Motor vehicle mapping and positioning method and device and computer readable storage medium

Abstract

The embodiment of the invention provides a motor vehicle mapping and positioning method and device and a computer readable storage medium. The method comprises the following steps: obtaining a look-around spliced two-dimensional diagram based on video images of multiple paths of vehicle-mounted cameras; performing semantic segmentation on the road marks in the look-around spliced two-dimensional graph to obtain line segment template images and parking space template images; performing image processing on the line segment template image and the parking space template image to obtain an initial two-dimensional semantic image; performing pixel clustering and parameterization processing on the initial two-dimensional semantic image to obtain a target two-dimensional semantic image; projecting the target two-dimensional semantic image to a vehicle coordinate system based on the speedometer pose to obtain an initial local map, and clustering and fusing continuous multiple frames of initial local maps to obtain a precise local map; projecting the accurate local map to a longitude and latitude coordinate system to obtain a target local map, and further obtaining a lightweight road semantic map; and matching and positioning the motor vehicle. According to the embodiment of the invention, mapping and positioning can be realized in a scene with less textures.

Description

Technical Field

[0001] The present invention relates to the field of motor vehicle mapping and positioning technology, and in particular to a method, apparatus and computer-readable storage medium for motor vehicle mapping and positioning. Background Technology

[0002] As motor vehicles become increasingly automated, car manufacturers are starting to equip them with valet parking systems at the factory to enable automatic parking. In practice, valet parking systems need to use SLAM (Simultaneous Localization and Mapping) technology to build parking maps of the parking area.

[0003] Currently, several methods are commonly used to locate vehicles in parking map construction: localization based on vehicle wheel speed odometer and onboard image fusion, monocular camera visual scene localization, image feature point matching localization based on onboard images, and vehicle localization methods combining lightweight semantic maps and particle filter frameworks. However, the first method suffers from significant pose calculation errors when wheels slip; the second method leads to localization errors in similar scenarios; the third method relies heavily on image features, affecting localization accuracy in poor lighting conditions or when scene feature points and textures are scarce; and the fourth method, through repeated sampling in large spaces, may accidentally lose particles near the correct location, leading to localization errors, and is also highly dependent on lighting conditions. Clearly, existing localization methods all have some shortcomings. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a method for mapping and locating motor vehicles, which can realize mapping and locating in scenes with few textures.

[0005] A further technical problem to be solved by the embodiments of the present invention is to provide a vehicle mapping and positioning device that can realize mapping and positioning in scenes with few textures.

[0006] A further technical problem to be solved by the embodiments of the present invention is to provide a computer-readable storage medium for storing a computer program that can perform mapping and positioning in scenes with few textures.

[0007] To address the aforementioned technical problems, this invention first provides the following technical solution: a method for mapping and locating motor vehicles, comprising the following steps: A surround-view stitched 2D map is obtained by processing video images collected from multiple vehicle-mounted cameras. A pre-stored semantic segmentation model is used to perform semantic segmentation on the road signs in the surround view stitched 2D image to obtain the line segment template image and parking space template image in the surround view stitched 2D image respectively; The line segment template image and the parking space template image are processed and merged to obtain an initial two-dimensional semantic image; The target two-dimensional semantic image is obtained by performing pixel clustering and parameterization on different road signs in the initial two-dimensional semantic image; Based on the odometry pose output by the integrated navigation and positioning system, the target two-dimensional semantic image is projected onto the vehicle coordinate system to obtain an initial local map. Clustering and fusing the same map features in multiple consecutive frames of the initial local map are used to obtain an accurate local map. A precise local map is projected onto a latitude and longitude coordinate system to obtain a target local map, and the target local map is corrected and optimized to obtain a lightweight road semantic map; and The vehicle is located by matching based on the lightweight road semantic map and the image features in the current image frame.

