updated, and the process proceeds to step S21 for the next iteration. The iteration terminates when the mean square error change falls below a preset threshold or the maximum number of iterations is reached, yielding the final rotation matrix R and translation vector T. Using this final rotation matrix R and translation vector T, the CAD theoretical model is accurately converted to the measurement coordinate system of the actual workpiece, forming a unified spatial reference and providing a precise coordinate foundation for subsequent steps. The converted theoretical contour point cloud is represented as follows: .

[0061] Step S24: Transfer the actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. ;

[0062] Because burrs remain on the spoke edges during the wheel hub manufacturing process, and the surface is highly reflective, the edge point cloud quality is poor when scanned by a 3D camera, resulting in abrupt and uneven contours. Therefore, it is necessary to filter out abnormal areas and remove abnormal point clouds. Specifically: for P s Each point in The KdTreeFLANN algorithm from the PCL algorithm library is used to calculate the theoretical contour point cloud. Neutral nearest point Calculation points With point The difference between the Euclidean distance Dis and the Z coordinate. If the Euclidean distance Dis is less than the set distance threshold And the difference Less than the deviation threshold along the Z-axis Then point If a point is considered valid, it is considered invalid. Filtering out all invalid points yields the actual contour point cloud. ,like Figure 5 As shown.

[0063] Step S3: Implement the actual contour point cloud using the CPD algorithm. With theoretical outline point cloud Non-rigid deformation registration between them yields the contour point cloud. Based on this contour point cloud Generate polishing trajectory.

[0064] Actual contour point cloud There will be discontinuities in the contour, and due to the deformation caused by wheel hub processing, there will be deformation deviations and missing point cloud deviations between the scanned wheel spoke point cloud and the theoretical model point cloud. Therefore, the CPD (Coherent Point Drift) algorithm is used to realize the actual contour point cloud. With theoretical outline point cloud Non-rigid deformation registration between the two point clouds is used to achieve high-precision fusion of the two point clouds, such as... Figure 6 As shown.

[0065] Specifically, the steps include the following:

[0066] Step S31: Convert the actual contour point cloud Set as target point cloud, theoretical contour point cloud The point cloud to be registered is set as the core model of Gaussian mixture theory, and the actual contour point cloud is then used. Viewed as a point cloud of theoretical outline Generated by sampling from a Gaussian mixture distribution; motion consistency constraints are introduced to refine the theoretical contour point cloud. A displacement field is applied to each point in the cloud to gradually approximate the actual contour point cloud while maintaining the consistency of the topological structure. Location;

[0067] Step S32: Calculate the actual contour point cloud Each point in the middle corresponds to a theoretical contour point cloud. The posterior probabilities of each Gaussian component are used to construct a soft assignment matrix for point correspondences. Based on the current posterior probabilities, the optimal displacement vector is solved by minimizing the negative log-likelihood function and adding motion smoothness constraints. The theoretical contour point cloud is then updated based on this optimal displacement vector. The positions of each point are determined, and the process proceeds to step S32 for the next iteration, re-estimating the parameters of the Gaussian mixture theory model. The iteration terminates when the registration error is below a preset threshold or the maximum number of iterations is reached, yielding the non-rigidly registered contour point cloud. ;

[0068] The theoretical model's contour point cloud is continuous and complete. After non-rigid registration, the theoretical model's point cloud has maximized its overlap with the scanned actual point cloud, effectively overcoming the discontinuity and deviation problems caused by processing deformation and scanning defects. This registered contour point cloud... This is the point cloud after fusion.

[0069] Step S33, due to the contour point cloud As a closed loop structure, the contour point cloud is calculated first. geometric centroid Subsequently, the contour point cloud was calculated separately. various points in the middle With geometric centroid The connecting vector between them is determined by the angle between the connecting vector and a preset reference direction (such as the positive X-axis direction). Size, for contour point cloud The points are arranged in ascending order to form a continuous and non-intersecting ordered list of points along the circumferential contour of the wheel spokes. ;

[0070] Step S34: Arrange the ordered points As control vertices, a theoretical model of a non-uniform rational B-spline (NURBS) curve is constructed. By specifying the curve order, node vectors, and control vertex weights, a non-uniform rational B-spline curve is fitted and generated. This curve is a smooth, continuous, and geometrically accurate closed contour curve. To meet the grinding system's requirement for equally spaced trajectory points, parameters are sampled along this B-spline curve at a preset fixed step size to obtain a series of discrete trajectory point sets with equal arc length spacing, ensuring smooth motion and constant speed during the grinding process. Figure 7 As shown, the discrete trajectory point set is output to the control system as the actual executable grinding path trajectory, which drives the grinding tool to perform high-precision, continuous, and stable automated grinding operations along the spokes.

[0071] In this invention, the terms "first," "second," and "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. The use of terms such as "upper," "lower," "left," "right," "front," and "rear" to indicate orientation or positional relationships is based on the orientation or positional relationships shown in the accompanying drawings and is only for the convenience of describing the invention, not to indicate or imply that the device referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation on the scope of protection of this invention. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0072] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0073] The above are merely specific embodiments of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial modifications made to the present invention using this concept shall be considered as infringing upon the protection scope of the present invention.

