Mixed reality-based vehicle chassis maintenance training method

Abstract

The present application relates to the technical field of mixed reality, in particular to a vehicle chassis maintenance practical training method based on mixed reality, comprising the following steps: constructing a chassis surface directed distance field data according to a vehicle chassis three-dimensional grid model in a mixed reality space; extracting three-dimensional space coordinates of a plurality of discrete pipeline interaction points input by a user in the mixed reality space in the chassis surface directed distance field data and performing interpolation to obtain a plurality of spatial interpolation nodes. The present application combines topological adjacent connection vertices to perform smoothing processing, uses compressed deformation vertex coordinates and non-penetrating pipeline peripheral vertex coordinates to replace original vertex coordinates, so that the generated deformation adaptive three-dimensional pipeline grid not only maintains the pipeline space path, but also takes into account the chassis surface avoidance relationship and geometric appearance continuity, thereby improving the consistency of pipeline path construction, space collision avoidance, deformation feedback and three-dimensional display in vehicle chassis maintenance practical training.

Description

Technical Field

[0001] This invention relates to the field of mixed reality technology, and more particularly to a mixed reality-based training method for vehicle chassis maintenance. Background Technology

[0002] Mixed reality technology is a technology that integrates real physical space, virtual 3D content, spatial positioning perception, human-computer interaction perception, and real-time rendering display.

[0003] Existing technologies primarily focus on the integration of real physical space, virtual 3D content, spatial positioning perception, human-computer interaction perception, and real-time rendering display in practical operation. While they can achieve the overlay display and basic interaction of virtual objects in a real environment, when faced with complex curved surface boundaries, narrow assembly spaces, and pipeline attachment paths in vehicle chassis pipeline maintenance training, they typically rely on virtual model position display and manual drag-and-drop interaction to maintain spatial relationships. The lack of continuous constraints between virtual pipelines and the vehicle chassis surface based on boundary distance, internal / external orientation, and vertex-level spatial states easily leads to problems such as pipelines penetrating the chassis structure, floating on the target surface, cross-sectional distortion at turns, or discontinuous connecting surfaces. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a vehicle chassis maintenance training method based on mixed reality.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a vehicle chassis maintenance training method based on mixed reality, comprising the following steps: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, oriented distance field data of the chassis surface is constructed. Extract the three-dimensional spatial coordinates of multiple discrete pipeline interaction points input by the user in the mixed reality space from the directed distance field data on the chassis surface and interpolate them to obtain multiple spatial interpolation nodes. Generate a three-dimensional spatial pipeline curve based on the multiple spatial interpolation nodes. A local pipeline frame is generated based on the tangent vector, principal normal vector, and secondary normal vector corresponding to each spatial interpolation node on the three-dimensional pipeline curve. The coordinates of the extruded vertices are obtained by projecting the preset two-dimensional circular contour vertices onto the local frame coordinate system of the pipeline. The vertex connection relationship is constructed based on the extruded vertices coordinates to generate the initial tubular three-dimensional geometric mesh. The coordinates of all outer vertices of the pipelines within the initial tubular three-dimensional geometric mesh are input into the directed distance field data of the chassis surface for sampling. The outer vertices of the target pipelines with negative distance field values ​​are extracted, and the corresponding coordinates of the outer vertices of the target pipelines are obtained. The displacement compensation value is calculated based on the spatial gradient vector corresponding to the outer vertex of the target pipeline, the sampled distance field value, and the preset displacement constant, and a pipeline anti-penetration displacement vector is generated. The coordinates of the outer vertex of the target pipeline are combined with the corresponding pipeline anti-penetration displacement vector to obtain the penetration compensation updated vertex; the topological adjacent connected vertices of the penetration compensation updated vertex are extracted and smoothed to generate the coordinates of the compressive deformation vertex; The coordinates of the compressed deformation vertices and the coordinates of the outer vertices of the unpenetrated pipeline with non-negative distance field values ​​are used to replace the original vertex coordinates in the initial tubular three-dimensional geometric mesh to generate a deformation-adaptive three-dimensional pipeline mesh.

[0006] Preferably, the step of acquiring the oriented distance field data of the chassis surface is as follows: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, the vertex coordinates of the boundary triangular facets of the 3D mesh model of the vehicle chassis are extracted. The space of the 3D mesh model of the vehicle chassis is divided into voxel nodes. The center coordinates of each voxel node are read one by one. The spatial distance from the center coordinates of the voxel node to the projection point of each boundary triangular facet is calculated. The minimum spatial distance is selected as the nearest Euclidean distance. The spatial direction sign is recorded according to the inner and outer positions of the voxel node center coordinates relative to the model boundary. The spatial direction sign and the nearest Euclidean distance are combined to generate the directional distance field data of the chassis surface.

[0007] Preferably, the step of obtaining the three-dimensional spatial pipeline curve is as follows: Read multiple discrete pipeline interaction points continuously input by the user in the mixed reality space, record the discrete pipeline interaction point identifiers according to the input order, retrieve the voxel neighborhood position of each discrete pipeline interaction point in the directed distance field data on the chassis surface, extract the three-dimensional spatial coordinates corresponding to the discrete pipeline interaction points, continuously interpolate the three-dimensional spatial coordinates according to the input order, fill in the spatial transition position between adjacent three-dimensional spatial coordinates, and obtain multiple spatial interpolation nodes; Establish a node arrangement sequence according to the input order of the multiple spatial interpolation nodes, read the three-dimensional spatial coordinates of adjacent spatial interpolation nodes one by one, verify whether the connection direction between adjacent spatial interpolation nodes is consistent with the pipeline input direction, establish a continuous spatial connection relationship for adjacent spatial interpolation nodes with consistent connection directions, and readjust the connection order of adjacent spatial interpolation nodes with inconsistent connection directions according to the node arrangement sequence to generate a three-dimensional spatial pipeline curve.

