Map data updating method and device, equipment and storage medium

Abstract

This application discloses a map data updating method, apparatus, device, and storage medium, relating to the field of map navigation technology. The method includes: acquiring multiple image sequences of the same geographical location through multi-pass acquisition; aligning the positions of the same ground vehicle information feature in different images within the multiple image sequences, and generating an ordered database based on the alignment results, recording the status information of the ground vehicle information feature at different acquisition time points; writing road element entities from the ordered database into a map master database, establishing a one-to-one correspondence between the ordered database and the map master database, where road element entities include point elements and relational elements; and automatically converting relational elements written into the map master database into road segment connectivity relationships for navigation path planning based on the arrow meaning and road network topology. This solution significantly reduces manual operation costs and substantially improves the efficiency of map data updating.

Description

Technical Field

[0001] This application relates to the field of map navigation technology, and in particular to a map data updating method, apparatus, device and storage medium. Background Technology

[0002] Map data is the foundation for intelligent navigation and autonomous driving, and its core lies in the accuracy of road features. Road features include point features such as speed limit signs, as well as relational features such as ground vehicle information (e.g., straight arrows, turn arrows). With rapid urban development and frequent changes in road features, how to efficiently and accurately update the map database with these real-world changes has become a persistent technical challenge in this field.

[0003] Currently, map data updates typically employ an "image recognition + manual editing" model. This involves acquiring road images using vehicle-mounted data acquisition equipment, utilizing deep learning networks to perform semantic recognition on the images, and extracting the category and location information of road features. Subsequently, professionals manually review and correct the data, and the confirmed features are then updated to the map master database.

[0004] However, the above solutions primarily target the identification and updating of point features. Relational features, such as ground vehicle information, essentially describe the turning relationships between lanes in different directions at intersections and cannot be directly mounted to the map's master database as point features. This makes it impossible to effectively and automatically convert relational features into road network topology relationships (such as road segment connectivity), resulting in a significant need for manual intervention in processing relational features and low map data update efficiency. Summary of the Invention

[0005] This application provides a map data update method, apparatus, device, and storage medium. The technical solution provided by this application includes the following aspects.

[0006] According to one aspect of the embodiments of this application, a map data updating method is provided, the method comprising: Multiple image sequences of the same geographical location are acquired through multi-pass acquisition. The positions of the same ground vehicle information element in different images in the multiple image sequences are aligned, and an ordered database containing the state information of the ground vehicle information element at different acquisition time points is generated based on the alignment results. Write the road element entities in the order library into the map parent library to establish a one-to-one correspondence between the road element entities in the order library and the map parent library. The road element entities include point elements and relational elements. For relational elements written into the map master library, they are automatically converted into road segment connectivity relationships for navigation route planning based on the meaning of the arrows and the road network topology.

[0007] According to one aspect of the embodiments of this application, a map data updating apparatus is provided, the apparatus comprising: The acquisition module is used to acquire multiple image sequences of the same geographical location through multi-pass acquisition. The generation module is used to align the positions of the same ground vehicle signal element in different images in the multiple image sequences, and generate an ordered library based on the alignment results, which records the state information of the ground vehicle signal element at different acquisition time points. The relationship module is used to write road element entities from the order library into the map master library and establish a one-to-one correspondence between road element entities between the order library and the map master library. The road element entities include point elements and relational elements. The conversion module is used to automatically convert relational elements written into road segment connectivity relationships for navigation route planning, based on the arrow meanings and road network topology of the relational elements.

[0008] According to one aspect of the embodiments of this application, a computer device is provided, the computer device including a processor and a memory, the memory storing a computer program, the computer program being loaded and executed by the processor to implement the above-described map data update method.

[0009] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, the computer program being loaded and executed by a processor to implement the above-described map data update method.

[0010] According to one aspect of the embodiments of this application, a computer program product is provided, the computer program product including a computer program executed by a processor to implement the above-described map data update method.

[0011] This application provides a map data update scheme. By constructing an order database, it aligns and records the status of ground vehicle information elements collected from multiple routes, and establishes a one-to-one correspondence between element entities in the order database and the parent map database, achieving automated identification and fusion of relational elements. Simultaneously, for relational elements such as ground vehicle information, it automatically converts them into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology, eliminating the need for manual intervention in topology construction for turning relationships. When changes occur in the real world, the binding mechanism between the order database and the parent map database automatically updates the parent map database by updating the order database, forming a fully automated closed loop from change detection to data update. This scheme significantly reduces manual operation costs and substantially improves the efficiency of map data updates. Attached Figure Description

[0012] Figure 1 This is a structural block diagram of a computer system provided according to an embodiment of this application; Figure 2 This is a flowchart of a map data update method provided according to an embodiment of this application; Figure 3 This is a flowchart of another map data updating method provided according to an embodiment of this application; Figure 4 This is a schematic diagram illustrating the determination of a homography transformation matrix according to an embodiment of this application; Figure 5 This is a schematic diagram of a process for constructing an order library according to an embodiment of this application; Figure 6 This is a schematic diagram of an image comparison provided according to an embodiment of this application; Figure 7 This is a schematic diagram illustrating a fully automated writing of an order database into a parent database according to an embodiment of this application; Figure 8 This is a schematic diagram illustrating the transformation of a parent database entity into a parent database semantics according to an embodiment of this application; Figure 9 This is a flowchart illustrating a map data update scheme provided according to an embodiment of this application; Figure 10 This is an overall architecture diagram of a map data update scheme provided according to an embodiment of this application; Figure 11 This is a block diagram of a map data updating device according to an embodiment of this application; Figure 12 This is a structural block diagram of a computer device provided according to an embodiment of this application. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0014] Figure 1 This is a structural block diagram of a computer system according to an embodiment of this application. The computer system 100 includes: a first terminal 110, a server 120, and a second terminal 130.