[0008] Furthermore, the step of processing video images acquired by multiple vehicle-mounted cameras to obtain a surround-view stitched two-dimensional image specifically includes: extracting image frames frame by frame from the video images acquired and transmitted by the multiple vehicle-mounted cameras, and performing image processing on the image frames to obtain a surround-view stitched two-dimensional image. The image processing includes image correction, viewpoint transformation, and image stitching.

[0009] Furthermore, the line segment template image includes: horizontal line segments, vertical line segments, and arrow markers; the parking space template image includes: pixel coordinates of the four corner points of the parking space; the step of processing and merging the line segment template image and the parking space template image to obtain the initial semantic image specifically includes: detecting and filtering various line segments in the line segment template image and the pixel coordinates of each parking space corner point in the parking space template image, and incorporating all horizontal and vertical line segments with lengths greater than the corresponding average length, all arrow markers, and parking lines generated based on the pixel coordinates of each parking space corner point into the same template to form the initial two-dimensional semantic image.

[0010] Furthermore, in the target two-dimensional semantic image, the parameter information of each arrow after clustering includes the reference point, main direction, bounding box, and uncertainty of the relative distance between the bounding box lines; the parameter information of each horizontal and vertical line segment after clustering includes multiple short line segments and uncertainty of the endpoints of each short line segment.

[0011] Furthermore, the clustering fusion of identical map features in multiple consecutive frames of the initial local map specifically includes: clustering and fusing identical arrow markers in multiple frames of the initial local map based on the intersection of bounding boxes in the parameter information of the arrow markers; and clustering and fusing identical line features in multiple frames of the initial local map based on the local direction, lateral distance, and longitudinal distance of neighboring points.

[0012] Furthermore, the step of correcting and optimizing the target local map to obtain a lightweight road semantic map specifically includes: fusing the target local maps constructed at multiple different time periods to correct the deviations of the target local maps.

[0013] Furthermore, the process of matching and locating motor vehicles based on the lightweight road semantic map and image features in the current image frame specifically includes: Based on the lightweight road semantic map, within a predetermined range centered on the initial pose of the vehicle provided by the integrated navigation and positioning system, map feature matching is performed using image features from the current image frame to correct the initial pose and obtain a coarse positioning pose; and The coarse positioning pose is iteratively optimized and re-associated using a sliding window factor graph to obtain the fine positioning pose of the vehicle. The factor graph includes odometer factors and map feature matching factors.

[0014] Furthermore, the step of using a sliding window factor graph to iteratively optimize and reassociate the initial pose of the vehicle to obtain the precise positioning pose of the vehicle specifically includes: The odometry factor is used to provide relative pose constraints, longitudinal position constraints are provided for the lateral line segments, and lateral position constraints are provided for the longitudinal line segments. Based on the map feature matching factors, point-to-line distances are constructed and optimized. For horizontal and vertical line segments, the point-to-line distance between each sampling point on the line segment in the current image frame and the corresponding matching horizontal or vertical line segment in the lightweight road semantic map is optimized. For arrow markers, the point-to-line distance between each sampling point on the bounding box of the arrow marker in the current image frame and the bounding box lines of the corresponding matching arrow marker in the target positioning map is optimized. The minimum value of the sum of the distances between the points and lines is obtained by iteratively solving the problem. During the iteration process, map feature matching and association are performed again based on the coarse localization pose optimized in each iteration.

[0015] On the other hand, in order to solve the above-mentioned further technical problems, the present invention provides the following technical solution: a motor vehicle mapping and positioning device, which is connected to a combined navigation and positioning system and a multi-channel vehicle camera installed on a motor vehicle, respectively. The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the motor vehicle mapping and positioning method as described in any of the above claims.

[0016] Furthermore, in order to solve the aforementioned technical problems, the present invention provides the following technical solution: a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the motor vehicle mapping and positioning method as described above.