[0008] Preferably, the step of obtaining the pipeline local marker is as follows: The three-dimensional spatial coordinates of each spatial interpolation node on the three-dimensional pipeline curve are read one by one. The difference between the three-dimensional spatial coordinates of the previous spatial interpolation node and the next spatial interpolation node is extracted according to the interpolation order. The tangent vector corresponding to each spatial interpolation node is determined. The change in tangent vector of adjacent spatial interpolation nodes is read. The principal normal vector is determined according to the change in tangent vector. The perpendicularity of tangent vector and principal normal vector is determined. The binormal vector is extracted according to the perpendicularity determination result. The tangent vector, principal normal vector and binormal vector are adjusted to mutually perpendicular unit directions to generate a local pipeline frame.

[0009] Preferably, the step of obtaining the initial tubular three-dimensional geometric mesh is as follows: Based on the pipeline local frame, spatial interpolation nodes are located one by one along the three-dimensional spatial pipeline curve. The coordinate system of the pipeline local frame corresponding to each spatial interpolation node is read. The planar coordinates of the preset two-dimensional circular contour vertices are called. According to the arrangement order of the two-dimensional circular contour vertices in the plane, each two-dimensional circular contour vertex is mapped to the pipeline local frame coordinate system of the corresponding spatial interpolation node. The position of the mapped spatial vertex is recorded to obtain the extrusion vertex coordinates. Divide adjacent sections according to the interpolation order of the three-dimensional spatial pipeline curve, read the coordinates of the extrusion vertices with the same number in each adjacent section, establish the longitudinal vertex connection relationship along the pipeline direction, read the coordinates of the extrusion vertices with adjacent numbers in the same section, establish the circumferential vertex connection relationship along the circular contour direction, determine the four-point closed facets according to the longitudinal vertex connection relationship and the circumferential vertex connection relationship, merge all four-point closed facets, and generate the initial tubular three-dimensional geometric mesh.

[0010] Preferably, the step of obtaining the outer vertices of the target pipeline is as follows: The pipeline perimeter vertex indices inside the initial tubular 3D geometric mesh are read one by one. The corresponding pipeline perimeter vertex coordinates are extracted according to the pipeline perimeter vertex indices. The pipeline perimeter vertex coordinates are located within the range of adjacent voxel nodes of the directed distance field data on the chassis surface. The distance field values ​​corresponding to the adjacent voxel nodes are read. The sampled distance field values ​​are calculated according to the spatial proportion of the pipeline perimeter vertex coordinates within the range of the adjacent voxel nodes. The pipeline perimeter vertex coordinates with negative sampled distance field values ​​are filtered to obtain the target pipeline perimeter vertex.

[0011] Preferably, the step of obtaining the pipeline penetration-resistant displacement vector is as follows: Read the three-dimensional spatial coordinates of the outer vertex of the target pipeline in the directed distance field data on the chassis surface. Select positive offset sampling positions and negative offset sampling positions along the three coordinate axes of the three-dimensional spatial coordinates respectively. Extract the distance field values ​​corresponding to the positive offset sampling positions and the distance field values ​​corresponding to the negative offset sampling positions respectively. Calculate the difference between the positive distance field values ​​and the negative distance field values ​​in the same coordinate axis direction. Combine the difference values ​​in the three coordinate axis directions according to the coordinate axis order to obtain the spatial gradient vector. The sampled distance field values ​​corresponding to the outer vertices of the target pipeline are called, the absolute values ​​of the sampled distance field values ​​are extracted, and the absolute values ​​of the sampled distance field values ​​are superimposed with a preset shift constant to obtain the displacement compensation value. The amplitude ratio of each directional component in the spatial gradient vector is calculated, and the spatial gradient vector is normalized according to the amplitude ratio of each directional component. The normalized spatial gradient vector is multiplied by the displacement compensation value to generate the pipeline anti-penetration displacement vector.

[0012] Preferably, the steps for obtaining the deformation-adaptive three-dimensional pipeline mesh are as follows: The coordinates of the outer vertices of the target pipeline are matched one by one with the corresponding pipeline anti-penetration displacement vector according to the same vertex index. The horizontal coordinate value, vertical coordinate value, and vertical coordinate value in the coordinates of the outer vertices of the target pipeline are read, and the displacement components in the corresponding directions in the pipeline anti-penetration displacement vector are superimposed respectively. The vertex index is written back for the superimposed horizontal coordinate value, vertical coordinate value, and vertical coordinate value, keeping the vertex index and topology connection number unchanged, to obtain the penetration compensation updated vertex. Based on the vertex connection relationships stored in the initial tubular three-dimensional geometric mesh, the topological adjacent connected vertices of the penetration compensation update vertex are extracted. The three-dimensional coordinate values ​​of all topological adjacent connected vertices are read one by one. The average coordinate values ​​of all topological adjacent connected vertices in the horizontal, vertical and vertical directions are calculated respectively to form the geometric center of all topological adjacent connected vertices. The vector difference from the penetration compensation update vertex to the geometric center of all topological adjacent connected vertices is calculated. The normal Laplace smoothing calculation is performed along the direction corresponding to the vector difference to generate the coordinates of the compressive deformation vertex. Read the coordinates of the outer vertices of the unpenetrated pipeline when the sampled distance field values ​​are non-negative. According to the vertex indices of the original vertex coordinates in the initial tubular 3D geometric mesh, replace the original vertex coordinates of the sampled distance field when the sampled distance field values ​​are negative with the vertex coordinates of the compressive deformation vertex. Use the vertex coordinates of the outer vertices of the unpenetrated pipeline to retain the original vertex coordinates of the sampled distance field when the sampled distance field values ​​are non-negative. Rewrite the 3D coordinate values ​​corresponding to all vertex indices to generate a deformation adaptive 3D pipeline mesh.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: This invention constructs oriented distance field data of the vehicle chassis surface based on a 3D mesh model of the vehicle chassis within a mixed reality space. This establishes computable constraints on the vehicle chassis surface boundaries, internal and external spatial orientation signs, and nearest Euclidean distance. Multiple discrete pipeline interaction points input by the user are then mapped to 3D spatial coordinates and interpolated, transforming discrete interaction behaviors into continuous 3D spatial pipeline curves. This improves the spatial continuity and interactive controllability of pipeline path generation. Furthermore, a local pipeline frame is formed based on the tangent vectors, principal normal vectors, and secondary normal vectors of the spatial interpolation nodes. The vertices of the 2D circular surface contour are projected into the corresponding coordinate system to generate extrusion vertex coordinates, ensuring that the pipeline cross-section maintains a stable orientation along the curve direction, reducing pipeline torsion, cross-sectional misalignment, and geometric fractures. The problem of cracking is addressed by sampling the coordinates of the outer vertices of the pipeline using the directional distance field data of the chassis surface. This identifies the outer vertices of the target pipeline when the distance field value is negative. Based on the spatial gradient vector, the sampled distance field value, and the shift constant, a pipeline anti-penetration displacement vector is generated, allowing the penetration position to be compensated and updated along the outer normal of the chassis surface. Combined with topological adjacent connection vertices, smoothing is performed. The coordinates of the deformed vertices and the outer vertices of the non-penetrated pipeline are used to replace the original vertex coordinates. This ensures that the generated deformation adaptive 3D pipeline mesh maintains the pipeline spatial path while taking into account the avoidance relationship and geometric appearance continuity of the chassis surface. This improves the consistency of pipeline path construction, spatial collision avoidance, deformation feedback, and 3D display in vehicle chassis maintenance training. Attached Figure Description

