Vehicle positioning method, computer device, server, storage medium and vehicle

Abstract

The present application relates to the technical field of automatic driving and unmanned driving, and particularly provides a vehicle positioning method, computer equipment, a server, a storage medium and a vehicle, aiming to solve the problem of reducing the real-time performance and reliability of positioning caused by too long loading time and large memory resource occupation. The method provided by the present application comprises loading a serialized file formed by serializing a data structure of a point cloud probability distribution map, the data structure being formed by encapsulating a voxel index of a point cloud map and normal distribution parameters of point clouds in the voxel; performing deserialization processing on the serialized file to obtain the data structure; and using a positioning method based on an NDT algorithm to perform point cloud registration on a laser radar point cloud frame collected by the vehicle according to the data structure, so as to determine the pose of the vehicle. Through the above method, the data loading time can be greatly reduced, and after the loading is completed, too much memory resource will not be occupied, and finally the real-time performance and reliability of vehicle positioning are improved.

Description

Technical Field

[0001] This invention relates to the fields of autonomous driving and driverless technology, specifically to a vehicle positioning method, computer equipment, server, storage medium, and vehicle. Background Technology

[0002] In laser point cloud localization technology based on the NDT (Normal Distributions Transform) algorithm, it is typically necessary to load the point cloud map stored on the computer's hard drive into memory. The loaded point cloud map is then processed to obtain a point cloud probability distribution map. Finally, this probability distribution map is used to perform point cloud registration on each frame of the LiDAR point cloud acquired by the vehicle to determine the vehicle's pose. Since the point cloud map is usually quite large, loading it from the hard drive into memory takes a considerable amount of time and consumes significant memory resources, which is detrimental to improving the real-time performance and reliability of laser point cloud localization.

[0003] Accordingly, a new technical solution is needed in this field to solve the above problems. Summary of the Invention

[0004] To overcome the above-mentioned defects, this invention is proposed to provide a vehicle positioning method, computer equipment, server, storage medium, and vehicle that solves or at least partially solves the technical problem of how to avoid the reduction in real-time performance and reliability of positioning due to excessively long data loading time and large data memory resource consumption when using NDT algorithm-based laser point cloud positioning technology for vehicle positioning.

[0005] In a first aspect, a vehicle positioning method is provided, the method being applied to a vehicle, the method comprising:

[0006] A serialized file is formed by serializing the data structure of the point cloud probability distribution map. The point cloud probability distribution map includes an index of each voxel formed after voxelization of the point cloud map and normal distribution parameters of the point cloud located in each voxel. Each data structure is encapsulated by an index of a voxel and normal distribution parameters of the point cloud located in the voxel.

[0007] The serialized file is deserialized to obtain each of the data structures;

[0008] A localization method based on the NDT (Normal Distributions Transform) algorithm is adopted. The point cloud registration of the LiDAR point cloud frame acquired by the vehicle is performed according to the index of the voxel encapsulated in each data structure and the normal distribution parameters of the point cloud located in each voxel, so as to determine the pose of the vehicle.

[0009] In one technical solution of the above vehicle positioning method, the data structure is formed by encapsulating the index of effective voxels and the normal distribution parameters of the point cloud located within the effective voxels, wherein the effective voxels are voxels containing a number of point clouds greater than or equal to a preset number threshold.

[0010] In one technical solution of the above vehicle positioning method, the serialized file is formed by serializing the data structure of the point cloud probability distribution map using the Google Protocol Buffer data serialization protocol. The step of "deserializing the serialized file to obtain each data structure" specifically includes:

[0011] The serialized file is deserialized using the Google Protocol Buffer data serialization protocol to obtain the data structure of the point cloud probability distribution map.

[0012] In one technical solution of the above vehicle positioning method, the normal distribution parameters encapsulated in the data structure include the mean and covariance information of the point cloud coordinates located within the voxel;

[0013] The covariance information includes the covariance matrix of the point cloud coordinates, the inverse matrix of the covariance matrix, multiple eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix, and the eigenvector corresponding to each eigenvalue.