[0017] After adopting the above technical solution, the embodiments of the present invention have at least the following beneficial effects: The embodiments of the present invention process images captured by various vehicle-mounted cameras to form a surround-view stitched two-dimensional image, and then sequentially perform semantic segmentation, image processing and merging, pixel clustering, and parameterization to obtain a target two-dimensional semantic image. In scenarios with less texture, the map can be constructed based on fixed ground markings such as parking spaces, lane lines, and speed bumps in the target two-dimensional semantic image, improving the robustness of the system. Furthermore, the target two-dimensional semantic image is further projected onto the vehicle coordinate system using the odometer pose output by the vehicle's integrated navigation and positioning system. An initial local map is obtained, and the odometry pose and map features are tightly coupled, which can effectively improve the accuracy and efficiency of mapping and positioning, and avoid pose estimation errors caused by wheel slippage. Finally, a high-precision accurate local map is obtained through clustering and fusion, and the accurate local map is projected onto the latitude and longitude coordinate system to obtain the target local map. The target local map is corrected and optimized to obtain a lightweight road semantic map. Based on the lightweight road semantic map and the image features in the current image frame, the vehicle is matched and located. The impact of data errors (e.g., cumulative error, RTK signal drift) is small and the positioning accuracy is high. Attached Figure Description

[0018] Figure 1 This is a flowchart of an optional embodiment of the vehicle mapping and positioning method of the present invention.

[0019] Figure 2 The flowchart below shows a specific step S7 of an optional embodiment of the vehicle mapping and positioning method of the present invention.

[0020] Figure 3 The flowchart below shows a specific step S72 of an optional embodiment of the vehicle mapping and positioning method of the present invention.

[0021] Figure 4 This is a schematic diagram of an optional embodiment of the vehicle mapping and positioning device of the present invention.

[0022] Figure 5 This is a functional module diagram of an optional embodiment of the vehicle mapping and positioning device of the present invention. Detailed Implementation

[0023] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the following illustrative embodiments and descriptions are only used to explain the present invention and are not intended to limit the present invention. Moreover, the embodiments and features in the embodiments of the present application can be combined with each other unless otherwise specified.

[0024] like Figure 1 As shown, an optional embodiment of the present invention provides a method for mapping and locating motor vehicles, comprising the following steps: S1: Process the video images collected by the multi-channel vehicle-mounted camera 1 to obtain a surround-view stitched 2D map; S2: Use a pre-stored semantic segmentation model to perform semantic segmentation on the road signs in the surround view stitched 2D image to obtain the line segment template image and parking space template image in the surround view stitched 2D image respectively; S3: Perform image processing and merge the line segment template image and the parking space template image to obtain an initial two-dimensional semantic image; S4: Perform pixel clustering and parameterization on the different road signs in the initial two-dimensional semantic image to obtain the target two-dimensional semantic image; S5: Based on the odometer pose output by the integrated navigation positioning system 2 (P-Box), the target two-dimensional semantic image is projected onto the vehicle coordinate system to obtain an initial local map. Clustering and fusing the same map features in multiple consecutive frames of the initial local map are used to obtain an accurate local map. S6: Project the precise local map onto a latitude and longitude coordinate system to obtain a target local map, and correct and optimize the target local map to obtain a lightweight road semantic map; and S7: Match and locate the motor vehicle based on the lightweight road semantic map and the image features in the current image frame.

[0025] This invention, in its embodiments, processes images captured by various vehicle-mounted cameras 1 to form a surround-view stitched 2D map. Then, it sequentially performs semantic segmentation, image processing and merging, pixel clustering, and parameterization to obtain a target 2D semantic image. In scenarios with limited texture, the map can be constructed based on fixed ground markers such as parking spaces, lane lines, and speed bumps in the target 2D semantic image, improving the system's robustness. Furthermore, the target 2D semantic image is projected onto the vehicle coordinate system using the odometer pose output by the vehicle's integrated navigation and positioning system 2 to obtain an initial local map. The odometer pose and map features are tightly coupled, effectively improving the accuracy and efficiency of mapping and positioning, and avoiding pose estimation errors caused by wheel slippage. Finally, a high-precision accurate local map is obtained through clustering and fusion, and this accurate local map is projected onto a latitude and longitude coordinate system to obtain the target local map. Correcting and optimizing the target local map yields a lightweight road semantic map. Based on this lightweight road semantic map and image features in the current image frame, vehicle matching and positioning are performed, minimizing the impact of data errors (e.g., cumulative errors, RTK signal drift) and achieving high positioning accuracy.