[0015] The first terminal 110 has a client 111 installed and running that supports viewing map data; for example, the client 111 could be a navigation client. When the first terminal runs the client 111, the user interface of the client 111 is displayed on the screen of the first terminal 110. The first terminal 110 is the terminal used by the first user 112.

[0016] The second terminal 130 has a client 131 installed and running that supports viewing map data; this client 131 can be a navigation client. When the second terminal 130 runs the client 131, the user interface of the client 131 is displayed on the screen of the second terminal 130. The second terminal 130 is the terminal used by the second user 132. The user can use the second terminal 130 to perform collaborative navigation with the first terminal 110.

[0017] Optionally, the clients installed on the first terminal 110 and the second terminal 130 are the same, or the clients installed on the two terminals are the same type of client on different operating system platforms. The first terminal 110 can refer to one of a plurality of terminals, and the second terminal 130 can refer to another of a plurality of terminals; this embodiment only uses the first terminal 110 and the second terminal 130 as examples. The device types of the first terminal 110 and the second terminal 130 may be the same or different, and these device types include at least one of: smartphones, tablets, e-book readers, MP3 players, MP4 players, laptops, and desktop computers. The following embodiments use smartphones as examples.

[0018] Those skilled in the art will understand that the number of terminals described above can be more or less. For example, there may be only one terminal, or there may be six, eight, or more terminals. This application does not limit the number of terminals or the type of device.

[0019] Figure 1 Only two terminals are shown in the diagram, but in different embodiments, multiple other terminals 140 can access the server 120. Optionally, one or more terminals 140 may also be terminals corresponding to developers, on which a development and editing platform for a client that supports viewing map data is installed. Developers can edit and update the client on the terminal 140 and transmit the updated client installation package to the server 120 via wired or wireless network. The first terminal 110 and the second terminal 130 can download the client installation package from the server 120 to update the client.

[0020] The first terminal 110, the second terminal 130, and other terminals 140 are connected to the server 120 via a wireless network or a wired network.

[0021] Server 120 includes at least one of a single server, multiple servers, a cloud computing platform, and a virtualization center. Server 120 is used to provide backend services for clients that support viewing map data. Optionally, server 120 undertakes the main computing work, and the terminal undertakes the secondary computing work; or, server 120 undertakes the secondary computing work, and the terminal undertakes the main computing work; or, server 120 and the terminal use a distributed computing architecture for collaborative computing.

[0022] In some embodiments, server 120 includes processor 122, user account database 123, navigation service module 124, and user-facing input / output interface (I / O interface) 125. Processor 122 loads instructions stored in server 120 and processes data in user account database 123 and navigation service module 124. User account database 123 stores user account data used by first terminal 110, second terminal 130, and other terminals 140, such as user account avatars and nicknames. Navigation service module 124 provides background services for navigation functions. User-facing I / O interface 125 establishes communication and exchanges data with first terminal 110 and / or second terminal 130 via wireless or wired network.

[0023] Figure 2 This is a flowchart of a map data update method provided according to an embodiment of this application. The method is performed by a computer device, such as... Figure 1 The server 120 shown executes this method. This method may include the following steps.

[0024] Step 201: Obtain multiple image sequences of the same geographical location through multi-pass acquisition.

[0025] In this embodiment, under the multi-path acquisition mechanism, multiple time-division shooting of the road scene is performed using an onboard camera to obtain multiple image sequences of the same geographical location. This is a fundamental data acquisition step for constructing an order database and realizing automated map data updates. This step, by repeatedly passing through the same road segment, forms an image sequence with temporal information and perspective differences, providing data support for subsequent multi-path fusion, position alignment, and change detection.

[0026] Multi-pass acquisition refers to the repeated acquisition of images of the same road segment or geographical location. Compared with single-pass acquisition, multi-pass acquisition obtains multiple sets of image data of the same road element at different times and with different travel trajectories by passing through the same location multiple times. This acquisition method can effectively overcome the limitations of single-pass acquisition, which results in incomplete image information in a single frame due to factors such as changes in lighting, vehicle obstruction, and equipment noise.

[0027] The same geographical location does not refer to absolutely identical latitude and longitude coordinates, but rather to the same road segment or intersection area that has a corresponding relationship in the road network topology. Optionally, by recording the location information at the time of each acquisition using high-precision positioning equipment (such as a global navigation satellite system receiver, inertial measurement unit, etc.), and combining it with map matching technology, it is possible to associate image sequences acquired from different times with the same geographic spatial range.

[0028] During data acquisition, each acquisition generates a continuous image sequence, with each frame containing a timestamp and spatial location information. Multiple image sequences from various acquisitions together constitute a complete observation dataset for that geographic location. Within this dataset, the same ground vehicle information features (such as ground arrows, lane lines, etc.) are presented under different shooting angles and imaging conditions in different images.

[0029] Step 202: Align the positions of the same ground vehicle information element in different images in multiple image sequences, and generate an ordered database based on the alignment results, which records the status information of the ground vehicle information element at different acquisition time points.

