Positioning method and device of autonomous vehicle, readable storage medium and vehicle

Abstract

The present application relates to the technical field of automatic driving, and particularly provides a positioning method and device of an automatic driving vehicle, a readable storage medium and a vehicle, aiming at solving the problem of how to perform accurate global positioning of the vehicle based on a laser radar. To this end, the present application acquires a light-weight point cloud map according to an initial positioning result obtained by a fusion positioning module of the vehicle, and acquires a global positioning result of the vehicle according to the light-weight point cloud map and point clouds collected by the laser radar. The present application can realize global positioning of the vehicle by matching the point clouds collected by the laser radar of the vehicle with the light-weight point cloud map, and can also give an accurate global positioning result of the vehicle in the case of poor RTK signal. When matching the light-weight point cloud map with the point clouds collected by the laser radar, the complexity of calculation in the matching process can be reduced, so that the global positioning result of the vehicle can have higher accuracy, and centimeter-level positioning can be realized.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and specifically provides a method, apparatus, readable storage medium, and vehicle for locating autonomous vehicles. Background Technology

[0002] Autonomous vehicles typically use a fusion positioning module installed on the vehicle to output converged positioning information, thereby assisting in the realization of autonomous driving. Before the fusion positioning module outputs converged positioning information, it is necessary to acquire vehicle information, such as vehicle position velocity and acceleration, and also to input the vehicle's initial relative pose and global pose for fusion filtering to obtain converged positioning information.

[0003] In outdoor environments, global pose is typically provided by RTK (Real-time kinematic). However, in situations with poor RTK signal (e.g., when the vehicle is in a tunnel or underground parking garage), global observation information is lost, making it impossible to provide an accurate global pose for the vehicle. Furthermore, due to limitations in vehicle hardware, current technologies do not utilize LiDAR for localization in autonomous vehicles.

[0004] Accordingly, there is a need in the field for a new positioning scheme for autonomous vehicles to address the aforementioned problems. Summary of the Invention

[0005] To overcome the above-mentioned shortcomings, the present invention is proposed to provide a solution, or at least a partial solution, to the problem of how to perform accurate global positioning of vehicles based on lidar.

[0006] In a first aspect, the present invention provides a method for locating an autonomous vehicle, the method comprising:

[0007] Based on the fusion positioning module installed on the vehicle, the initial positioning result of the vehicle is obtained;

[0008] Based on the initial positioning result, a lightweight point cloud map within a preset distance range of the initial positioning result is obtained;

[0009] The lightweight point cloud map is matched with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle.

[0010] In one technical solution of the above-mentioned autonomous vehicle positioning method, the step of "matching the lightweight point cloud map with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle" includes:

[0011] Based on the parameters of the lightweight point cloud map, the point cloud collected by the lidar installed on the vehicle is processed to obtain NDT point cloud data.

[0012] The NDT-D2D algorithm is applied to match the lightweight point cloud map with the NDT point cloud data to obtain the global positioning result of the vehicle.

[0013] In one technical solution of the above-mentioned autonomous vehicle positioning method, the step of "applying the NDT-D2D algorithm to match the lightweight point cloud map with the NDT point cloud data to obtain the global positioning result of the vehicle" includes:

[0014] Based on the initial positioning results, obtain the transformation matrix for converting the NDT point cloud data to the map coordinate system of the lightweight point cloud map;

[0015] The NDT-D2D algorithm is applied and iterated multiple times to optimize the transformation matrix, so that the NDT point cloud data and the lightweight point cloud map are matched to obtain the final transformation matrix.

[0016] Based on the final transformation matrix, the global positioning result of the vehicle is obtained.