Claims

1. A method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model, characterized in that: Includes the following steps: Step S1: Scan the actual wheel hub workpiece to obtain the set of wheel spoke contour points, and then obtain the actual contour point cloud P of the wheel spokes. s Based on the theoretical model of the wheel hub, multiple spoke edges are identified, and the contours of each spoke segment are uniformly discretized to obtain the theoretical contour point cloud P. t ; Step S2: Perform ICP registration between the actual contour point cloud and the theoretical contour point cloud, transforming the theoretical model to the measurement coordinate system of the actual wheel hub workpiece, to obtain the transformed theoretical contour point cloud. The actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. ; Step S3: Implement the actual contour point cloud using the CPD algorithm. With theoretical outline point cloud Non-rigid deformation registration between them yields the contour point cloud. Based on this contour point cloud Generate polishing trajectory; In step S3, the contour point cloud is obtained through the following steps. : Step S31: Using the Gaussian mixture theory model as the core, convert the actual contour point cloud... Viewed as a point cloud of theoretical outline Generated by sampling from a Gaussian mixture distribution; motion consistency constraints are introduced to refine the theoretical contour point cloud. A displacement field is applied to each point in the cloud to gradually approximate the actual contour point cloud while maintaining the consistency of the topological structure. Location; Step S32: Calculate the actual contour point cloud Each point in the middle corresponds to a theoretical contour point cloud. The posterior probabilities of each Gaussian component are obtained. Based on these posterior probabilities, the optimal displacement vector is solved by minimizing the negative log-likelihood function and applying motion smoothness constraints. The theoretical contour point cloud is then updated based on this optimal displacement vector. The positions of each point are determined, and the process proceeds to step S32 for the next iteration. After the iteration is completed, the contour point cloud is obtained. ; In step S3, based on the contour point cloud The polishing trajectory is obtained through the following steps: Step S33: Calculate the contour point cloud geometric centroid Calculate the contour point cloud respectively various points in the middle With geometric centroid The connecting vectors between them are determined by the angle between the connecting vectors and the preset reference direction. Size, for contour point cloud The points are arranged in ascending order to form a continuous and non-intersecting ordered list of points along the circumferential contour of the wheel spokes. ; Step S34: Arrange the ordered points As the control vertex, a non-uniform rational B-spline curve is fitted, and parameters are sampled along the B-spline curve according to a preset fixed step size to obtain a series of discrete trajectory point sets with equal arc length spacing. The grinding trajectory is the discrete trajectory point set.

2. The method for generating wheel spoke grinding trajectory based on the fusion of point cloud and theoretical model according to claim 1, characterized in that: In step S1, a binocular surface scanning 3D camera is used to scan the wheel hub to generate three-dimensional point cloud data of the wheel hub. .

3. The method for generating wheel spoke grinding trajectory based on the fusion of point cloud and theoretical model according to claim 2, characterized in that: In step S1, the actual contour point cloud P is obtained through the following steps. s : Step S11: Apply statistical filtering to the 3D point cloud data. Processing is performed to eliminate isolated noise points introduced during data acquisition; Step S12: Use voxel grid filtering to uniformly downsample the statistically filtered point cloud data to obtain point cloud data. ; Step S13: Based on point cloud data The relationship between the normal directions of each point and its neighboring points is used to determine whether the point lies on the spoke contour boundary, thus obtaining the actual contour point cloud P. s .

4. The method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model according to claim 1, 2, or 3, characterized in that: In step S1, the theoretical contour point cloud P is obtained through the following steps. t : Step S14: Read multiple triangular facets from the STL file storing the CAD theoretical model of the wheel hub; Step S15: For edges with shared edge E AB Two adjacent triangular facets T A and T B According to the formula Calculate T A and T B Angle between normal vectors ,in, and T respectively A and T B The normal vector; Step S16, when the included angle When the angle is less than the set threshold, the shared edge E AB Let P be a segment of the wheel spoke profile. For each profile, it is uniformly discretized according to a pre-defined discrete interval to obtain the theoretical profile point cloud P. t .

5. The method for generating wheel spoke grinding trajectories based on the fusion of point cloud and theoretical model according to claim 1, 2, or 3, characterized in that: Step S2 specifically includes the following steps: Step S21: For the theoretical contour point cloud P t Each point in the actual contour point cloud P s Search for the point with the closest Euclidean distance in the middle and establish a point-to-point correspondence; Step S22: Based on the established point-to-point correspondence, construct an optimization function with mean square error as the objective, and solve for the optimal rigid body transformation parameters that minimize the mean square error. The optimal rigid body transformation parameters include the rotation matrix and translation vector. Step S23: Apply the rotation matrix and translation vector to the theoretical contour point cloud P. t Update the spatial position of each point, proceed to step S21 for the next iteration, until the mean square error change is lower than a preset threshold or the maximum number of iterations is reached, then terminate the iteration, obtaining the final rotation matrix R and translation vector T. The transformed theoretical contour point cloud is represented as follows: ; Step S24: Transfer the actual contour point cloud P s With theoretical outline point cloud Compare and remove actual contour point cloud P s The outliers in the data are used to obtain the filtered actual contour point cloud. .

6. The method for generating wheel spoke grinding trajectory based on the fusion of point cloud and theoretical model according to claim 5, characterized in that: Step S24 specifically involves: For P... s Each point in Calculate theoretical contour point cloud Neutral nearest point Calculation points With point The difference between the Euclidean distance Dis and the Z coordinate. If the Euclidean distance Dis is less than the set distance threshold and the difference is... If the deviation is less than the threshold along the Z-axis, then the point If a point is considered valid, it is considered invalid. Filtering out all invalid points yields the actual contour point cloud. .

Images