[0030] In this embodiment of the application, based on the completion of multi-stage image sequence acquisition, this step aligns the positions of the same ground vehicle information element in different images in multiple image sequences, and generates an order library based on the alignment results.

[0031] In multi-pass image acquisition, factors such as lateral offset and speed variations in vehicle trajectories result in the same ground vehicle information element exhibiting different positions, angles, and scales in images acquired at different times. Alignment processing aims to eliminate these differences and establish spatial correspondences for the same element in different images. Optionally, a projection alignment method based on the homography transformation matrix can be used for alignment processing.

[0032] Subsequently, based on the alignment results, the identification results of the same ground vehicle information element at different time points are merged to form an order database. The core of the order database is to record the status information of each element at different collection time points, including the element's precise location, lane number, arrow type and other attributes, as well as the element's status changes on the time axis (addition, content change or disappearance).

[0033] Step 203: Write the road element entities in the order library into the map master library to establish a one-to-one correspondence between the road element entities in the order library and the map master library. The road element entities include point elements and relational elements.

[0034] In this embodiment of the application, after the order library is constructed, this step writes the road element entities in the order library into the map master library and establishes a one-to-one correspondence between the element entities in the order library and the map master library.

[0035] The road element entities in the order database are precise results after multi-process fusion, containing the geometric location, attribute information, and status change records of the elements. When writing to the map master database, the system automatically maps each element entity to its corresponding spatial location in the master database. For point elements (such as speed limit signs and speed cameras), they are directly written to the corresponding location of the corresponding Link (road segment) in the master database; for relational elements (such as ground vehicle signals), they are written to the corresponding lane or intersection area in the master database, retaining their attribute information for subsequent semantic transformation. This writing process requires no manual intervention, realizing automated data migration from the order database to the master database.

[0036] Subsequently, a one-to-one correspondence between the feature entities in the order database and the parent map database is established. That is, a bidirectional association mapping is established for each road feature entity: on the one hand, the mapping relationship between the feature entity in the order database and the corresponding location feature entity in the parent database is recorded; on the other hand, the unique identifier of the feature entity in the order database is stored in the parent database, forming a bidirectional index.

[0037] It's important to note that road feature entities are categorized into two types based on their representation in the map data model: point features and relational features. Point features are directly attached to the corresponding positions of links in the master database, their spatial location having a direct geometric dependency on the links. Once written, they can be directly used in navigation applications. Relational features, however, describe the turning relationships between lanes in different directions at intersections and cannot be directly expressed as point features. Therefore, after being written to the master database, relational features undergo a subsequent semantic transformation process. Based on their arrow meanings and road network topology, they are automatically converted into road connectivity relationships (also known as link-link relationships) for navigation route planning.

[0038] Step 204: For relational features written into the map master library, automatically convert them into road segment connectivity relationships for navigation route planning based on the meaning of the arrows and the road network topology.

[0039] In this embodiment of the application, after the relational elements are written into the map master library, this step automatically converts them into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology.

[0040] The semantic essence of relational elements such as ground vehicle information is not an attribute expression of a single location point, but rather a description of the turning connectivity between lanes in different directions at an intersection. For example, a straight arrow indicates that a vehicle entering the intersection from the current road segment can proceed to the opposite road segment; a left-turn arrow indicates that it can proceed to the intersecting road segment on the left. This turning relationship cannot be expressed by simply attaching point elements; it must be modeled based on the topological structure of the road network (road segment-node-road segment) in order to be understood and calculated by the navigation engine.

[0041] Optionally, the automatic conversion process includes: First, based on the spatial location of the relational element in the parent database, the intersection node corresponding to the relational element is located by searching forward through the road network topology. This intersection node is the junction of multiple road segments in the road network and also the anchor point of the turning relationship. Then, the meaning of the arrow of the relational element is analyzed to identify the type of traffic direction it indicates (straight, left turn, right turn, U-turn, straight-left combination, etc.). Finally, based on the topological structure of the intersection node and the arrow direction, all relevant entry and exit road segments matching the turning intention are automatically determined to generate road segment connectivity relationships.

[0042] Optionally, the generated road segment connectivity relationships are stored in the parent database as binary pairs of "entry road segment - exit road segment". Each connectivity relationship represents a valid turning path, which can be directly used for route determination in navigation path planning. For example, when a user plans a route from point A to point B, the navigation engine quickly determines the subsequent road segments that are passable in the current road segment by querying the connectivity relationship set at the intersection node, thus achieving real-time and efficient route generation.

[0043] This application provides a map data update scheme. By constructing an order database, it aligns and records the status of ground vehicle information elements collected from multiple routes, and establishes a one-to-one correspondence between element entities in the order database and the parent map database, achieving automated identification and fusion of relational elements. Simultaneously, for relational elements such as ground vehicle information, it automatically converts them into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology, eliminating the need for manual intervention in topology construction for turning relationships. When changes occur in the real world, the binding mechanism between the order database and the parent map database automatically updates the parent map database by updating the order database, forming a fully automated closed loop from change detection to data update. This scheme significantly reduces manual operation costs and substantially improves the efficiency of map data updates.

[0044] The above Figure 2 The main flow of the map data update method provided in this application embodiment is as follows. The map data update method will be further described below. Figure 3 This is a flowchart of another map data update method provided according to an embodiment of this application. The method is performed by a computer device, such as... Figure 1 The server 120 shown executes this method. This method may include the following steps.