[0014] Figure 1 A schematic diagram of the distance field values ​​obtained by sampling the vertices of the pipeline perimeter; Figure 2 This is a schematic diagram showing the displacement compensation values ​​corresponding to the outer vertices of the target pipeline. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] Please see Figure 1-2 This invention provides a technical solution: a vehicle chassis maintenance training method based on mixed reality, comprising the following steps: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, the oriented distance field data of the chassis surface is constructed; the 3D spatial coordinates of multiple discrete pipeline interaction points input by the user in the mixed reality space are extracted in the oriented distance field data of the chassis surface and interpolated to obtain multiple spatial interpolation nodes; and 3D spatial pipeline curves are generated based on the multiple spatial interpolation nodes. A local frame for the pipeline is generated based on the tangent vector, principal normal vector, and secondary normal vector corresponding to each spatial interpolation node on the three-dimensional pipeline curve. The coordinates of the extruded vertices are obtained by projecting the preset two-dimensional circular contour vertices onto the coordinate system of the local frame for the pipeline. The vertex connection relationship is constructed based on the extruded vertices coordinates to generate the initial tubular three-dimensional geometric mesh. The coordinates of all outer vertices of the pipelines within the initial tubular 3D geometric mesh are input into the directed distance field data on the chassis surface for sampling. The outer vertices of the target pipelines with negative distance field values ​​are extracted, and the corresponding coordinates of the outer vertices of the target pipelines are obtained. The displacement compensation value is calculated based on the spatial gradient vector corresponding to the outer vertices of the target pipelines, the sampled distance field values, and the preset displacement constant, and the pipeline anti-penetration displacement vector is generated. The coordinates of the outer vertices of the target pipeline are combined with the corresponding pipeline anti-penetration displacement vector to obtain the penetration compensation updated vertex; the topological adjacent connected vertices of the penetration compensation updated vertex are extracted and smoothed to generate the coordinates of the compressive deformation vertex; the coordinates of the compressive deformation vertex, and the coordinates of the outer vertices of the non-penetrated pipeline in the state where the sampled distance field value is non-negative, are used to replace the original vertex coordinates in the initial tubular three-dimensional geometric mesh to generate a deformation adaptive three-dimensional pipeline mesh.

[0017] The steps for obtaining the oriented distance field data of the chassis surface are as follows: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, the vertex coordinates of the boundary triangular facets of the 3D mesh model of the vehicle chassis are extracted. The space of the 3D mesh model of the vehicle chassis is divided into voxel nodes. The center coordinates of each voxel node are read one by one. The spatial distance from the center coordinates of the voxel node to the projection point of each boundary triangular facet is calculated. The minimum spatial distance is selected as the nearest Euclidean distance. The spatial direction sign is recorded according to the inner and outer positions of the voxel node center coordinates relative to the model boundary. The spatial direction sign and the nearest Euclidean distance are combined to generate the directional distance field data of the chassis surface.

[0018] Specifically, based on the 3D mesh model of the vehicle chassis in mixed reality space, the coordinates of all boundary triangular faces and their vertices constituting the model are first extracted. Then, the space in which the model is located is divided into 3D meshes, generating a series of closely arranged voxel nodes. The fineness of the meshing is set according to the model size. For example, for a chassis with dimensions of 5m x 2m x 1m, it can be divided into a 512x256x128 voxel mesh, with a voxel size of approximately 1 cm. Next, each voxel node in the mesh is traversed one by one, and the 3D spatial coordinates of its center point are read. For each center point, the shortest spatial distance from it to all boundary triangular faces is calculated. Specifically, the projection point from the center point to the plane containing each triangular face is calculated. If the projection point is located inside the triangular facet, ... The projected distance is then used as a candidate distance. If it is not found, the shortest distance from the center point to the three sides of the triangular facet is calculated as the candidate distance. The minimum value is selected from all candidate distances and recorded as the nearest Euclidean distance. Then, the position of the voxel node center point relative to the model is determined. A ray intersection test method is used to emit a virtual ray from the center point along any fixed direction (e.g., the positive Z-axis direction). The number of intersections between the ray and the triangular facet of the model boundary is counted. If the number of intersections is odd, the point is determined to be inside the model, and the spatial direction sign is recorded as -1. If it is even (including 0), the point is determined to be outside the model, and the spatial direction sign is recorded as +1. Finally, the spatial direction sign corresponding to each voxel node is multiplied by the calculated nearest Euclidean distance to generate the oriented distance field data of the chassis surface.