[0026] It is understandable that, since the integrated navigation and positioning system 1 relies on IMU data, RTK signals, and odometer information to achieve positioning, the positioning accuracy output by the integrated navigation and positioning system 1 can be at the centimeter level when the vehicle is in a location with strong RTK signals; when the vehicle is in a location with relatively weak RTK signals (e.g., an underground parking lot), the positioning accuracy output by the integrated navigation and positioning system 1 can be at the meter level; the semantic segmentation model can be a semantic segmentation model based on deep learning, such as FCN, U-Net, DeepLab series, PSPNet, SegNet, etc.

[0027] In an optional embodiment of the present invention, step S1 specifically includes: extracting image frames frame by frame from the video images acquired and transmitted by the multiple vehicle-mounted cameras 1, and performing image processing on the image frames to obtain a surround-view stitched two-dimensional image. The image processing includes image correction, viewing angle transformation, and image stitching. In this embodiment, by performing image processing such as image correction, viewing angle transformation, and image stitching on the image frames of the multiple vehicle-mounted cameras 1, a surround-view stitched two-dimensional image can be obtained simply and efficiently. In specific implementation, each vehicle-mounted camera 1 should be pre-calibrated with the ground geometry (camera height, camera installation angle).

[0028] In an optional embodiment of the present invention, the line segment template image includes: horizontal line segments, vertical line segments, and arrow markers; the parking space template image includes: pixel coordinates of the four corner points of the parking space; the step of processing and merging the line segment template image and the parking space template image to obtain an initial semantic image specifically includes: detecting and filtering various line segments in the line segment template image and the pixel coordinates of each corner point of the parking space template image, and incorporating all horizontal and vertical line segments with lengths greater than the corresponding average length, all arrow markers, and parking lines generated based on the pixel coordinates of each corner point of the parking space into the same template to form an initial two-dimensional semantic image. In this embodiment, a line segment template image is formed by extracting horizontal line segments, vertical line segments, and arrow markers from the image. The average length of each line segment is calculated first, and then the length of each line segment is compared with the average length. This retains only horizontal and vertical lines whose actual length is greater than the corresponding average length, thus filtering out a large number of short lines with no practical reference value, improving processing efficiency and reducing interference. Arrow markers, being directional and highly referential lines, are all retained for subsequent map construction. The pixel coordinates of the four corner points of a parking space determine its location and are all retained. The filtered features are then fused into the same template to obtain the initial two-dimensional semantic image.

[0029] In an optional embodiment of the present invention, in the target two-dimensional semantic image, the parameter information of each arrow identifier after clustering includes a reference point, main direction, bounding box, and uncertainty of the relative distance between the bounding box lines; the parameter information of each horizontal and vertical line segment after clustering includes multiple short line segments and uncertainty of the endpoints of each short line segment. In this embodiment, to facilitate subsequent map construction and matching, the clustered pixel blocks in the target two-dimensional semantic image are converted from pixel-level information to parameterized information. Specifically, the parameterized arrow marker mainly includes a reference point, a main direction, a bounding box, and the uncertainty of the relative distances between the bounding box's lines. The reference point is the geometric center or tip of the arrow, the main direction is the direction the arrow points in, which is the direction information for the subsequent map, and the bounding box is the smallest bounding rectangle of the arrow marker. The uncertainty of the relative distances between the bounding box's lines mainly characterizes the confidence level or error range of the arrow marker's position, size, and other parameters when it is detected. For the parameterized horizontal and vertical line segments, a series of short line segments are used to fit and replace a single horizontal or vertical line segment. The uncertainty of the endpoints of each short line segment characterizes the confidence level or error range of the position at each endpoint.