[0017] In one technical solution of the above-mentioned positioning method for autonomous vehicles, the lightweight point cloud map includes multiple map voxels, and the NDT point cloud data includes multiple point cloud data voxels. The step of "applying the NDT-D2D algorithm, performing multiple iterations, and optimizing the transformation matrix to match the NDT point cloud data with the lightweight point cloud map to obtain the final transformation matrix" includes:

[0018] For each iteration, the NDT-D2D algorithm is applied to transform the NDT point cloud data into the map coordinate system of the lightweight point cloud map according to the transformation matrix, and a matching pair consisting of map voxels and point cloud data voxels is obtained.

[0019] Based on the distribution distance of the matching pairs, construct the residual function for the current iteration;

[0020] Calculate the gradient of the current iteration based on the residual function;

[0021] The gradient descent method is applied to update the transformation matrix based on the gradient, and then the next iteration is performed.

[0022] When the iteration meets the preset cutoff condition, the iteration ends to obtain the final transformation matrix.

[0023] In one technical solution of the aforementioned autonomous vehicle localization method, the step of "obtaining a matching pair consisting of a one-to-one map voxel and a point cloud data voxel" includes:

[0024] For each point cloud data voxel transformed to the map coordinate system, the nearest neighbor algorithm is applied to obtain multiple matching map voxels;

[0025] Based on the normal distribution parameters of the map voxels and the point cloud data voxels, a similarity analysis is performed on multiple matching map voxels to obtain map voxels that correspond one-to-one with the point cloud data voxels, thus forming matching pairs.

[0026] In one technical solution of the above-mentioned autonomous vehicle localization method, the step of "constructing the residual function for the current iteration based on the distribution distance of the matching pairs" includes:

[0027] Construct the residual function for the current iteration based on the distribution distances of all matching pairs;

[0028] Wherein, the residual function is positively correlated with the distribution distance; the distribution distance includes the mean distance of the normal distribution and the covariance difference of the normal distribution; and / or,

[0029] The cutoff condition for the iteration is that the gradient value of the iteration is less than a preset gradient threshold.

[0030] In one technical solution of the above-mentioned autonomous vehicle positioning method, the lightweight point cloud map is a point cloud map represented by a normal distribution. The step of "processing the point cloud collected by the lidar installed on the vehicle according to the parameters of the lightweight point cloud map to obtain NDT point cloud data" includes:

[0031] The point cloud collected by the lidar is preprocessed to obtain preprocessed point cloud data;

[0032] Based on the parameters of the lightweight point cloud map, the preprocessed point cloud data is converted into NDT point cloud data with the same normal distribution expression as the lightweight point cloud map.

[0033] In one technical solution of the above-mentioned positioning method for autonomous vehicles, the step of "preprocessing the point cloud collected by the lidar to obtain preprocessed point cloud data" includes:

[0034] The point cloud is downsampled to obtain a downsampled point cloud;

[0035] The downsampled point cloud is then subjected to distortion correction to obtain the distortion-corrected point cloud.

[0036] The distortion-free point cloud is segmented into a ground point cloud to obtain a ground point cloud and a non-ground point cloud. The ground point cloud and the non-ground point cloud are used as the preprocessed point cloud data.

[0037] In one technical solution of the above-mentioned autonomous vehicle positioning method, after the step of "matching the lightweight point cloud map with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle", the method further includes:

[0038] The control module performs fusion filtering based on the global positioning result and data collected by other sensors installed on the vehicle.

[0039] In a second aspect, a control device is provided, comprising 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 executed by the processor to perform the positioning method for an autonomous vehicle as described in any of the above-described technical solutions for the positioning method of an autonomous vehicle.

[0040] In a third 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 positioning method for an autonomous vehicle as described in any of the above-described technical solutions.

[0041] In a fourth aspect, a vehicle is provided, the vehicle including the control device described in the above-mentioned control device technical solution.