[0014] In a second aspect, a vehicle positioning method is provided, the method being applied to a server, the method comprising:

[0015] Obtain a point cloud probability distribution map of a point cloud map, wherein the point cloud probability distribution map includes an index of each voxel formed after voxelization of the point cloud map and normal distribution parameters of the point cloud located in each voxel.

[0016] The index of each voxel and the normal distribution parameters of the point cloud located within each voxel are encapsulated into a data structure of the point cloud probability distribution map.

[0017] The data structure of the point cloud probability distribution map is serialized to form a serialized file, which the vehicle can load. The serialized file is then deserialized to obtain the data structure. A localization method based on the NDT (Normal Distributions Transform) algorithm is then used to perform point cloud registration on the lidar point cloud frames collected by the vehicle, based on the index of the voxel encapsulated in each data structure and the normal distribution parameters of the point cloud within each voxel, to determine the vehicle's pose.

[0018] In one technical solution of the above vehicle positioning method, the step of "encapsulating the index of each voxel and the normal distribution parameters of the point cloud located in each voxel into a data structure of the point cloud probability distribution map" specifically includes:

[0019] Select voxels whose number of point clouds is greater than or equal to a preset threshold as effective voxels, and encapsulate the index of each effective voxel and the normal distribution parameters of the point cloud within each effective voxel into a data structure of the point cloud probability distribution map.

[0020] And / or, the method further includes serializing the data structure to form a serialized file by using the Google Protocol Buffer data serialization protocol;

[0021] And / or, the normal distribution parameters encapsulated in the data structure include the mean and covariance information of the point cloud coordinates located within the voxel. The covariance information includes the covariance matrix of the point cloud coordinates, the inverse matrix of the covariance matrix, multiple eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix, and the eigenvector corresponding to each eigenvalue. The step of "encapsulating the index of each voxel and the normal distribution parameters of the point cloud located within each voxel into a data structure of the point cloud probability distribution map" specifically includes:

[0022] The index of each voxel and the normal distribution parameters of the point cloud located within each voxel are encapsulated into a data structure of the point cloud probability distribution map.

[0023] And / or, prior to the step of "obtaining the point cloud probability distribution map of the point cloud map", the method further includes: deleting point clouds in the point cloud map whose point cloud height is greater than a preset upper limit value and deleting point clouds whose point cloud height is less than a preset lower limit value.

[0024] In a third aspect, a computer device is provided, the computer device including a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and run by the processor to perform the vehicle positioning method described in any of the technical solutions of the vehicle positioning method of the first aspect above.

[0025] In a fourth aspect, a server is provided, the server including a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and run by the processor to perform the vehicle positioning method described in any of the technical solutions of the vehicle positioning method of the second aspect above.

[0026] In a fifth aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the vehicle positioning method described in any one of the technical solutions of the vehicle positioning method of the first aspect above, or the program codes being adapted to be loaded and run by a processor to perform the vehicle positioning method described in any one of the technical solutions of the vehicle positioning method of the second aspect above.

[0027] In a sixth aspect, a vehicle is provided that includes the aforementioned computer equipment.

[0028] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects:

[0029] In implementing the technical solution of this invention, a serialized file is formed by serializing a data structure (each data structure is encapsulated by an index of a voxel and the normal distribution parameters of the point cloud within each voxel) of a point cloud probability distribution map loaded by the vehicle. The serialized file is then deserialized to obtain each data structure. Finally, a positioning method based on the NDT (Normal Distributions Transform) algorithm is used to perform point cloud registration on the lidar point cloud frames collected by the vehicle, based on the voxel index encapsulated in each data structure and the normal distribution parameters of the point cloud within each voxel, to determine the vehicle's pose.