[0045] Step 301: By using the vehicle-mounted camera to repeatedly pass through the same road segment and take photos at different times, images of the road scene ahead are collected, resulting in multiple image sequences with time-series information.

[0046] In this embodiment of the application, this step obtains multiple sets of image data of the same road element at different times and from different perspectives by repeatedly traversing the same geographical area, which fundamentally solves the problem of incomplete information caused by factors such as changes in lighting, vehicle obstruction, and equipment noise in a single acquisition.

[0047] Multi-pass acquisition refers to the repeated acquisition of images of the same road segment or geographical location. Essentially, it involves repeatedly passing through the same location to obtain multiple sets of image data of the same road features under different driving trajectories. Unlike traditional single-pass acquisition, multi-pass acquisition does not rely on a single imaging session but rather compensates for the deficiencies of a single acquisition by using redundant information from multiple observations.

[0048] In engineering practice, multi-process data collection is usually completed by professional data collection vehicles traveling in a cycle along a predetermined route, or by a large number of social vehicles accumulating data naturally during normal driving through crowdsourcing.

[0049] From a data perspective, the multi-path acquisition method constructs a three-dimensional spatiotemporal data cube: spatially, it covers different locations on the same road segment; temporally, it records different acquisition times at the same location; and visually, it presents different observation angles of the same element. This data structure provides rich redundant information for subsequent algorithms, enabling location correction, content fusion, and state change detection through multi-path fusion.

[0050] The following explains how to achieve the function of passing through the same road segment multiple times.

[0051] The first method is the professional data acquisition mode. The data acquisition vehicle travels cyclically along a preset route, covering key road sections multiple times. During each acquisition run, the vehicle maintains a relatively stable speed and acquisition frequency to ensure the continuity of the image sequence and the integrity of spatial coverage. High-precision positioning equipment records the trajectory information for each acquisition, providing a coordinate reference for subsequent spatial alignment.

[0052] The second approach is crowdsourced data collection. This method utilizes trajectory and image data generated by a large number of vehicles during normal driving, and uses spatiotemporal clustering to associate data collected from different vehicles at different times with the same road segment. This approach can achieve wider and higher-frequency road segment coverage, but the data quality varies, requiring multi-path fusion algorithms for quality screening and correction.

[0053] It should be noted that, regardless of the mode used, the core of multi-process data acquisition lies in establishing a set of observation data for the same road segment at different time points, so as to provide a temporally continuous record of the status of elements for the subsequent construction of the order database.

[0054] In some embodiments, for any road segment, each time the road segment is passed, the on-board camera captures images of the road scene containing the same road elements. Due to the difference in the lateral position of the vehicle in the lane, the same road elements in the captured images present different shooting angles at different times of passing.

[0055] The reason is that, during driving, a vehicle's lateral position within the lane is not fixed due to factors such as lane width, driving habits, and obstacle avoidance. The same vehicle may veer to the left or right of the lane on different trips through the same road segment; different vehicles, due to differences in model and driving style, exhibit vastly different trajectories. This natural difference in lateral position causes the onboard camera to change its shooting angle for the same road element: when the vehicle veers to the left, the element appears to the right in the image and is presented from a left-side perspective; when the vehicle veers to the right, the element appears to the left in the image and is presented from a right-side perspective.

[0056] Using different shooting angles offers two main advantages: First, the diversity of perspectives can effectively overcome the occlusion problem under a single perspective. When an element is obscured by a vehicle in front during a certain acquisition, other acquisitions may avoid the obstruction due to different perspectives and obtain complete element information. Second, multi-view observation provides a foundation for 3D information reconstruction. By aligning the positions of multi-view images, the precise 3D position of the element can be deduced, improving geometric accuracy.

[0057] The following is an explanation of the multiple image sequences obtained.

[0058] Each acquired image sequence contains temporal information, including an acquisition timestamp and the sequential relationship between image frames. The timestamp records the precise acquisition time of each frame, used for subsequent temporal analysis; the sequential relationship between frames reflects the continuous changes in the vehicle's spatial position during its movement, used to construct spatial relationships between images.

[0059] Temporal information plays a crucial role in the construction of the ordered database. On the one hand, timestamps allow observations of the same element across different acquisitions to be sorted chronologically, forming a lifecycle record of the element. On the other hand, temporal information supports the tracing of changes in element status, enabling the identification of dynamic events such as additions, content changes, and disappearances. When newly acquired image sequences are compared with historical records in the ordered database, temporal information determines the time baseline for change detection.

[0060] Step 302: Align the positions of the same ground vehicle information feature in different images within multiple image sequences.

[0061] In this embodiment, after completing multi-path image acquisition, it is necessary to align the positions of the same ground vehicle information element in different images from multiple image sequences. That is, to establish the spatial correspondence of the same element in different images, eliminate geometric offsets caused by differences in vehicle trajectories, changes in viewing angles, etc., and provide a spatially consistent data foundation for subsequent multi-path fusion and order library construction.