[0019] The steps for obtaining three-dimensional spatial pipeline curves are as follows: Read multiple discrete pipeline interaction points continuously input by the user in the mixed reality space, record the discrete pipeline interaction point identifiers according to the input order, retrieve the voxel neighborhood position of each discrete pipeline interaction point in the directed distance field data on the chassis surface, extract the three-dimensional spatial coordinates corresponding to the discrete pipeline interaction points, continuously interpolate the three-dimensional spatial coordinates according to the input order, fill in the spatial transition position between adjacent three-dimensional spatial coordinates, and obtain multiple spatial interpolation nodes; Establish a node arrangement sequence according to the input order of multiple spatial interpolation nodes, read the three-dimensional spatial coordinates of adjacent spatial interpolation nodes one by one, verify whether the connection direction between adjacent spatial interpolation nodes is consistent with the pipeline input direction, establish a continuous spatial connection relationship for adjacent spatial interpolation nodes with consistent connection directions, and readjust the connection order of adjacent spatial interpolation nodes with inconsistent connection directions according to the node arrangement sequence to generate a three-dimensional spatial pipeline curve.

[0020] Specifically, the system reads multiple discrete pipeline interaction points selected by the user on the surface of a virtual chassis model using interactive devices (such as handheld controllers or gesture recognition) in the mixed reality space. Each interaction point is assigned a unique identifier starting from 0 and its input timestamp is recorded to ensure the accuracy of the input order. Then, for each recorded discrete pipeline interaction point, its 3D coordinates are used to perform a fast position search in the previously constructed directed distance field data on the chassis surface to determine its voxel unit and its eight surrounding neighboring voxels. Since the position of the interaction point is a continuous floating-point coordinate, while the directed distance field is discrete, the directed distance field value corresponding to the precise position of the interaction point is calculated using trilinear interpolation. Using the gradient information of the directed distance field, the position of the interaction point is finely adjusted along the gradient direction, so that it is attached to the isosurface with a value of zero in the distance field, i.e., the virtual surface of the vehicle chassis, to obtain accurate three-dimensional spatial coordinates. After the surface attachment of all points is completed, the three-dimensional spatial coordinate sequence arranged in the input order is processed by cubic B-spline interpolation or Catmull-Rom spline interpolation algorithm to generate a series of new points between every two consecutive three-dimensional spatial coordinates. The density of interpolation points is determined by the preset step size. For example, if the step size is set to 5 mm, enough points are generated between adjacent coordinates so that the distance between any two adjacent new points is less than 5 mm, thus filling the spatial transition positions between coordinate points and obtaining multiple spatial interpolation nodes.

[0021] An initial node arrangement sequence is established based on the input order of multiple spatial interpolation nodes. To verify and optimize the smoothness and rationality of the sequence, a global pipeline routing reference vector is first calculated, pointing from the start node to the end node of the sequence. Then, the system iterates through each pair of adjacent spatial interpolation nodes in the sequence, for example, node... and Calculate the local connectivity vectors between them. Then, the dot product of the local connection vector and the global direction reference vector is calculated, and the result is normalized to obtain the cosine value of the angle between the two. A direction consistency judgment threshold is set, which is empirically set, for example, 0.2, indicating that the maximum allowable angle between the local direction and the global direction is approximately 78.5 degrees. If the calculated cosine value is greater than the threshold, the connection direction of this pair of adjacent nodes is considered to be consistent with the overall pipeline input direction, and their connection relationship is retained. If the cosine value is less than or equal to the threshold, the connection direction is determined to be inconsistent, possibly due to a path inflection point or loop caused by user error. At this time, the connection order readjustment logic is triggered, and the system will temporarily disconnect the node. and The connection, and from Begin searching backwards, calculating sequentially. With subsequent nodes ( The cosine of the angle between the local vector and the global direction vector is calculated until the first node is found whose cosine value is greater than the direction consistency threshold. Then the node With this node Establish a new connection and discard intermediate nodes (from...). arrive After traversing and adjusting the entire node sequence, the resulting continuous, loop-free node sequence constitutes the three-dimensional spatial pipeline curve.

[0022] The steps for obtaining a local pipeline marker are as follows: The three-dimensional spatial coordinates of each spatial interpolation node on the three-dimensional pipeline curve are read one by one. The difference between the three-dimensional spatial coordinates of the previous spatial interpolation node and the next spatial interpolation node is extracted according to the interpolation order. The tangent vector corresponding to each spatial interpolation node is determined. The change of tangent vector of adjacent spatial interpolation nodes is read. The principal normal vector is determined according to the change of tangent vector. The perpendicularity of tangent vector and principal normal vector is determined. The binormal vector is extracted according to the perpendicularity determination result. The tangent vector, principal normal vector and binormal vector are adjusted to mutually perpendicular unit directions to generate a local pipeline frame.

[0023] Specifically, the 3D spatial coordinates of each spatial interpolation node on the 3D pipeline curve are read one by one. In order to construct a stable local coordinate system without gimbal lock problem at each node, the rotation minimization frame (RMF) technique is used to generate the local frame of the pipeline. First, for any non-endpoint spatial interpolation node on the curve... By using its central difference method, that is, by utilizing its previous node and the next node Calculate the difference vector of coordinates And after normalizing the vector, it is used as a node. Tangent vector at point For the starting and ending nodes, forward and backward differences are used respectively. Then, the normal vector of the first node is initialized, for example, by performing a cross product between the initial tangent vector of the curve and the Y-axis (0, 1, 0) of the world coordinate system. Then, starting from the second node, the normal vectors of all subsequent nodes are calculated using parallel transmission. Specifically, for each node... Its principal normal vector The principal normal vector of the previous node Based on arrive The result is obtained by rotational transformation and then subjected to Schmidt orthogonalization to ensure... and Strict perpendicularity ensures that the normal vector experiences minimal twisting as it travels along the curve, further enhanced by adjusting the tangent vector. With principal normal vector Perform cross product operation ( ), to obtain the binormal vector Then normalize the tangent vector Principal normal vector and binormal vector Taken together, these three mutually perpendicular unit vectors together form the node Localized pipeline markers at the location.