[0042] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects:

[0043] In implementing the technical solution of this invention, a lightweight point cloud map is obtained based on the initial positioning result acquired by the vehicle's fusion positioning module, and a global positioning result for the vehicle is obtained based on the lightweight point cloud map and the point cloud collected by the LiDAR. Through the above configuration, this invention can achieve global positioning of the vehicle by matching the point cloud collected by the vehicle's LiDAR with the lightweight point cloud map, providing accurate global positioning results even when the RTK signal is poor. Furthermore, because the lightweight point cloud map significantly reduces the map size, matching the lightweight point cloud map with the point cloud collected by the LiDAR reduces the computational complexity of the matching process while achieving higher accuracy in the vehicle's global positioning result, reaching centimeter-level positioning. Attached Figure Description

[0044] 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:

[0045] Figure 1This is a schematic flowchart of the main steps of a positioning method for an autonomous vehicle according to an embodiment of the present invention;

[0046] Figure 2 This is a schematic flowchart of the main steps of a positioning method for an autonomous vehicle according to one embodiment of the present invention. Detailed Implementation

[0047] 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.

[0048] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. 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. Non-transitory 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. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0049] See appendix Figure 1 , Figure 1 This is a schematic flowchart illustrating the main steps of a positioning method for an autonomous vehicle according to an embodiment of the present invention. Figure 1 As shown, the positioning method for autonomous vehicles in this embodiment of the invention mainly includes the following steps S101-S103.

[0050] Step S101: Obtain the initial positioning result of the vehicle based on the fusion positioning module installed on the vehicle.

[0051] In this embodiment, the initial positioning result of the vehicle can be obtained based on the fusion positioning module installed on the vehicle. The initial positioning result represents the initial state of the vehicle's positioning as obtained by the fusion positioning module based on the vehicle information.

[0052] Step S102: Based on the initial positioning results, obtain a lightweight point cloud map within a preset distance range of the initial positioning results.

[0053] In this embodiment, a lightweight point cloud map within a preset distance range of the initial positioning result can be obtained based on the initial positioning result. Those skilled in the art can set the value of the preset distance according to the needs of actual applications.

[0054] In one implementation, the lightweight point cloud map is a point cloud map represented using a normal distribution. Specifically, voxels of a preset size are selected, the point cloud map is divided according to voxel size, and the mean and variance of the point cloud within each voxel are calculated. The mean and variance are then used to represent the normal distribution of the point cloud within the current voxel, thus obtaining the lightweight point cloud map. Processing the point cloud map into a lightweight form can significantly reduce its size while preserving its geometric features. For example, taking a service area as an example, the size of a lightweight point cloud map for a one-square-kilometer service area can be compressed to approximately 10MB, greatly reducing map download bandwidth and increasing the possibility of large-scale applications.

[0055] In one implementation, the preset distance is 150m.

[0056] Step S103: Match the lightweight point cloud map with the point cloud collected by the LiDAR installed on the vehicle to obtain the global positioning result of the vehicle.

[0057] In this embodiment, a lightweight point cloud map can be matched with the point cloud collected by the LiDAR installed on the vehicle to obtain the vehicle's global positioning result. That is, global positioning of the vehicle is achieved by matching the lightweight point cloud map with the collected point cloud. Here, global positioning refers to the vehicle's location within a global map.

[0058] In one implementation, the NDT (Normal-Distributions Transform) algorithm can be used to match the lightweight point cloud map with the point cloud acquired by LiDAR.

[0059] In one implementation, the fusion positioning module can perform fusion filtering based on data collected by other sensors installed on the vehicle and the global positioning results. The data collected by the other sensors on the vehicle may include the vehicle's gear speed, acceleration, initial positioning results, etc.

[0060] Based on steps S101-S103 above, this embodiment of the invention obtains a lightweight point cloud map based on the initial positioning result acquired by the vehicle's fusion positioning module, and obtains the vehicle's global positioning result based on the lightweight point cloud map and the point cloud collected by the LiDAR. Through this configuration, this embodiment of the invention can achieve global vehicle positioning by matching the point cloud collected by the vehicle's LiDAR with the lightweight point cloud map, providing accurate global positioning results even when the RTK signal is poor. Furthermore, because the lightweight point cloud map significantly reduces the map size, matching the lightweight point cloud map with the point cloud collected by the LiDAR reduces the computational complexity of the matching process while achieving higher accuracy in the vehicle's global positioning result, reaching centimeter-level positioning.