[0062] In multi-pass image acquisition, the same ground vehicle information element exhibits different geometric shapes in images captured in different sessions. Taking a straight-ahead arrow at an intersection as an example: in the first acquisition, the vehicle is in the center of the lane, and the arrow is centered in the image; in the second acquisition, the vehicle veers to the left of the lane to avoid an obstacle, and the arrow appears to the right with some perspective distortion; in the third acquisition, under backlighting conditions, the arrow's edges are blurred and the contrast is reduced. These differences cause the algorithm to classify even the same element as different targets when directly comparing images, resulting in numerous false positives and false negatives. Alignment processing aims to eliminate these geometric inconsistencies caused by differences in acquisition conditions, enabling the algorithm to accurately determine whether two bounding boxes in different images point to the same physical element.

[0063] In some embodiments, aligning the positions of the same ground vehicle information feature in different images within a multiple image sequence includes: First, select ground feature matching points and determine the homography transformation matrix.

[0064] The alignment process begins by selecting ground feature matching points to determine the homography transformation matrix. The homography transformation matrix is ​​a mathematical tool that describes the projection relationship of the same plane between two images from different viewpoints. Its prerequisite is the existence of a plane in the scene—and feature points on the ground, such as lane lines, markings, and road textures, happen to satisfy this planar assumption.

[0065] For example, see Figure 4 , Figure 4 This is a schematic diagram illustrating the determination of a homography transformation matrix according to an embodiment of this application. Figure 4 As shown, an exemplary example is an image 401 taken from a left-view perspective and an image 402 taken from a right-view perspective. The system extracts ground feature points (such as road texture corners, lane line intersections, etc.) from the two images respectively, and finds corresponding point pairs through a feature matching algorithm, such as... Figure 4 O in L and O RBased on these point pairs, a homography transformation matrix H is determined, which expresses the projection mapping relationship from the first image plane to the second image plane. Taking a real-world scenario as an example, when the data acquisition vehicle passes an intersection for the first time, the straight arrow in the image is located at pixel coordinates (500, 300); the second time it passes, because the vehicle is veering to the right, the same arrow moves to (550, 280) in the image. The homography transformation matrix can establish the correspondence between these two coordinate positions, accurately projecting any point in the first image to the corresponding position in the second image.

[0066] Then, the homography transformation matrix is ​​used to determine the projection box of the ground vehicle information feature identification box in the first image onto the projection box in the second image.

[0067] In this process, after obtaining the homography transformation matrix, the matrix is ​​used to determine the projection boxes of the ground vehicle information feature recognition boxes in the first image onto the second image. Let's assume a straight arrow has been identified in the first image using a target detection algorithm, and its bounding box is a rectangle A. Using the homography transformation matrix, the four vertices of rectangle A are projected onto the second image to obtain projection box A'.

[0068] Finally, when the projection frame hits the ground vehicle information feature recognition box in the second image, it confirms that the ground vehicle information feature is aligned in the first and second images.

[0069] In this algorithm, projection frame A' represents the expected location of an element in the first image in the second image. The algorithm then checks whether projection frame A' hits a ground vehicle signal element recognition box in the second image. If a straight arrow recognition box B also exists in the second image, and the intersection-union ratio (IUU) of projection frame A' and recognition box B is greater than a preset threshold, then it is determined that the projection frame "hit" recognition box B, thus confirming that the element in the first image and the element in the second image are the same ground vehicle signal element.

[0070] For example, at an intersection, there is a combined arrow for both straight and left turns. During the first image capture, the coordinates of the recognition box A are (480, 290) to (520, 310). During the second image capture, due to vehicle trajectory shift, the actual position of the arrow in the image becomes (530, 280) to (570, 300). By projecting the recognition box A onto the second image using a homography transformation matrix, the resulting projection box A' exactly covers the area from (528, 278) to (568, 298), forming a high degree of overlap with the recognition box B. Based on this, the system determines that the two boxes correspond to the same element, thus completing the alignment confirmation.

[0071] By accurately associating the same element from different perspectives using a homography transformation matrix, the mismatch problem caused by perspective differences in single-image matching is overcome, providing accurate correspondences for multi-path fusion. Based on the hit-based mechanism of the projection box, the alignment problem is transformed into a quantifiable geometric calculation, realizing the automation and standardization of the alignment operation without manual intervention. The alignment results record the positional mapping relationship of the same element in different images, providing a spatial benchmark for positional weighting correction in subsequent multi-path fusion, enabling the construction of the order library to output more accurate element positions based on the geometric consistency of multiple observations.

[0072] Step 303: Generate an ordered database based on the alignment results, which records the status information of ground vehicle information elements at different collection time points.

[0073] In this embodiment, based on cross-image position alignment of the same ground vehicle information element in multiple image sequences, an ordered database recording the state information of the ground vehicle information element at different acquisition time points is generated based on the alignment results. This step aggregates the scattered recognition results into complete and accurate element entities through multi-process fusion, and constructs the lifecycle archive of the elements through time-series state records.

[0074] Alignment processing establishes the correspondence of the same element in different images, but the recognition results from each acquisition are still scattered, and there are problems such as positional drift and inconsistent content recognition. To solve this problem, multi-process fusion is performed first, then state changes are identified, and finally the state changes are recorded.

[0075] In some embodiments, generating an ordered database containing state information of ground vehicle information elements at different acquisition time points based on the alignment results includes the following sub-steps.

[0076] Step 303a: Based on the alignment results, the recognition results of the same ground vehicle information element in multiple image sequences at different time points are fused to obtain the fusion result.

[0077] In some embodiments, multi-process fusion includes two parallel technical operations: position-weighted correction and content fusion, which address geometric accuracy and semantic accuracy issues, respectively.