[0030] Compared to point cloud maps, point cloud probability distribution maps have a much smaller data volume. Furthermore, serializing point cloud probability distribution maps can further reduce their data volume. This significantly reduces data loading time when loading the serialized point cloud probability distribution map file into the vehicle's memory, and it does not consume excessive memory resources after loading. This overcomes the problem of existing technologies that first load the point cloud map into the vehicle's memory and then process the point cloud map in the vehicle's memory to obtain the point cloud probability distribution map, which consumes a lot of data loading time and memory resources, ultimately reducing the real-time performance and reliability of vehicle positioning. Attached Figure Description

[0031] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Wherein:

[0032] Figure 1 This is a schematic diagram of the hardware environment for a vehicle positioning method according to an embodiment of the present invention;

[0033] Figure 2 This is a schematic flowchart of the main steps of a vehicle positioning method according to an embodiment of the present invention;

[0034] Figure 3 This is a schematic flowchart of the main steps of a method for obtaining a data structure according to an embodiment of the present invention;

[0035] Figure 4 This is a schematic flowchart of the main steps of a vehicle positioning method according to another embodiment of the present invention;

[0036] Figure 5 This is a schematic flowchart of the main steps of a vehicle positioning method according to another embodiment of the present invention. Detailed Implementation

[0037] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0038] In the description of this invention, "processor" can include hardware, software, or a combination of both. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B.

[0039] The vehicle positioning method embodiments provided by this invention can be applied to, for example... Figure 1 The diagram shows a hardware environment consisting of vehicles and servers. For example... Figure 1 As shown, the server connects to the vehicle via a network and can be used to provide services (such as application services) to the vehicle or clients installed on the vehicle. A database can be set up on the server or independently to provide data storage services to the server. Cloud computing and / or edge computing services can be configured on the server or independently to provide data processing services to the server.

[0040] The following describes the vehicle positioning method embodiments applied to the server and vehicle in the above hardware environment.

[0041] I. Implementation Examples of Vehicle Positioning Methods Applied to Servers

[0042] See appendix Figure 2 , Figure 2 This is a schematic flowchart illustrating the main steps of a vehicle positioning method applied to a server according to an embodiment of the present invention. Figure 2 As shown, the vehicle positioning method in this embodiment of the invention mainly includes the following steps S101 to S103.

[0043] Step S101: Obtain the point cloud probability distribution map of the point cloud map, wherein the point cloud probability distribution map may include the index of each voxel formed after voxelization of the point cloud map and the normal distribution parameters of the point cloud located in each voxel.

[0044] A point cloud map refers to a map built using LiDAR point clouds. LiDAR point clouds are the point cloud data of each environmental reflection point in the environment obtained by scanning the vehicle's surroundings using a LiDAR scanner. This invention does not specifically limit the method for building a point cloud map; any method that can obtain a point cloud map and derive a point cloud probability distribution map from it is acceptable.

[0045] The coordinate system of a point cloud map is typically a three-dimensional coordinate system formed by the X, Y, and Z axes. For example, the map coordinate system can be a three-dimensional coordinate system formed by longitude, latitude, and height (longitude, latitude, and height correspond to the X, Y, and Z axes respectively), or it can be a northeast-sky coordinate system (a three-dimensional rectangular coordinate system where the X-axis points east, the Y-axis points north, and the Z-axis points to the zenith). When a vehicle travels on a road, its relative height to the road usually does not change abruptly. Therefore, before voxelizing the point cloud map, noisy point clouds in the Z-axis direction can be removed first, and then voxelization can be performed, thereby improving the accuracy and efficiency of vehicle positioning. Specifically, in this embodiment of the invention, point clouds with heights greater than a preset upper limit and point clouds with heights less than a preset lower limit can be deleted; that is, point clouds with excessively high or low heights can be deleted.