[0061] The following is a further explanation of step S103.

[0062] In one embodiment of the present invention, step S103 may further include the following steps S1031 and S1032:

[0063] Step S1031: Based on the parameters of the lightweight point cloud map, process the point cloud collected by the lidar installed on the vehicle to obtain NDT point cloud data.

[0064] In this embodiment, step S1031 may further include the following steps S10311 and S10312:

[0065] Step S10311: Preprocess the point cloud data collected by the lidar to obtain preprocessed point cloud data.

[0066] In this embodiment, step S10311 may further include steps S103111 to S103113:

[0067] Step S103111: Downsample the point cloud to obtain the downsampled point cloud.

[0068] In this embodiment, downsampling the point cloud can remove a large amount of redundant data and reduce the computation time. Common point cloud downsampling methods in the art can be applied, such as voxel mesh downsampling, uniform downsampling, geometric downsampling, and random downsampling.

[0069] Step S103112: Perform distortion correction processing on the downsampled point cloud to obtain the distortion-corrected point cloud.

[0070] In this embodiment, distortion correction processing is performed on the point cloud to correct point cloud distortion caused by motion and other factors. Common point cloud distortion correction methods in the art can be applied, such as pure estimation methods, sensor-assisted methods, and fusion methods.

[0071] Step S103113: Perform ground point cloud segmentation on the distortion-removed point cloud to obtain ground point cloud and non-ground point cloud, and use the ground point cloud and non-ground point cloud as preprocessed point cloud data.

[0072] In this embodiment, the distortion-reduced point cloud can be segmented into a ground point cloud to obtain a ground point cloud and a non-ground point cloud. The ground point cloud and the non-ground point cloud are used as preprocessed point cloud data to facilitate matching of point clouds of the same category during matching. Commonly used ground point cloud segmentation methods in the art can be applied for ground point cloud segmentation, such as planar raster method, point cloud normal vector method, model fitting method, and surface mesh method.

[0073] Step S10312: Based on the parameters of the lightweight point cloud map, convert the preprocessed point cloud data into NDT point cloud data with the same normal distribution expression as the lightweight point cloud map.

[0074] In this embodiment, the preprocessed point cloud data can be converted into a lightweight point cloud map with the same normal distribution representation based on the voxel size of the lightweight point cloud, thereby obtaining NDT point cloud data. Converting point cloud data to NDT point cloud data can reduce the amount of observation information and decrease the computational complexity of matching with the lightweight point cloud map without losing the geometric information of the point cloud data.

[0075] Step S1032: Apply the NDT-D2D algorithm to match the lightweight point cloud map with the NDT point cloud data to obtain the global positioning result of the vehicle.

[0076] In this embodiment, step S1032 may further include steps S10321 to S10323:

[0077] Step S10321: Based on the initial positioning results, obtain the transformation matrix for converting NDT point cloud data to the map coordinate system of lightweight point cloud map.

[0078] In this embodiment, a transformation matrix for converting NDT point cloud data to a lightweight point cloud map coordinate system can be obtained based on the initial positioning results.

[0079] Step S10322: Apply the NDT-D2D algorithm, perform multiple iterations, and optimize the transformation matrix to match the NDT point cloud data with the lightweight point cloud map, so as to obtain the final transformation matrix.

[0080] In this embodiment, the lightweight point cloud map includes multiple map voxels, and the NDT point cloud data includes multiple point cloud data voxels. The NDT-D2D (Distribution to Distribution) algorithm is a distribution-to-distribution NDT algorithm. The NDT-D2D algorithm voxels the source point cloud and the target point cloud, calculates the distance between the distributions of the source point cloud and the target point cloud, and uses cumulative distance error for control.