[0030] In an optional embodiment of the present invention, the clustering and fusion of identical map features in multiple consecutive frames of the initial local map specifically includes: clustering and fusing identical arrow markers in multiple frames of the initial local map based on the intersection of bounding boxes in the parameter information of the arrow markers; and clustering and fusing identical line features in multiple frames of the initial local map based on the local direction, lateral distance, and longitudinal distance of neighboring points. In this embodiment, when fusing arrow markers in the initial local map, the intersection of bounding boxes is used to determine whether the same arrow marker is seen multiple times, which is a simple method; while for line features, judgment and fusion are performed based on geometric rules such as proximity and directional consistency, resulting in high fusion accuracy.

[0031] In an optional embodiment of the present invention, step S6 specifically includes: fusing the target local maps constructed at multiple different time periods to correct the deviations in the target local maps. In this embodiment, by fusing the target local maps at multiple different time periods, the deviations in the target local maps are quickly corrected, thereby improving image accuracy.

[0032] In an optional embodiment of the present invention, such as Figure 2 As shown, step S7 specifically includes: S71: Based on the lightweight road semantic map, within a predetermined range (e.g., a 10m radius) centered on the initial pose of the vehicle provided by the integrated navigation and positioning system, map feature matching is performed using image features from the current image frame to correct the initial pose and obtain a coarse positioning pose; and S72: The coarse positioning pose is iteratively optimized and re-associated using a sliding window factor graph to obtain the fine positioning pose of the vehicle. The factor graph includes odometer factors and map feature matching factors.

[0033] In this embodiment, after determining the target positioning map, the approximate initial pose output by the integrated navigation and positioning system 2 is first combined with the spatial distribution mode to perform feature matching search and association, and output a sub-meter level coarse positioning pose; then, a sliding window factor graph optimization is used to finally output a fine positioning pose, improving accuracy and stability.

[0034] In an optional embodiment of the present invention, such as Figure 3 As shown, step S72 specifically includes: S721: The odometry factor is used to provide relative pose constraints, to provide longitudinal position constraints for the lateral line segment, and to provide lateral position constraints for the longitudinal line segment. S722: Based on the map feature matching factors, construct and optimize the point-to-line distance. For horizontal and vertical line segments, optimize the point-to-line distance between each sampling point on the line segment in the current image frame and the corresponding matching horizontal or vertical line segment in the lightweight road semantic map. For arrow markers, optimize the point-to-line distance between each sampling point on the bounding box of the arrow marker in the current image frame and the bounding box lines of the corresponding matching arrow marker in the target positioning map; and S723: Iteratively solve for the minimum value of the sum of the distances between the points and lines to obtain the fine localization pose, and in the process of iteration, re-perform map feature matching and association based on the coarse localization pose optimized in each iteration.

[0035] In this embodiment, during precise pose optimization, the odometry factor is used to provide relative pose constraints to achieve constraints between image frames. A graph feature matching factor is used to construct relative distances. Specifically: for line features, the distance from sampling points on line segments in the current image frame to the corresponding line segments in the target positioning map is calculated. Horizontal lines primarily constrain the vehicle's Y (longitudinal) coordinate, and longitudinal lines primarily constrain the vehicle's X (lateral) coordinate. For arrow markers, the distance from sampling points on the arrow bounding box in the current image frame to the four sides of the corresponding arrow bounding box in the target positioning map is calculated. Arrow markers can simultaneously constrain the vehicle's (lateral) X coordinate, (longitudinal) Y coordinate, and yaw angle. Then, the precise positioning pose is obtained by iteratively solving for the minimum sum of each distance. Furthermore, during the iteration process, map feature matching and association are re-performed based on the coarse positioning pose optimized in each iteration. That is, if an abnormally large matching error is found during optimization (potentially a mismatch), this matching relationship can be discarded, and a correct association can be found again, thereby improving the system's robustness in handling similar scenarios.