[0024] The steps for obtaining the initial tubular 3D geometric mesh are as follows: Based on the local frame of the pipeline, spatial interpolation nodes are located one by one along the three-dimensional spatial pipeline curve. The coordinate system of the local frame of the pipeline corresponding to each spatial interpolation node is read. The planar coordinates of the preset two-dimensional circular contour vertices are called. According to the arrangement order of the two-dimensional circular contour vertices in the plane, each two-dimensional circular contour vertex is mapped to the local frame coordinate system of the corresponding spatial interpolation node. The position of the spatial vertex after mapping is recorded to obtain the coordinates of the extruded vertex. Divide adjacent sections according to the interpolation order of the three-dimensional spatial pipeline curve, read the coordinates of the extrusion vertices with the same number in each adjacent section, establish the longitudinal vertex connection relationship along the pipeline direction, read the coordinates of the extrusion vertices with adjacent numbers in the same section, establish the circumferential vertex connection relationship along the circular contour direction, determine the four-point closed facets according to the longitudinal vertex connection relationship and the circumferential vertex connection relationship, merge all four-point closed facets, and generate the initial tubular three-dimensional geometric mesh.

[0025] Specifically, based on the local framework of the pipeline previously generated for each spatial interpolation node, a scan generation operation of the tubular model is performed. First, a standard two-dimensional circular contour is defined, which consists of a set of preset vertices. For example, on a two-dimensional plane centered at the origin, 16 vertices are defined by parametric equations, and their planar coordinates are... From the formula and Given, among which The value range is from 0 to 15. The preset pipe radius, for example, is set to 15 millimeters. This set of two-dimensional vertices defines the shape of the pipe's cross-section. Then, each spatial interpolation node is visited sequentially along the three-dimensional pipe curve. At each node, read its corresponding local pipeline frame, which is composed of tangent vectors. Principal normal vector and binormal vector The composition is defined for each vertex of a predefined two-dimensional circular contour. This maps it from a two-dimensional plane to a three-dimensional space. The specific mapping rule is to map the two-dimensional coordinates... Considered as being based on the principal normal and secondary normal Using the coordinate components of the basis vectors on the plane, the position of the contour vertex in three-dimensional space is calculated. This calculation method involves using the node positions... Adding the offset composed of the contour coordinates and the normal vector, i.e. Performing this mapping on all 16 contour vertices yields a vertex located at node For a circular cross-section that is perpendicular to the pipeline direction, record the positions of all cross-section vertices generated by all spatial interpolation nodes to obtain the coordinates of the extruded vertex.

[0026] Following the interpolation order of the 3D spatial pipeline curves, the extruded vertex coordinates generated in the previous process are considered as a series of continuous cross-sections. Each cross-section contains the same number of vertices, for example, 16, and has a circumferential number from 0 to 15. To construct the topology of the tubular mesh, it is necessary to establish the connection relationships between vertices. First, the longitudinal vertex connection relationships along the pipeline direction are established. Specifically, for any two adjacent cross-sections (e.g., the first...), the connection relationships between vertices are... The and the first (a cross section), connecting the vertices with the same circumferential number on them, for example, connecting the first cross section... The first section The vertex and the first The first section Establish connections between vertices, performing this operation for all cyclic vertices numbered 0 to 15. Then, establish cyclic vertex connections within the same cross section; that is, within any cross section, connect vertices with adjacent cyclic vertices. For example, connect the 1st vertex... The vertex and the first Connect the vertices, and connect the last vertex (number 15) to the first vertex (number 0) to form a closed loop. Through these two connection relationships, a four-point closed face consisting of four vertices can be uniquely identified. These four vertices are the vertices of the first vertex (number 15) and the last vertex (number 0). The first section The vertex, the first The first section The vertex, the first The first section +1 vertex and the first The first section Each vertex is determined by traversing all cross-section indices. and circular index This process generates all possible four-point closed patches, splits each four-point closed patch into two triangular patches, merges all these triangular patches, and finally generates a complete, closed initial tubular 3D geometric mesh.

[0027] The steps to obtain the outer vertices of the target pipeline are as follows: Read the pipe perimeter vertex indices one by one inside the initial tubular 3D geometric mesh, extract the corresponding pipe perimeter vertex coordinates according to the pipe perimeter vertex index, locate the pipe perimeter vertex coordinates within the range of adjacent voxel nodes of the directed distance field data on the chassis surface, read the distance field values ​​corresponding to the adjacent voxel nodes, calculate the sampled distance field values ​​according to the spatial proportion of the pipe perimeter vertex coordinates within the range of adjacent voxel nodes, filter the pipe perimeter vertex coordinates with negative sampled distance field values, and obtain the target pipe perimeter vertex.

[0028] Specifically, the index numbers of all outer vertices of the pipelines within the initial tubular 3D geometric mesh are read one by one, and the 3D spatial coordinates of each vertex are extracted based on the index. For each vertex coordinate, its distance value in the directed distance field data on the chassis surface needs to be queried to determine whether it has penetrated the chassis model. Since the vertex coordinates are continuous, while the directed distance field data is stored in a discrete voxel mesh, trilinear interpolation is used to obtain accurate distance field sampling values. The specific operation is as follows: first, the voxel cell in which the vertex is located is located based on its 3D coordinates. This cell is surrounded by 8 adjacent voxel nodes. The distance field values ​​stored in each of these 8 nodes are read, and then the relative position of the vertex coordinates within this voxel cell (i.e., in X, Y, ...) is determined. The fractional coordinates of the three axes (Z, Z, and Z) are used as interpolation weights to perform a weighted average of the distance field values ​​of the eight nodes, and the accurate sampled distance field value of the vertex position is calculated. A negative value indicates that the vertex is inside the chassis model, i.e., penetration has occurred, while zero or a positive value indicates that the vertex is on the surface or outside of the model, and no penetration has occurred. Then, a penetration judgment threshold is set, such as -0.001, and all pipeline perimeter vertices with sampled distance field values ​​less than the threshold are filtered out. These filtered vertices and their coordinate sets together constitute the target pipeline perimeter vertices.