[0081] In one embodiment, step S10322 may further include steps S103221 to S103225:

[0082] Step S103221: For each iteration, apply the NDT-D2D algorithm to transform the NDT point cloud data to the map coordinate system of the lightweight point cloud map according to the transformation matrix, and obtain matching pairs consisting of map voxels and point cloud data voxels.

[0083] In this embodiment, step S103221 may further include the following steps S1032211 and S1032212:

[0084] Step S1032211: For each point cloud data voxel transformed to the map coordinate system, apply the nearest neighbor algorithm to obtain multiple matching map voxels.

[0085] Step S1032212: Based on the normal distribution parameters of map voxels and point cloud data voxels, perform similarity analysis on multiple matching map voxels to obtain map voxels that correspond one-to-one with point cloud data voxels, so as to form matching pairs.

[0086] In this embodiment, for each point cloud data voxel transformed to the map coordinate system, a nearest neighbor algorithm can be applied first to obtain multiple matching map voxels. Then, based on the overall distribution parameters, a similarity analysis is performed on the multiple matching map voxels to obtain map voxels that correspond one-to-one with the point cloud data voxels. In this way, the one-to-one corresponding point cloud data voxels and map voxels can form matching pairs. The normal distribution parameters may include the mean and covariance.

[0087] In one implementation, similarity analysis can be performed based on the KL (Kullback-Leibler) divergence; the smaller the KL divergence, the smaller the difference between the two variables. The KL divergence is obtained from the probability distribution function of the two random variables, which in turn can be derived from the normal distribution parameters (mean, covariance).

[0088] In one implementation, when matching NDT point cloud data with lightweight point cloud maps, matching is required for point cloud types of the same category. For example, if a map voxel is a ground point cloud, it needs to be matched with point cloud data voxels that are also ground point clouds; if a map voxel is a non-ground point cloud, it needs to be matched with point cloud data voxels that are also non-ground point clouds.

[0089] Step S103222: Construct the residual function for the current iteration based on the distribution distance of the matching pairs.

[0090] In this embodiment, step S103222 can be further configured as follows:

[0091] Based on the distribution distances of all matching pairs, construct the residual function for the current iteration; where the residual function is positively correlated with the distribution distance; the distribution distance includes the mean distance of the normal distribution and the covariance difference of the normal distribution.

[0092] In one implementation, the residual function is the sum of the distribution distances of all matched pairs. The distribution distance can include the physical distance between the two normal distributions in the matched pair (measured by the mean distance) and the KL divergence (measured by applying the mean distance and covariance difference).

[0093] Step S103223: Calculate the gradient of the current iteration based on the residual function.

[0094] Step S103224: Apply gradient descent to update the transformation matrix based on the gradient, and then proceed to the next iteration.

[0095] In this embodiment, the gradient of the current iteration can be calculated based on the residual function, and the gradient descent method can be applied to update the transformation matrix based on the gradient, thereby proceeding to the next iteration.

[0096] Step S103225: When the iteration meets the preset cutoff condition, the iteration ends to obtain the final transformation matrix.

[0097] In this embodiment, if the iteration meets the cutoff condition, the iteration can be terminated, and the transformation matrix obtained at the end of the iteration is the final transformation matrix.

[0098] In one implementation, the iteration is terminated when the gradient value of the iteration is less than a preset gradient threshold. Those skilled in the art can set the value of the gradient threshold according to the needs of the actual application.

[0099] Step S10323: Obtain the global positioning result of the vehicle based on the final transformation matrix.

[0100] In this embodiment, the global positioning result of the vehicle in the map coordinate system of the lightweight point cloud map can be obtained based on the final transformation matrix.