[0046] When performing voxelization on point cloud maps, conventional voxelization methods in the field of point cloud data processing technology can be used. For example, firstly, the distribution boundaries of the point cloud on the X, Y, and Z axes can be determined based on the distribution of the point cloud on the X, Y, and Z axes. The minimum coordinate x of the point cloud on the X-axis is then defined. min and maximum coordinate x max This forms the distribution boundary of the point cloud on the X-axis, and the minimum coordinate y of the point cloud on the Y-axis. min and the maximum coordinate y max This forms the distribution boundary of the point cloud on the Y-axis, and the minimum coordinate z of the point cloud on the Z-axis. min and the maximum coordinate z max This establishes the distribution boundary of the point cloud along the Z-axis. Then, based on this distribution boundary, a large cube is generated that can contain all point clouds (all point clouds after noise removal). The length, width, and height of this large cube are (x, y, y) and (x, y, y). max -x min ), (y max -y min ) and (z max -z min Finally, the large cube is divided into multiple point cloud map voxels with preset side lengths. size The point cloud map consists of small cubes, which are called voxels; each small cube represents a voxel. Furthermore, each voxel has its own corresponding index, which contains at least the voxel's coordinates. The corresponding voxel can be retrieved by querying the index value. It should be noted that those skilled in the art can flexibly set the preset point cloud map resolution (voxel) according to actual needs. size The specific values ​​are not specifically limited in the embodiments of the present invention.

[0047] The normal distribution parameters of a point cloud refer to the parameters that determine the normal distribution function of the point cloud. For example, the normal distribution parameters of a point cloud may include the mean and variance of the point cloud coordinates located within a voxel. In some implementations, the normal distribution parameters may include the mean and covariance matrix of the point cloud coordinates located within a voxel. Wherein, the mean p of the point cloud coordinates... k The formulas for calculating the covariance matrix ∑k of the point cloud coordinates are shown in equations (1) and (2) below.

[0048]

[0049]

[0050] The meanings of the parameters in formulas (1) and (2) are as follows:

[0051] p k M represents the mean of the point cloud coordinates located in the k-th voxel. k x represents the total number of point clouds located within the k-th voxel. ki Let represent the coordinates of the i-th point cloud within the k-th voxel, ∑k represent the covariance matrix of the point cloud coordinates within the k-th voxel, and t represent the transpose of the matrix.

[0052] Step S102: Encapsulate the index of each voxel and the normal distribution parameters of the point cloud within each voxel into a data structure ProbMap for a point cloud probability distribution map.

[0053] In this embodiment of the invention, conventional structure encapsulation methods in the field of computer technology can be used to encapsulate the voxel index and normal distribution parameters together to form a data structure ProbMap. This embodiment of the invention will not elaborate on the above-mentioned structure encapsulation method.

[0054] When using a localization method based on the NDT (Normal Distributions Transform) algorithm to register point cloud frames acquired by LiDAR for a vehicle based on the aforementioned normal distribution parameters to determine the vehicle's pose, in order to reduce the computational load of point cloud registration based on the aforementioned normal distribution parameters, after obtaining the covariance matrix of the point cloud coordinates within the voxels, some parameters required for point cloud registration can be calculated in advance and encapsulated into a data structure ProbMap. This way, after obtaining the ProbMap data structure, these parameters can be directly retrieved from it without further calculation. Specifically, the aforementioned parameters may include the inverse of the covariance matrix, multiple eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix, and the eigenvectors corresponding to each eigenvalue. After obtaining these parameters, the index of each voxel, the mean of the point cloud coordinates within each voxel, the covariance matrix, the inverse of the covariance matrix, the eigenvalues, and the eigenvectors can be encapsulated into a point cloud probability distribution map data structure ProbMap. In some implementations, the eigenvalues ​​and their corresponding eigenvectors in the normal distribution parameters of each voxel can be arranged in descending order of eigenvalues, and then encapsulated to form a data structure ProbMap.

[0055] It should be noted that, in the embodiments of the present invention, the conventional matrix eigenvalue decomposition method in the field of mathematical technology can be used to decompose the covariance matrix into eigenvalues ​​to obtain the eigenvalues ​​of the covariance matrix and the eigenvectors corresponding to each eigenvalue. The embodiments of the present invention will not elaborate on the eigenvalue decomposition method.