[0029] The steps for obtaining the pipeline penetration-resistant displacement vector are as follows: Read the three-dimensional spatial coordinates of the outer vertex of the target pipeline in the directed distance field data on the chassis surface. Select positive offset sampling positions and negative offset sampling positions along the three coordinate axes of the three-dimensional spatial coordinates respectively. Extract the distance field values ​​corresponding to the positive offset sampling positions and the distance field values ​​corresponding to the negative offset sampling positions respectively. Calculate the difference between the positive distance field values ​​and the negative distance field values ​​in the same coordinate axis direction. Combine the difference values ​​in the three coordinate axis directions according to the coordinate axis order to obtain the spatial gradient vector. The sampled distance field values ​​corresponding to the outer vertices of the target pipeline are called, the absolute values ​​of the sampled distance field values ​​are extracted, and the absolute values ​​of the sampled distance field values ​​are numerically superimposed with the preset shift constant to obtain the displacement compensation value. The magnitude ratio of each directional component in the spatial gradient vector is calculated, and the spatial gradient vector is normalized according to the magnitude ratio of each directional component. The normalized spatial gradient vector is multiplied by the displacement compensation value to generate the pipeline anti-penetration displacement vector.

[0030] Specifically, the three-dimensional spatial coordinates of the outer vertices of each previously selected target pipeline in the directed distance field data on the chassis surface are read. To calculate the spatial gradient vector at that point and determine the fastest direction to push the vertex away from the chassis surface, the central difference method is used to approximate the gradient of the directed distance field. A small offset is first set for the difference calculation. This value must be set smaller than one voxel size of the directed range field data. For example, if the voxel size is 1 cm, then... It can be set to 0.1 mm, for a coordinate system. For the outermost vertex of the target pipeline, positive and negative offset samples are performed along the three coordinate axes of its current position, respectively, to obtain its position in... , , , , and The range field values ​​at these six sampling locations were also obtained from the directed range field data using trilinear interpolation. Then, the rate of change of the range field value along each coordinate axis was calculated. The gradient component in the X direction was obtained through... The gradient components in the Y and Z directions are calculated using a similar method. Then, the gradient components in these three directions are combined into a three-dimensional vector according to the coordinate axes. This vector is the spatial gradient vector at that vertex.

[0031] Call the distance field value obtained by sampling the corresponding outer vertex of each target pipeline Since this vertex is a penetrating vertex, its distance value For negative numbers, their absolute value This represents the depth at which the vertices are embedded in the chassis model. To push the vertices off the model surface and add a safety gap, a push-off constant needs to be set. This constant serves as a fixed offset, and its value is set according to the required accuracy of the training, for example, 1 millimeter, to ensure sufficient clearance between the pipeline and the chassis. The absolute value of the penetration depth is added to this displacement constant to obtain the total displacement compensation value. This value determines the total distance the vertex needs to move, and then the spatial gradient vector calculated in the previous step is called. This vector points in the direction of the fastest increase in the distance field value, that is, the direction perpendicular to the model surface and outwards. In order to use only its directional information, this vector needs to be normalized. Specifically, this is calculated by dividing each of its components by the magnitude of the vector. This yields the normalized direction vector. Then, the displacement compensation value representing the magnitude of the displacement is... With the normalized gradient vector representing the displacement direction Multiplying these together yields the final pipeline penetration-resistant displacement vector. This vector defines the displacement required to move the penetrating vertex out of the model and reach the preset safety gap.

[0032] The steps for obtaining a deformation-adaptive 3D pipeline mesh are as follows: Match the coordinates of the outer vertices of the target pipeline with the corresponding pipeline anti-penetration displacement vectors one by one according to the same vertex index. Read the horizontal, vertical, and longitudinal coordinate values ​​in the coordinates of the outer vertices of the target pipeline, and superimpose the displacement components in the corresponding directions of the pipeline anti-penetration displacement vectors. Write back the vertex indexes of the superimposed horizontal, vertical, and longitudinal coordinate values, keeping the vertex indexes and topology connection numbers unchanged, to obtain the penetration compensation updated vertices. Based on the vertex connection relationships stored in the initial tubular 3D geometric mesh, the topological adjacent connected vertices of the penetration compensation update vertex are extracted. The 3D coordinate values ​​of all topological adjacent connected vertices are read one by one. The average coordinate values ​​of all topological adjacent connected vertices in the horizontal, vertical and vertical directions are calculated to form the geometric center of all topological adjacent connected vertices. The vector difference from the penetration compensation update vertex to the geometric center of all topological adjacent connected vertices is calculated. The normal Laplace smoothing calculation is performed along the direction corresponding to the vector difference to generate the coordinates of the compressive deformation vertex. Read the coordinates of the outer vertices of the unpenetrated pipeline when the sampled distance field values ​​are non-negative. Replace the original vertex coordinates when the sampled distance field values ​​are negative with the vertex indices of the original vertex coordinates in the initial tubular 3D geometric mesh using the vertex coordinates of the original vertex coordinates in the initial tubular 3D geometric mesh. Retain the original vertex coordinates when the sampled distance field values ​​are non-negative using the vertex coordinates of the outer vertices of the unpenetrated pipeline. Rewrite the 3D coordinate values ​​corresponding to all vertex indices to generate a deformation adaptive 3D pipeline mesh.

[0033] Specifically, the original coordinates of all target pipeline perimeter vertices are matched with their corresponding pipeline anti-penetration displacement vectors. This matching process is based on the unique index of the vertex in the initial grid. For each matching pair, the original horizontal, vertical, and longitudinal coordinate values ​​of the target pipeline perimeter vertices are read. Simultaneously, the displacement components of the corresponding pipeline anti-penetration displacement vector on the three coordinate axes are read. Then, each component of the original coordinates is added to the corresponding component of the displacement vector to calculate the new vertex coordinates. ,in , , These new coordinates are the coordinates used to move the vertex from the penetration position to a safe position outside the model. After the calculation is completed, a vertex index write-back operation is performed, which means using the calculated new coordinates in the array or list storing the vertex data of the entire tubular 3D geometric mesh. Replace the original coordinates pointed to by the vertex index During this process, the vertex index number and the topological connection relationship of the mesh (i.e. how the vertices are connected to form triangular patches) remain completely unchanged. Only the position of the vertex that is penetrated is updated. The set of vertices after this step is the penetration compensation updated vertex.