[0036] On the other hand, such as Figure 4 As shown, this embodiment of the invention provides a vehicle mapping and positioning device 3, which is connected to a combined navigation and positioning system 2 and a multi-channel vehicle camera 1 installed on a vehicle. It includes a processor 30, a memory 32, and a computer program stored in the memory 32 and configured to be executed by the processor 30. When the processor 30 executes the computer program, it implements the vehicle mapping and positioning method as described in any of the above embodiments.

[0037] For example, the computer program can be divided into one or more modules / units, which are stored in the memory 32 and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the motor vehicle mapping and positioning device 3. For example, the computer program can be divided into... Figure 5 The functional modules in the motor vehicle mapping and positioning device 3 include the two-dimensional map generation module 41, the semantic segmentation module 42, the initial semantic image generation module 43, the target semantic image generation module 44, the precise local map generation module 45, the positioning map generation module 46, and the matching positioning module 47, which respectively perform the above steps S1-S7.

[0038] The vehicle mapping and positioning device 3 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The vehicle mapping and positioning device 3 may include, but is not limited to, a processor 30 and a memory 32. Those skilled in the art will understand that the schematic diagram is merely an example of the vehicle mapping and positioning device 3 and does not constitute a limitation on the vehicle mapping and positioning device 3. It may include more or fewer components than shown in the diagram, or combine certain components, or use different components. For example, the vehicle mapping and positioning device 3 may also include input / output devices, network access devices, buses, etc.

[0039] The processor 30 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor 30 is the control center of the vehicle mapping and positioning device 3, connecting all parts of the vehicle mapping and positioning device 3 via various interfaces and lines.

[0040] The memory 32 can be used to store the computer program and / or modules. The processor 30 implements various functions of the vehicle mapping and positioning device 3 by running or executing the computer program and / or modules stored in the memory 32 and calling the data stored in the memory 32. The memory 32 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as image recognition function, image overlay function, etc.), etc.; the data storage area may store data created according to the use of the control device (such as image data, etc.). In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0041] If the functions described in the embodiments of the present invention are implemented in the form of software functional modules or units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the embodiments of the present invention can implement all or part of the processes in the methods described above, or they can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by the processor 30, 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 files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable 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, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0042] In another aspect, embodiments of the present invention provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the motor vehicle mapping and positioning method as described in any of the above embodiments.

[0043] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0044] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the scope of protection of the present invention.

Claims

1. A method for mapping and locating motor vehicles, characterized in that, The method includes the following steps: A surround-view stitched 2D map is obtained by processing video images collected from multiple vehicle-mounted cameras. A pre-stored semantic segmentation model is used to perform semantic segmentation on the road signs in the surround view stitched 2D image to obtain the line segment template image and parking space template image in the surround view stitched 2D image respectively; The line segment template image and the parking space template image are processed and merged to obtain an initial two-dimensional semantic image; The target two-dimensional semantic image is obtained by performing pixel clustering and parameterization on different road signs in the initial two-dimensional semantic image; Based on the odometry pose output by the integrated navigation and positioning system, the target two-dimensional semantic image is projected onto the vehicle coordinate system to obtain an initial local map. Clustering and fusing the same map features in multiple consecutive frames of the initial local map are used to obtain an accurate local map. The precise local map is projected onto a latitude and longitude coordinate system to obtain a target local map, and the target local map is corrected and optimized to obtain a lightweight road semantic map; and The vehicle is located by matching based on the lightweight road semantic map and the image features in the current image frame.