[0078] First, for any ground vehicle information element, the trajectory drift of the ground vehicle information element is corrected by multi-path position weighting.

[0079] Due to factors such as vehicle positioning errors and cumulative errors in inertial navigation, the positions of elements recorded in each data acquisition exhibit varying degrees of drift. Taking a straight arrow at an intersection as an example, the coordinates recorded in the first acquisition deviated from the actual position by approximately 0.2 meters, the second by approximately -0.15 meters, and the third by approximately 0.05 meters. Using only a single acquisition result would introduce a 0.2-meter positioning error. However, by using multi-path position weighted correction, which weights the coordinates from multiple acquisitions according to their confidence levels (e.g., assigning weights based on factors such as image clarity, recognition confidence, and positioning accuracy), random drift can be effectively offset, bringing the final position closer to the true value.

[0080] The mathematical essence of position-weighted correction is to integrate multiple independent observations and use the law of large numbers to reduce random errors. Multi-pass measurement refers to the data collection vehicle traveling the same route multiple times, i.e., collecting data on the same road segment multiple times. Multi-pass position-weighted correction refers to applying position-weighted correction to the same road segment collected multiple times. In practical applications, for road segments collected a large number of times, the corrected position accuracy can reach the centimeter level, far exceeding the accuracy level of a single data collection.

[0081] Then, the lane number and arrow category of the ground vehicle information elements are determined through multi-process content fusion.

[0082] Among them, the content recognition of ground vehicle information elements is easily affected by factors such as occlusion, wear and tear, and lighting. Taking a combination of straight and left-turn arrows at an intersection as an example: In the first data collection, the arrow was clear and complete, and the recognition model output "straight and left-turn" with a 95% confidence level; In the second data collection, the arrow was partially obscured by the vehicle in front, and only the left-turn part was visible, and the recognition result output "left-turn" with a 70% confidence level; In the third data collection, it was backlit, and the arrow outline was blurred, and the recognition result output "straight" with a 60% confidence level.

[0083] Single-shot identification is insufficient to guarantee accuracy, while multi-stage content fusion, by integrating multiple observation results, can output more reliable conclusions. Optionally, a multi-stage voting mechanism can be used: if two out of three acquisitions identify the arrow as containing a left turn component and one identifies it as only going straight, the fusion result is determined as "straight and left turn". Alternatively, a weighted fusion mechanism can be used: weights are assigned based on factors such as image quality and recognition confidence, and the results of multiple identifications are weighted and combined to output the arrow category with the highest overall confidence.

[0084] Step 303b: Compare the fusion result with the historical records of the same geographical location in the order database to identify the status changes of ground vehicle information elements. Status changes include additions, content changes, or disappearances.

[0085] The order repository not only records the current state of elements, but its core function is to record the state changes of elements over time. This function is achieved by comparing the fusion result with historical records.

[0086] Accordingly, after the newly acquired multiple image sequences are aligned and fused, the system compares the fusion result with the historical records of the same geographical location in the order database element by element. The comparison includes attributes such as the existence, location, arrow type, and lane number of the element.

[0087] Optionally, based on the comparison results, three types of state change can be identified.

[0088] The first type is "new". When a certain element exists in the fusion result but not in the historical record, it is determined to be a new element. For example, if a new left-turn arrow is painted at an intersection, and the arrow appears in the multi-path fusion result but is not recorded in the historical record, the system marks the element as "new".

[0089] The second type is content change. When the same element exists in both the fusion result and the historical records, but its attributes have changed, it is judged as a content change. For example, an intersection originally had a combination of straight and left-turn arrows, but after road reconstruction, it was changed to a straight-only arrow. The arrow category in the multi-path fusion result is inconsistent with the historical records. The system marks this element as "content change" and records the specific attributes before and after the change.

[0090] The third type is disappearance. When an element exists in the historical records but not in the fusion result, it is determined to be a disappeared element. For example, if a right-turn arrow at an intersection is erased due to road reconstruction, and the arrow is not detected in the multi-path fusion result, but it exists in the historical records, the system will mark the element as "disappeared".

[0091] Step 303c: Update the status information of the ground vehicle information element. This status information is used to record the status changes of the ground vehicle information element on the time axis.

[0092] After the state change is identified, the order database will update the state information of the ground vehicle information element to form a complete lifecycle profile.

[0093] Optionally, the status information of each ground vehicle information element includes three levels of information: first, the geometric location and attribute information of the element (the optimal result after multi-process fusion); second, timestamp information, which records the time of each observation and the time point when the status change occurs; and third, the status change log, which records every addition, content change or disappearance event experienced by the element.

[0094] For example, taking a straight-ahead arrow at an intersection as an example, its status information can include the following records: the arrow was added when it was first collected in January of XX year; when it was collected in June of XX year, the arrow was found to be worn but the content remained unchanged, and the status remained unchanged; when it was collected in October of XX year, the arrow was found to have been erased, and the status was marked as disappeared. This accumulation of time-series records makes the historical trajectory of each element traceable and verifiable.

[0095] This step, through multi-location weighted correction and content fusion, transforms discrete, error-prone multiple observations into accurate and reliable feature entities, fundamentally improving the quality of map feature data. Through state change identification and recording, a lifecycle archive of features is constructed, transforming map data from a static snapshot into a dynamic database with temporal traceability.