[0056] Step S103: Serialize the data structure ProbMap of the point cloud probability distribution map to form a serialized file so that the vehicle can load it and deserialize the serialized file to obtain the data structure ProbMap. Then, using the NDT algorithm-based localization method, the vehicle's pose is determined by registering the LiDAR point cloud frames acquired by the vehicle with the index of the voxel encapsulated in each data structure ProbMap and the normal distribution parameters of the point cloud in each voxel.

[0057] In this embodiment of the invention, the Google Protocol Buffer data serialization protocol can be used to serialize the ProbMap data structure to form a serialized file. Alternatively, those skilled in the art can use other conventional data serialization methods in the field of computer technology to serialize the ProbMap data structure to form a serialized file, as long as serialization can be achieved, facilitating the transmission of the ProbMap data structure and reducing its data size.

[0058] The following is a brief description of the method for determining vehicle pose using a localization method based on the NDT algorithm and the ProbMap data structure of a point cloud probability distribution map. Specifically, in this embodiment of the invention, the vehicle pose can be determined through the following steps 11 to 14.

[0059] Step 11: Acquire the LiDAR point cloud frame collected by the vehicle. Based on the vehicle's initial pose, project the point cloud in the LiDAR point cloud frame to the corresponding voxel in the data structure ProbMap. That is, perform coordinate transformation on the point cloud in the LiDAR point cloud frame, transforming the coordinates of each point cloud in the LiDAR point cloud frame to the coordinates in the voxel.

[0060] In this embodiment of the invention, the relative pose between two adjacent lidar point cloud frames obtained from the laser odometry on the vehicle can be selected as the initial pose of the vehicle.

[0061] Step 12: Calculate the distribution probability of each point cloud projected into the voxel based on the normal distribution parameters encapsulated in the ProbMap data structure. The formula for calculating the distribution probability is shown in Equation (3) below.

[0062]

[0063] The meanings of each parameter in formula (3) are as follows:

[0064] x i Let p(x) represent the coordinates of the i-th point cloud. i Let p represent the probability distribution of the i-th point cloud. k Let represent the mean of the point cloud coordinates within the k-th voxel, and ∑k represent the covariance matrix of the point cloud coordinates within the k-th voxel. Let x' be the inverse of the covariance matrix ∑k. i P represents the coordinates of the i-th point cloud projected onto the corresponding projection point in the voxel. vehicle The vehicle pose to be determined is represented by T, which is the transformation matrix from the coordinate system of the lidar point cloud frame to the coordinate system of the voxel.

[0065] Step 13: Add up the distribution probabilities of all point clouds in the lidar point cloud frame to obtain the objective function shown in the following formula (4).

[0066]

[0067] In formula (4), n represents the total number of voxels, and M k This represents the total number of point clouds located within the k-th voxel.

[0068] Step 14: Use Newton's method to find the vehicle pose P that maximizes the objective function. vehicle As the final vehicle position.

[0069] Based on the method described in steps S101 to S103 above, a serialized file of the point cloud probability distribution map that is conducive to the rapid loading and utilization of vehicles can be obtained. This greatly reduces the data loading time when the vehicle loads this serialized file in memory, and does not occupy too much memory resources after loading, thereby improving the real-time performance and reliability of vehicle positioning.

[0070] The following provides a further explanation of step S102.

[0071] After voxelization of the point cloud map, the number of points within each voxel varies. Voxels with fewer points have less impact on vehicle pose, so these voxels can be treated as invalid voxels, and the remaining voxels as valid voxels. Then, only the data structure ProbMap of the valid voxels is obtained, and these ProbMap data structures are serialized. This improves the efficiency of serialization and reduces the size of the serialized file. For details, see Appendix. Figure 3 In some embodiments of step S102 above, the data structure ProbMap can be obtained through the following steps S1021 to S1022.