[0034] Based on the pre-stored vertex connection relationships (i.e., mesh topology) within the initial tubular 3D geometric mesh, for each vertex whose position was updated in the previous step, the penetration compensation update vertex is updated. The process involves finding and extracting all topologically adjacent vertices directly connected to the vertex via edges, forming a neighborhood vertex set. Then, the 3D coordinates of each topologically adjacent vertex in this set are read one by one, and the average coordinates of these vertices on the horizontal, vertical, and center axes are calculated to obtain the geometric center point of this neighborhood vertex set. Then calculate the vertex update from the penetration compensation. Pointing to its neighborhood geometric center The vector difference, i.e., the Laplace vector. To make the deformation more natural and prevent unnecessary shrinkage or collapse of the pipe wall during the smoothing process, normal Laplacian smoothing calculations need to be performed. This means that the smoothing is constrained to occur only on the tangent plane perpendicular to the vertex normal. First, the vertex is calculated... Vertex normal at the location Then the Laplace vector Projecting onto the normal line yields the normal component. Subtracting the normal component from the original Laplacian vector yields the tangential smoothing vector. Then, the tangential smoothing vector is multiplied by a preset smoothing weight coefficient (e.g., 0.5) and added back to the coordinates of the vertex updated by the penetration compensation. This process can be iterated several times (e.g., 3 to 5 times) to more smoothly pass the deformation effect to neighboring vertices and generate the final coordinates of the deformed vertex.

[0035] Based on the distance field sampling results obtained from all pipeline perimeter vertices in the previous steps, the vertices are divided into two categories: one category consists of non-penetrating pipeline perimeter vertices with sampled distance field values ​​that are non-negative (greater than or equal to 0), and the other category consists of negative sampled values ​​whose coordinates have been processed by anti-penetration displacement and Laplace smoothing to generate compressive deformation vertex coordinates. Next, the vertex data for the final deformation-adaptive 3D pipeline mesh is constructed. Specifically, all original vertex indices of the initial tubular 3D geometric mesh are traversed, starting from index 0 until the last vertex. For each vertex index, the category is determined. If the vertex was initially marked as a penetrating vertex (with a negative sampled value), then in the new vertex... In the data list, the coordinates of the index position are set to the corresponding smoothed coordinates of the compressive deformation vertex. If the vertex is marked as a non-penetrating vertex (the sampled value is non-negative), its position does not need to be changed. Its original vertex coordinates in the initial tubular 3D geometric mesh are directly written to the corresponding index position in the new vertex data list. In this way, the conditional replacement or retention of the coordinates of all vertices is completed, and finally a brand new and complete vertex coordinate list is obtained. The vertex positions in this list have been adaptively fitted to the chassis surface. Then, this new vertex coordinate list is combined with the topological connection relationship of the initial tubular 3D geometric mesh (i.e., the vertex index definition of the triangular facet) to generate a deformation adaptive 3D pipeline mesh.

[0036] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A vehicle chassis maintenance training method based on mixed reality, characterized in that, Includes the following steps: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, oriented distance field data of the chassis surface is constructed. Extract the three-dimensional spatial coordinates of multiple discrete pipeline interaction points input by the user in the mixed reality space from the directed distance field data on the chassis surface and interpolate them to obtain multiple spatial interpolation nodes. Generate a three-dimensional spatial pipeline curve based on the multiple spatial interpolation nodes. A local pipeline frame is generated based on the tangent vector, principal normal vector, and secondary normal vector corresponding to each spatial interpolation node on the three-dimensional pipeline curve. The coordinates of the extruded vertices are obtained by projecting the preset two-dimensional circular contour vertices onto the local frame coordinate system of the pipeline. The vertex connection relationship is constructed based on the extruded vertices coordinates to generate the initial tubular three-dimensional geometric mesh. The coordinates of all outer vertices of the pipelines within the initial tubular three-dimensional geometric mesh are input into the directed distance field data of the chassis surface for sampling. The outer vertices of the target pipelines with negative distance field values ​​are extracted, and the corresponding coordinates of the outer vertices of the target pipelines are obtained. The displacement compensation value is calculated based on the spatial gradient vector corresponding to the outer vertex of the target pipeline, the sampled distance field value, and the preset displacement constant, and a pipeline anti-penetration displacement vector is generated. The coordinates of the outer vertex of the target pipeline are combined with the corresponding pipeline anti-penetration displacement vector to obtain the penetration compensation updated vertex; the topological adjacent connected vertices of the penetration compensation updated vertex are extracted and smoothed to generate the coordinates of the compressive deformation vertex; The coordinates of the compressed deformation vertices and the coordinates of the outer vertices of the unpenetrated pipeline with non-negative distance field values ​​are used to replace the original vertex coordinates in the initial tubular three-dimensional geometric mesh to generate a deformation-adaptive three-dimensional pipeline mesh.

2. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for acquiring the oriented distance field data of the chassis surface are as follows: Based on the 3D mesh model of the vehicle chassis in the mixed reality space, the vertex coordinates of the boundary triangular facets of the 3D mesh model of the vehicle chassis are extracted. The space of the 3D mesh model of the vehicle chassis is divided into voxel nodes. The center coordinates of each voxel node are read one by one. The spatial distance from the center coordinates of the voxel node to the projection point of each boundary triangular facet is calculated. The minimum spatial distance is selected as the nearest Euclidean distance. The spatial direction sign is recorded according to the inner and outer positions of the voxel node center coordinates relative to the model boundary. The spatial direction sign and the nearest Euclidean distance are combined to generate the directional distance field data of the chassis surface.

3. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the three-dimensional spatial pipeline curve are as follows: Read multiple discrete pipeline interaction points continuously input by the user in the mixed reality space, record the discrete pipeline interaction point identifiers according to the input order, retrieve the voxel neighborhood position of each discrete pipeline interaction point in the directed distance field data on the chassis surface, extract the three-dimensional spatial coordinates corresponding to the discrete pipeline interaction points, continuously interpolate the three-dimensional spatial coordinates according to the input order, fill in the spatial transition position between adjacent three-dimensional spatial coordinates, and obtain multiple spatial interpolation nodes; Establish a node arrangement sequence according to the input order of the multiple spatial interpolation nodes, read the three-dimensional spatial coordinates of adjacent spatial interpolation nodes one by one, verify whether the connection direction between adjacent spatial interpolation nodes is consistent with the pipeline input direction, establish a continuous spatial connection relationship for adjacent spatial interpolation nodes with consistent connection directions, and readjust the connection order of adjacent spatial interpolation nodes with inconsistent connection directions according to the node arrangement sequence to generate a three-dimensional spatial pipeline curve.

4. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the pipeline local marker are as follows: The three-dimensional spatial coordinates of each spatial interpolation node on the three-dimensional pipeline curve are read one by one. The difference between the three-dimensional spatial coordinates of the previous spatial interpolation node and the next spatial interpolation node is extracted according to the interpolation order. The tangent vector corresponding to each spatial interpolation node is determined. The change in tangent vector of adjacent spatial interpolation nodes is read. The principal normal vector is determined according to the change in tangent vector. The perpendicularity of tangent vector and principal normal vector is determined. The binormal vector is extracted according to the perpendicularity determination result. The tangent vector, principal normal vector and binormal vector are adjusted to mutually perpendicular unit directions to generate a local pipeline frame.

5. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the initial tubular three-dimensional geometric mesh are as follows: Based on the pipeline local frame, spatial interpolation nodes are located one by one along the three-dimensional spatial pipeline curve. The coordinate system of the pipeline local frame corresponding to each spatial interpolation node is read. The planar coordinates of the preset two-dimensional circular contour vertices are called. According to the arrangement order of the two-dimensional circular contour vertices in the plane, each two-dimensional circular contour vertex is mapped to the pipeline local frame coordinate system of the corresponding spatial interpolation node. The position of the mapped spatial vertex is recorded to obtain the extrusion vertex coordinates. Divide adjacent sections according to the interpolation order of the three-dimensional spatial pipeline curve, read the coordinates of the extrusion vertices with the same number in each adjacent section, establish the longitudinal vertex connection relationship along the pipeline direction, read the coordinates of the extrusion vertices with adjacent numbers in the same section, establish the circumferential vertex connection relationship along the circular contour direction, determine the four-point closed facets according to the longitudinal vertex connection relationship and the circumferential vertex connection relationship, merge all four-point closed facets, and generate the initial tubular three-dimensional geometric mesh.

6. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the outer vertices of the target pipeline are as follows: The pipeline perimeter vertex indices inside the initial tubular 3D geometric mesh are read one by one. The corresponding pipeline perimeter vertex coordinates are extracted according to the pipeline perimeter vertex indices. The pipeline perimeter vertex coordinates are located within the range of adjacent voxel nodes of the directed distance field data on the chassis surface. The distance field values ​​corresponding to the adjacent voxel nodes are read. The sampled distance field values ​​are calculated according to the spatial proportion of the pipeline perimeter vertex coordinates within the range of the adjacent voxel nodes. The pipeline perimeter vertex coordinates with negative sampled distance field values ​​are filtered to obtain the target pipeline perimeter vertex.

7. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the pipeline penetration-resistant displacement vector are as follows: Read the three-dimensional spatial coordinates of the outer vertex of the target pipeline in the directed distance field data on the chassis surface. Select positive offset sampling positions and negative offset sampling positions along the three coordinate axes of the three-dimensional spatial coordinates respectively. Extract the distance field values ​​corresponding to the positive offset sampling positions and the distance field values ​​corresponding to the negative offset sampling positions respectively. Calculate the difference between the positive distance field values ​​and the negative distance field values ​​in the same coordinate axis direction. Combine the difference values ​​in the three coordinate axis directions according to the coordinate axis order to obtain the spatial gradient vector. The sampled distance field values ​​corresponding to the outer vertices of the target pipeline are called, the absolute values ​​of the sampled distance field values ​​are extracted, and the absolute values ​​of the sampled distance field values ​​are superimposed with a preset shift constant to obtain the displacement compensation value. The amplitude ratio of each directional component in the spatial gradient vector is calculated, and the spatial gradient vector is normalized according to the amplitude ratio of each directional component. The normalized spatial gradient vector is multiplied by the displacement compensation value to generate the pipeline anti-penetration displacement vector.

8. The vehicle chassis maintenance training method based on mixed reality according to claim 1, characterized in that, The steps for obtaining the deformation-adaptive 3D pipeline mesh are as follows: The coordinates of the outer vertices of the target pipeline are matched one by one with the corresponding pipeline anti-penetration displacement vector according to the same vertex index. The horizontal coordinate value, vertical coordinate value, and vertical coordinate value in the coordinates of the outer vertices of the target pipeline are read, and the displacement components in the corresponding directions in the pipeline anti-penetration displacement vector are superimposed respectively. The vertex index is written back for the superimposed horizontal coordinate value, vertical coordinate value, and vertical coordinate value, keeping the vertex index and topology connection number unchanged, to obtain the penetration compensation updated vertex. Based on the vertex connection relationships stored in the initial tubular three-dimensional geometric mesh, the topological adjacent connected vertices of the penetration compensation update vertex are extracted. The three-dimensional coordinate values ​​of all topological adjacent connected vertices are read one by one. The average coordinate values ​​of all topological adjacent connected vertices in the horizontal, vertical and vertical directions are calculated respectively to form the geometric center of all topological adjacent connected vertices. The vector difference from the penetration compensation update vertex to the geometric center of all topological adjacent connected vertices is calculated. The normal Laplace smoothing calculation is performed along the direction corresponding to the vector difference to generate the coordinates of the compressive deformation vertex. Read the coordinates of the outer vertices of the unpenetrated pipeline when the sampled distance field values ​​are non-negative. According to the vertex indices of the original vertex coordinates in the initial tubular 3D geometric mesh, replace the original vertex coordinates of the sampled distance field when the sampled distance field values ​​are negative with the vertex coordinates of the compressive deformation vertex. Use the vertex coordinates of the outer vertices of the unpenetrated pipeline to retain the original vertex coordinates of the sampled distance field when the sampled distance field values ​​are non-negative. Rewrite the 3D coordinate values ​​corresponding to all vertex indices to generate a deformation adaptive 3D pipeline mesh.

Images