[0096] It should be noted that the above steps for generating the order library only need to be performed once. That is, the order library is constructed as a baseline by collecting multiple image sequences, and then the order library is updated through the following steps.

[0097] For example, see Figure 5 As shown, Figure 5 This is a schematic flowchart illustrating the construction of an order library according to an embodiment of this application. The process includes: 501, ground matching point selection; 502, determining the H-transformation matrix; 503, using the H-transformation matrix for position estimation; 504, target alignment.

[0098] Step 304: Evaluate the image quality of the newly acquired multiple image sequences to obtain the image quality evaluation results of the newly acquired images.

[0099] Before newly acquired image sequences enter the change detection process, they are first evaluated for image quality. This involves obtaining quality indicators such as sharpness, lighting conditions, and occlusion level for each frame, forming an image quality evaluation result. In other words, low-quality images are identified, and existing multi-path historical records in the order database are used as a reference benchmark to assist in differential judgment. This effectively filters out false changes caused by image quality degradation, avoiding misjudging image noise as real changes in road features.

[0100] In actual road data collection, low-quality images are inevitably present in newly acquired image sequences due to various factors such as weather changes, lighting conditions, vehicle obstruction, and equipment noise. For example, a straight arrow at an intersection is clear and complete with a distinct arrow outline when captured at noon on a sunny day; however, when captured on a rainy day, severe road surface reflection blurs the arrow edges; when captured in the evening with backlight, the arrow is covered by strong light, resulting in loss of detail; and during peak hours, the arrow is partially obscured by vehicles ahead. If these low-quality images are directly used for change detection, they are highly prone to misjudgment: a straight arrow that has not changed may be identified as "disappeared" due to blurring, or as having "content changes" due to reflection, generating numerous false changes and severely impacting the accuracy of map data updates.

[0101] Therefore, image quality assessment before change detection, identification of low-quality images, and corresponding processing measures are necessary prerequisites for achieving high-precision automated updates.

[0102] Optionally, image quality evaluation employs a multi-dimensional indicator system to comprehensively assess the recognizability of each frame. The evaluation indicators are as follows: First, sharpness evaluation. The degree of blur in an image is quantified by determining metrics such as high-frequency components, edge gradients, and Laplacian variance. Sharpness metrics significantly decrease when image detail is lost due to motion blur, inaccurate focus, or lens contamination.

[0103] Second, evaluate the lighting conditions. Analyze parameters such as the overall brightness distribution, contrast, and exposure uniformity of the image. Overexposed images lose information in bright areas, while underexposed images lack detail in dark areas, both of which affect the accuracy of feature identification.

[0104] Third, occlusion assessment. Dynamic objects (such as vehicles or pedestrians in front) in the image are identified using semantic segmentation or object detection algorithms, and the proportion of the area obscured by these objects is determined. When an arrow is completely obscured by a large vehicle, the observation of that element in that frame is invalid.

[0105] Fourth, confidence evaluation. The image is input into the target detection model to obtain a confidence score for the recognition results of ground vehicle-to-everything (TTL) features. When the model's confidence score for a certain feature is below a threshold, it indicates that the image quality of that frame is poor or the features themselves are unclear.

[0106] Step 305: If the image quality evaluation result is lower than the preset threshold, the multi-stage historical records of the corresponding geographical location in the order library are used as a reference benchmark to correct the change detection result, so as to filter out false detection results caused by the decline in image quality.

[0107] In this embodiment, when a newly acquired image is determined to be of low quality, the system no longer relies on a single frame image to directly determine the change. Instead, it calls upon the multi-path historical records of the corresponding geographical location in the order database as a reference benchmark, using multi-path information to assist in differential judgment. That is, by utilizing the multi-path historical observation data accumulated in the order database, a "standard knowledge" for a specific geographical location is formed. When a low-quality image cannot independently complete an accurate change judgment, the redundancy and consistency of the multi-path information corrects the misjudgments that may be introduced by a single low-quality image, ensuring the authenticity and reliability of the change detection results. The principle of this mechanism lies in the fact that the order database stores multiple high-quality observation results of the same geographical location at different time points, forming a "standard profile" of the element.

[0108] Optionally, the preset threshold is determined based on historical data statistics and engineering experience. For example, by analyzing a large amount of manually annotated sample data, it can be concluded that the feature recognition accuracy drops significantly when the sharpness is below 0.3 and the illumination uniformity is below 0.4, and corresponding thresholds are set accordingly. The threshold can be dynamically adjusted according to different scenarios: a higher threshold can be set for images collected at noon on a sunny day, while the threshold can be appropriately relaxed for images collected on rainy days or at dusk, in order to balance the relationship between false positives and false negatives.

[0109] When low-quality image-assisted differential processing is triggered, the system retrieves the corresponding geographical location's multi-path historical records from the order database based on the geographical location information of the newly acquired image. Each element entity in the order database records multiple historical observation results, including image features, recognition results, location coordinates, and status change records at different acquisition times.