[0072] Step S1021: Select voxels whose number of point clouds is greater than or equal to a preset number threshold as valid voxels.

[0073] Step S1022: Encapsulate the index of each valid voxel and the normal distribution parameters of the point cloud within each valid voxel into a data structure ProbMap for a point cloud probability distribution map.

[0074] Those skilled in the art can flexibly set the specific value of the preset quantity threshold according to actual needs, and the embodiments of the present invention do not impose specific limitations on this. For example, in some embodiments, the preset quantity threshold can be 6, where voxels with a point cloud quantity greater than or equal to 6 are considered valid voxels, while voxels with a point cloud quantity less than 6 are considered invalid voxels.

[0075] The above describes an embodiment of a vehicle positioning method applied to a server. The following describes an embodiment of a vehicle positioning method applied to a vehicle.

[0076] II. Implementation Examples of Vehicle Positioning Methods Applied to Vehicles

[0077] See appendix Figure 4 , Figure 4This is a schematic flowchart illustrating the main steps of a vehicle positioning method applied to a vehicle according to an embodiment of the present invention. Figure 4 As shown, the vehicle positioning method in this embodiment of the invention mainly includes the following steps S201 to S203.

[0078] Step S201: Load the serialized file formed by serializing the ProbMap data structure of the point cloud probability distribution map.

[0079] The point cloud probability distribution map and the data structure ProbMap have the same meanings as those in the aforementioned vehicle localization method embodiment applied to the server, and will not be repeated here. The serialized file refers to the serialized file obtained through the aforementioned vehicle localization method embodiment applied to the server. After obtaining the serialized file through the aforementioned method embodiment, the serialized file can be stored on the vehicle, for example, on the vehicle's hard drive.

[0080] Step S202: Deserialize the serialized file to obtain each data structure ProbMap.

[0081] In this embodiment of the invention, the method for deserializing the serialized file can be determined based on the serialization method used, so that the serialized file can be deserialized to obtain the data structure ProbMap. In some implementations, the serialized file can be formed by serializing the point cloud probability distribution map data structure ProbMap using the Google Protocol Buffer data serialization protocol. Therefore, the Google Protocol Buffer data serialization protocol can be used to deserialize the serialized file to obtain the point cloud probability distribution map data structure ProbMap.

[0082] Step S203: Using a localization method based on the NDT algorithm, point cloud registration is performed on the lidar point cloud frames acquired by the vehicle based on the index of the voxel encapsulated in each data structure ProbMap and the normal distribution parameters of the point cloud within each voxel, so as to determine the vehicle's pose.

[0083] In this step, the method for determining the vehicle's pose is the same as that described in steps 11 to 14 of the aforementioned method embodiment, and will not be repeated here.

[0084] Based on the method described in steps S201 to S203 above, the time for vehicle data loading can be greatly reduced and the data will not occupy too much memory resources after loading, thereby improving the real-time performance and reliability of vehicle positioning.

[0085] The above is a description of an embodiment of a vehicle positioning method applied to vehicles.

[0086] The following is in conjunction with the appendix Figure 5 For applications such as Figure 1 The overall steps of the vehicle localization method, which uses a hardware environment consisting of a server and a vehicle, are briefly explained.

[0087] See appendix Figure 5 , Figure 5 According to embodiments of the present invention, it is applied to, for example Figure 1 The diagram shows the main steps of a vehicle localization method within a hardware environment consisting of a server and a vehicle. Figure 5 As shown, the vehicle positioning method mainly includes the following steps S301 to S308.

[0088] Step S301: The server deletes point clouds in the point cloud map whose point cloud height is greater than the preset upper limit value and deletes point clouds whose point cloud height is less than the preset lower limit value.

[0089] Step S302: The server obtains the point cloud probability distribution map of the point cloud map, wherein the point cloud probability distribution map includes the index of each voxel formed after voxelization of the point cloud map and the normal distribution parameters of the point cloud located in each voxel.