[0101] In one implementation, see Appendix Figure 2 , Figure 2 This is a schematic flowchart illustrating the main steps of a positioning method for an autonomous vehicle according to one embodiment of the present invention. Figure 2 As shown, the localization method for autonomous vehicles can obtain initial localization results through a fusion localization module. A point cloud preprocessing module performs downsampling, ROI (region of interest) processing, point cloud distortion correction, and ground point cloud segmentation on the point cloud collected by LiDAR. The map engine obtains a lightweight point cloud map based on the initial localization results. A D2D optimization and registration module matches the lightweight point cloud map with the preprocessed point cloud to obtain global localization results, which are then sent to the fusion localization module for fusion filtering. The global localization result can achieve an accuracy of up to 20cm.

[0102] The fusion positioning module and the D2D optimization registration module are loosely coupled. The fusion positioning module provides the initial positioning result to the D2D optimization registration module, while the D2D optimization registration module provides the fusion positioning module with a more accurate global positioning result. This loose coupling ensures the fusion filtering is in a healthy state, maximizing the reuse of modules on the vehicle, optimizing computing power, and supporting multiple services.

[0103] 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.

[0104] 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.

[0105] Furthermore, the present invention also provides a control device. In one embodiment of the control device according to the present invention, the control device includes a processor and a storage device. The storage device can be configured to store a program for executing the positioning method of the autonomous vehicle of the above-described method embodiments. The processor can be configured to execute the program in the storage device, which includes, but is not limited to, a program for executing the positioning method of the autonomous vehicle of the above-described method embodiments. 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.

[0106] In embodiments of the present invention, the control device may be a control device device comprising various electronic devices. In some possible implementations, the control device may include multiple storage devices and multiple processors. The program executing the autonomous vehicle localization method of the above method embodiments can be divided into multiple subroutines, each subroutine can be loaded and run by a processor to execute different steps of the autonomous vehicle localization method of the above method embodiments. Specifically, each subroutine can be stored in different storage devices, and each processor can be configured to execute programs in one or more storage devices to jointly implement the autonomous vehicle localization method of the above method embodiments, that is, each processor executes different steps of the autonomous vehicle localization method of the above method embodiments to jointly implement the autonomous vehicle localization method of the above method embodiments.

[0107] The aforementioned multiple processors can be processors deployed on the same device. For example, the aforementioned control device can be a high-performance device composed of multiple processors, and the aforementioned multiple processors can be processors configured on that high-performance device. Alternatively, the aforementioned multiple processors can also be processors deployed on different devices. For example, the aforementioned control device can be a server cluster, and the aforementioned multiple processors can be processors on different servers within the server cluster.

[0108] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for performing the positioning method of an autonomous vehicle according to the above-described method embodiments. This program can be loaded and run by a processor to implement the positioning method of the autonomous vehicle described above. 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 can 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.

[0109] Furthermore, the present invention also provides a vehicle. In one embodiment of the present invention, the vehicle may include a control device as described in the control device embodiment.