2. The method for mapping and locating motor vehicles as described in claim 1, characterized in that, The process of processing video images acquired by multiple vehicle-mounted cameras to obtain a surround-view stitched two-dimensional image specifically includes: extracting image frames frame by frame from the video images acquired and transmitted by each vehicle-mounted camera, and performing image processing on each of the image frames to obtain a surround-view stitched two-dimensional image. The image processing includes image correction, viewpoint transformation, and image stitching.

3. The method for mapping and locating motor vehicles as described in claim 1, characterized in that, The line segment template image includes: horizontal line segments, vertical line segments, and arrow markers; the parking space template image includes: pixel coordinates of the four corner points of the parking space; the step of processing and merging the line segment template image and the parking space template image to obtain the initial semantic image specifically includes: detecting and filtering various line segments in the line segment template image and the pixel coordinates of each parking space corner point in the parking space template image, and incorporating all horizontal and vertical line segments with lengths greater than the corresponding average length, all arrow markers, and parking lines generated based on the pixel coordinates of each parking space corner point into the same template to form the initial two-dimensional semantic image.

4. The method for mapping and locating motor vehicles as described in claim 3, characterized in that, In the target two-dimensional semantic image, the parameter information of each arrow after clustering includes the reference point, main direction, bounding box, and uncertainty of the relative distance between the bounding box lines; the parameter information of each horizontal and vertical line segment after clustering includes multiple short line segments and uncertainty of the endpoints of each short line segment.

5. The method for mapping and locating motor vehicles as described in claim 4, characterized in that, The clustering fusion of identical map features in multiple consecutive frames of the initial local map specifically includes: clustering and fusing identical arrow markers in multiple frames of the initial local map based on the intersection of bounding boxes in the parameter information of the arrow markers; and clustering and fusing identical line features in multiple frames of the initial local map based on the local direction, horizontal distance, and vertical distance of neighboring points.

6. The method for mapping and locating motor vehicles as described in claim 1, characterized in that, The step of correcting and optimizing the target local map to obtain a lightweight road semantic map specifically includes: fusing the target local maps constructed at multiple different time periods to correct the deviations of the target local maps.

7. The method for mapping and locating motor vehicles as described in claim 1, characterized in that, The process of matching and locating motor vehicles based on the lightweight road semantic map and image features in the current image frame specifically includes: Based on the lightweight road semantic map, within a predetermined range centered on the initial pose of the vehicle provided by the integrated navigation and positioning system, map feature matching is performed using image features from the current image frame to correct the initial pose and obtain a coarse positioning pose; and The coarse positioning pose is iteratively optimized and re-associated using a sliding window factor graph to obtain the fine positioning pose of the vehicle. The factor graph includes odometer factors and map feature matching factors.

8. The method for mapping and locating motor vehicles as described in claim 7, characterized in that, The step of using a sliding window factor graph to iteratively optimize and reassociate the initial pose of the vehicle to obtain its precise positioning pose specifically includes: The odometry factor is used to provide relative pose constraints, longitudinal position constraints are provided for the lateral line segments, and lateral position constraints are provided for the longitudinal line segments. Based on the map feature matching factors, point-to-line distances are constructed and optimized. For horizontal and vertical line segments, the point-to-line distance between each sampling point on the line segment in the current image frame and the corresponding matching horizontal or vertical line segment in the lightweight road semantic map is optimized. For arrow markers, the point-to-line distance between each sampling point on the bounding box of the arrow marker in the current image frame and the bounding box lines of the corresponding matching arrow marker in the target positioning map is optimized. The minimum value of the sum of the distances between the points and lines is obtained by iteratively solving the problem. During the iteration process, map feature matching and association are performed again based on the coarse localization pose optimized in each iteration.

9. A vehicle mapping and positioning device, connected to a combined navigation and positioning system and a multi-channel vehicle-mounted camera installed on a vehicle, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the motor vehicle mapping and positioning method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the motor vehicle mapping and positioning method as described in any one of claims 1 to 8.

Images