[0090] Step S303: The server selects voxels whose number of point clouds is greater than or equal to a preset threshold as valid voxels.

[0091] Step S304: The server arranges the eigenvalues ​​and their corresponding eigenvectors in the normal distribution parameters of each voxel in descending order of eigenvalues. Then, it encapsulates the index of each voxel and the normal distribution parameters of the point cloud within each voxel into a data structure ProbMap for a point cloud probability distribution map.

[0092] Step S305: The server uses the Google Protocol Buffer data serialization protocol to serialize the ProbMap data structure into a serialized file. After obtaining the serialized file from the server, this serialized file is stored on the vehicle.

[0093] Step S306: The vehicle loads the serialized file.

[0094] Step S307: The vehicle uses the Google Protocol Buffer data serialization protocol to deserialize the serialized file to obtain the data structure ProbMap of the point cloud probability distribution map.

[0095] Step S308: The vehicle uses a localization algorithm based on the NDT algorithm. According to the index of the voxel encapsulated in each data structure ProbMap and the normal distribution parameters of the point cloud in each voxel, the point cloud frames collected by the LiDAR are registered to determine the vehicle's pose.

[0096] The implementation methods of steps S301 to S308 are the same as the relevant methods in the aforementioned vehicle positioning method embodiment applied to servers and vehicles, and will not be repeated here.

[0097] The above is a brief explanation of the overall steps and procedures of the vehicle positioning method.

[0098] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0099] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0100] Furthermore, the present invention also provides a computer device.

[0101] In an embodiment of a computer device according to the present invention, the computer device includes a processor and a storage device. The storage device may be configured to store a program for executing the vehicle positioning method of the above-described method embodiments, and the processor may be configured to execute the program in the storage device. This program includes, but is not limited to, a program for executing the vehicle positioning method of the above-described vehicle positioning method embodiments applied to a vehicle. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. This computer device may be a control device device comprising various electronic devices.

[0102] Furthermore, the present invention also provides a server.

[0103] In one embodiment of a server according to the present invention, the server includes a processor and a storage device. The storage device may be configured to store a program for executing the vehicle positioning method of the above-described method embodiments, and the processor may be configured to execute the program in the storage device. This program includes, but is not limited to, a program for executing the vehicle positioning method of the above-described vehicle positioning method embodiments applied to the server. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The server may be a control device comprising various electronic devices.

[0104] Furthermore, the present invention also provides a computer-readable storage medium.

[0105] In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium may be configured to store a program for performing the vehicle positioning method of the above-described method embodiments. This program may be loaded and run by a processor to implement the above-described vehicle positioning method applied to a vehicle, or the program may be loaded and run by a processor to implement the above-described vehicle positioning method applied to a server. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium may be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0106] Furthermore, the present invention also provides a vehicle. In one embodiment of a vehicle according to the present invention, the vehicle may include the computer equipment described in the above-described computer equipment embodiments. In this embodiment, the vehicle may be an autonomous vehicle, an unmanned vehicle, or the like. Moreover, according to the type of power source, the vehicle in this embodiment may be a gasoline vehicle, an electric vehicle, a hybrid vehicle using a mixture of electric and gasoline power, or a vehicle using other new energy sources, etc.

[0107] The technical solution of the present invention has been described above with reference to one embodiment shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions resulting from such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A vehicle positioning method, characterized in that, The method is applied to a vehicle, and the method includes: A serialized file is formed by serializing the data structure of the point cloud probability distribution map. The point cloud probability distribution map includes an index of each voxel formed after voxelization of the point cloud map and normal distribution parameters of the point cloud located in each voxel. Each data structure is encapsulated by an index of a voxel and normal distribution parameters of the point cloud located in the voxel. The serialized file is deserialized to obtain each of the data structures; A localization method based on the NDT algorithm is adopted. The point cloud registration of the LiDAR point cloud frame acquired by the vehicle is performed according to the index of the voxel encapsulated in each data structure and the normal distribution parameters of the point cloud located in each voxel, so as to determine the pose of the vehicle.