[0110] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0111] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0112] The technical solution of the present invention has been described above with reference to the preferred embodiments 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 after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for locating an autonomous vehicle, characterized in that, The method includes: Based on the fusion positioning module installed on the vehicle, the initial positioning result of the vehicle is obtained; Based on the initial positioning result, a lightweight point cloud map within a preset distance range of the initial positioning result is obtained; The lightweight point cloud map is matched with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle. The step of "matching the lightweight point cloud map with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle" includes: Based on the parameters of the lightweight point cloud map, the point cloud collected by the lidar installed on the vehicle is processed to obtain NDT point cloud data. The NDT-D2D algorithm is applied to match the lightweight point cloud map with the NDT point cloud data to obtain the global positioning result of the vehicle. The step of "applying the NDT-D2D algorithm to match the lightweight point cloud map with the NDT point cloud data to obtain the global positioning result of the vehicle" includes: Based on the initial positioning results, obtain the transformation matrix for converting the NDT point cloud data to the map coordinate system of the lightweight point cloud map; The NDT-D2D algorithm is applied and iterated multiple times to optimize the transformation matrix, so that the NDT point cloud data and the lightweight point cloud map are matched to obtain the final transformation matrix. Based on the final transformation matrix, the global positioning result of the vehicle is obtained; The lightweight point cloud map includes multiple map voxels, and the NDT point cloud data includes multiple point cloud data voxels. The step of "applying the NDT-D2D algorithm, performing multiple iterations, and optimizing the transformation matrix to match the NDT point cloud data with the lightweight point cloud map to obtain the final transformation matrix" includes: For each iteration, the NDT-D2D algorithm is applied to transform the NDT point cloud data into the map coordinate system of the lightweight point cloud map according to the transformation matrix, and a matching pair consisting of map voxels and point cloud data voxels is obtained. Based on the distribution distance of the matching pairs, construct the residual function for the current iteration; Calculate the gradient of the current iteration based on the residual function; The gradient descent method is applied to update the transformation matrix based on the gradient, and then the next iteration is performed. When the iteration meets the preset cutoff condition, the iteration ends to obtain the final transformation matrix.

2. The positioning method for an autonomous vehicle according to claim 1, characterized in that, The steps to "obtain a one-to-one matching pair consisting of point cloud map voxels and point cloud data voxels" include: For each point cloud data voxel transformed to the map coordinate system, the nearest neighbor algorithm is applied to obtain multiple matching map voxels; Based on the normal distribution parameters of the map voxels and the point cloud data voxels, a similarity analysis is performed on multiple matching map voxels to obtain map voxels that correspond one-to-one with the point cloud data voxels, thus forming matching pairs.

3. The positioning method for an autonomous vehicle according to claim 1, characterized in that, The step of "constructing the residual function for the current iteration based on the distribution distance of the matching pairs" includes: Construct the residual function for the current iteration based on the distribution distances of all matching pairs; Wherein, the residual function is positively correlated with the distribution distance; the distribution distance includes the mean distance of the normal distribution and the covariance difference of the normal distribution; and / or, The cutoff condition for the iteration is that the gradient value of the iteration is less than a preset gradient threshold.

4. The positioning method for an autonomous vehicle according to claim 1, characterized in that, The lightweight point cloud map is a point cloud map represented using a normal distribution. The step of "processing the point cloud collected by the lidar installed on the vehicle according to the parameters of the lightweight point cloud map to obtain NDT point cloud data" includes: The point cloud collected by the lidar is preprocessed to obtain preprocessed point cloud data; Based on the parameters of the lightweight point cloud map, the preprocessed point cloud data is converted into NDT point cloud data with the same normal distribution expression as the lightweight point cloud map.

5. The positioning method for an autonomous vehicle according to claim 4, characterized in that, The step of "preprocessing the point cloud acquired by the lidar to obtain preprocessed point cloud data" includes: The point cloud is downsampled to obtain a downsampled point cloud; The downsampled point cloud is then subjected to distortion correction to obtain the distortion-corrected point cloud. The distortion-free point cloud is segmented into a ground point cloud to obtain a ground point cloud and a non-ground point cloud. The ground point cloud and the non-ground point cloud are used as the preprocessed point cloud data.

6. The positioning method for an autonomous vehicle according to claim 1, characterized in that, After the step of "matching the lightweight point cloud map with the point cloud collected by the lidar installed on the vehicle to obtain the global positioning result of the vehicle", the method further includes: The control module performs fusion filtering based on the global positioning result and data collected by other sensors installed on the vehicle.

7. A control 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 positioning method for the autonomous vehicle according to any one of claims 1 to 6.

8. 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 localization method for an autonomous vehicle as described in any one of claims 1 to 6.

9. A vehicle, characterized in that, The vehicle includes the control device as described in claim 7.

Images