2. The vehicle positioning method according to claim 1, characterized in that, The data structure is formed by encapsulating the index of effective voxels and the normal distribution parameters of the point cloud located within the effective voxels, wherein the effective voxels are voxels containing a number of point clouds greater than or equal to a preset number threshold.

3. The vehicle positioning method according to claim 1, characterized in that, The serialized file is formed by serializing the data structure of the point cloud probability distribution map using the Google Protocol Buffer data serialization protocol. The step of "deserializing the serialized file to obtain each data structure" specifically includes: The serialized file is deserialized using the Google Protocol Buffer data serialization protocol to obtain the data structure of the point cloud probability distribution map.

4. The vehicle positioning method according to claim 1, characterized in that, The normal distribution parameters encapsulated in the data structure include the mean and covariance information of the point cloud coordinates located within the voxel; The covariance information includes the covariance matrix of the point cloud coordinates, the inverse matrix of the covariance matrix, multiple eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix, and the eigenvector corresponding to each eigenvalue.

5. A vehicle positioning method, characterized in that, The method is applied to a server, and the method includes: Obtain a point cloud probability distribution map of a point cloud map, wherein the point cloud probability distribution map includes an index of each voxel formed after voxelization of the point cloud map and normal distribution parameters of the point cloud located in each voxel. The index of each voxel and the normal distribution parameters of the point cloud located within each voxel are encapsulated into a data structure of the point cloud probability distribution map. The data structure of the point cloud probability distribution map is serialized to form a serialized file, which the vehicle can load. The serialized file is then deserialized to obtain the data structure. A localization method based on the NDT algorithm is then used to perform point cloud registration on the lidar point cloud frames collected by the vehicle, based on the index of the voxel encapsulated in each data structure and the normal distribution parameters of the point cloud within each voxel, to determine the vehicle's pose.

6. The vehicle positioning method according to claim 5, characterized in that, The step of "encapsulating the index of each voxel and the normal distribution parameters of the point cloud within each voxel into a data structure for the point cloud probability distribution map" specifically includes: Select voxels whose number of point clouds is greater than or equal to a preset threshold as valid voxels, and encapsulate the index of each valid voxel and the normal distribution parameters of the point cloud within each valid voxel into a data structure of the point cloud probability distribution map. And / or, The method further includes serializing the data structure to form a serialized file by using the Google Protocol Buffer data serialization protocol; And / or, The normal distribution parameters encapsulated in the data structure include the mean and covariance information of the point cloud coordinates located within the voxel. The covariance information includes the covariance matrix of the point cloud coordinates, the inverse matrix of the covariance matrix, multiple eigenvalues ​​obtained by eigenvalue decomposition of the covariance matrix, and the eigenvector corresponding to each eigenvalue. The step of "encapsulating the index of each voxel and the normal distribution parameters of the point cloud located within each voxel into a data structure of the point cloud probability distribution map" specifically includes: The index of each voxel and the normal distribution parameters of the point cloud located within each voxel are encapsulated into a data structure of the point cloud probability distribution map. And / or, Before the step of "obtaining the point cloud probability distribution map of the point cloud map", the method further includes: deleting point clouds in the point cloud map whose point cloud height is greater than a preset upper limit value and deleting point clouds whose point cloud height is less than a preset lower limit value.

7. A computer device comprising a processor and a storage device, said storage device being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the vehicle positioning method according to any one of claims 1 to 4.

8. A server comprising a processor and a storage device, said storage device being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the vehicle positioning method according to any one of claims 5 to 6.

9. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the vehicle positioning method according to any one of claims 1 to 4; Alternatively, the program code may be loaded and run by a processor to perform the vehicle positioning method of any one of claims 5 to 6.

10. A vehicle, characterized in that, The vehicle includes the computer equipment as described in claim 7.

Images