[0110] For example, taking a combination of straight and left-turn arrows at an intersection as an example, the system's data collection database has accumulated five historical capture records: the first capture was on a sunny day, with the arrow clearly and completely visible; the second capture was on a cloudy day, with the arrow slightly shadowed but still identifiable; the third capture was after rain, with the arrow reflecting light but its outline still discernible; the fourth capture was in the early morning, with soft light and a clear arrow; and the fifth capture was in the evening, with slightly dimmer light but the arrow still visible. These five historical records together constitute the "multi-path standard template" for the arrow, covering the imaging characteristics under different lighting conditions. When a newly captured image is heavily reflected in rain and the arrow is difficult to identify, the system marks that frame as low quality. In the change detection stage, the system does not directly determine whether the arrow has changed. Instead, it compares the features of the new image with the multi-stage historical records in the order database. If the blurry features in the new image basically match the arrow outline in the historical records, and the arrow has not changed recently in the historical records, it is determined that "the image quality is poor, resulting in unclear recognition, and the element has not changed," thus eliminating false changes. If the arrow cannot be detected at all in the new image, but clear arrows have been observed multiple times in the historical records, and although the new image is of low quality, other elements (such as lane lines) can be identified, then it is necessary to combine multiple low-quality images for comprehensive judgment, or trigger a manual review mechanism.

[0111] The following describes how to correct the change detection results of low-quality images based on the multi-process history of the calls.

[0112] The first method is benchmark comparison correction. This involves comparing the features identified in the low-quality image with historical records in the order database item by item. If a feature is not detected in the low-quality image due to blurring, but the feature has consistently existed and remained unchanged in recent observations across multiple historical records, the system determines that the feature "has not disappeared" and corrects the detection result to "exists and has not changed." If a false feature is misidentified in the low-quality image due to reflection, and the corresponding feature does not exist at that location in the historical records, the system determines this as a "false change" and corrects the detection result to "no change."

[0113] The second method is multi-stage voting correction. For elements with low confidence in low-quality images, the system uses a multi-stage voting mechanism for correction. For example, a ground arrow at an intersection might be identified as "straight ahead" in a low-quality image with 60% confidence; however, the three most recent historical observations in the database are "straight ahead, left turn," "straight ahead, left turn," and "straight ahead, left turn," with the voting result being "straight ahead, left turn." Based on this, the system corrects the detection result to "straight ahead, left turn" and determines that the arrow "has not changed its content." Through multi-stage voting, the statistical advantage of historical observations is used to compensate for the lack of information in a single low-quality image.

[0114] The third approach is temporal prediction correction. To address the uncertainty in change detection caused by multiple consecutive low-quality images, the system introduces a temporal prediction mechanism. Utilizing the time-series features of historical records in the order database, the expected state of elements at the current moment is predicted. If the detection result of a newly acquired low-quality image matches the prediction result, the detection result is accepted; otherwise, the prediction result is used for correction. For example, if an element has maintained a stable state with no recorded changes over the past year, its current state is predicted to remain unchanged; if a low-quality image detects "content change," the system classifies it as a false detection and corrects it to "no change."

[0115] After the above correction process, the system outputs the final change detection result. This result, along with the newly acquired data after image quality evaluation and low-quality judgment correction, enters the subsequent order database update and map master database writing process.

[0116] For example, see Figure 6 As shown, Figure 6 This is a schematic diagram of an image comparison provided according to an embodiment of this application. For example... Figure 6 As shown, the system performs content matching and location matching on the newly acquired Query image 601 and multiple Base historical records 602. Then, it performs low-quality judgment auxiliary differential analysis between the Base historical records (dynamic multi-process) and the newly acquired Query image. When the Query image quality is high, direct differential comparison is performed; when the Query image quality is low, high-quality historical records from the Base dynamic multi-process are used to assist in the judgment. By leveraging the redundancy and consistency of multi-process information, misjudgments that may be introduced by a single low-quality image are corrected. Finally, the final output result 603 is obtained.

[0117] Step 306: Perform change detection on the newly acquired multiple image sequences and the historical records in the order database, compare the state information of ground vehicle information elements at the same geographical location at different acquisition time points, and identify the element entities whose state has changed.

[0118] In this embodiment, after image quality evaluation and low-quality image-assisted differential correction, this step performs change detection on the newly acquired multiple image sequences and historical records in the order database. By comparing the state information of ground vehicle information elements at the same geographical location at different acquisition time points, the element entities whose state has changed are accurately identified. That is, through the comparison of element states in the spatiotemporal dimension, the real changes of road elements in the real world are discovered, providing an accurate change list for subsequent map master database writing and semantic transformation, ensuring that map data remains synchronized with the actual situation.

[0119] The basic principle of change detection is to compare the real-world state reflected in the newly acquired data with the historical baseline state stored in the order database element by element, and identify changes through difference analysis. Unlike traditional single-map difference schemes, change detection in this step is carried out after the order database has been built: the order database has stored the multi-path fusion results of the same geographical location at different acquisition time points, forming a "standard knowledge base" for that geographical location. Each ground vehicle information element records precise location coordinates, lane number, arrow type, and complete state change history.

[0120] The core operation of change detection is to compare the state information of the same ground vehicle information element at different collection time points in multiple dimensions. Optionally, the comparison dimensions include the following three dimensions.

[0121] The first dimension is existence comparison. This involves determining whether a certain element exists in the current observation and comparing it to whether that element exists in the historical records of the order database. Existence comparison is the foundation of change detection, used to identify the addition or disappearance of elements. For example, if an intersection originally had no left-turn arrow and there was no corresponding record in the order database; but a left-turn arrow is detected in the current observation, the system determines it as a newly added element. Conversely, if an intersection originally had a right-turn arrow and there was a clear record in the order database; but this arrow is not detected in the current observation and has been verified through multiple rounds to have been erased, the system determines it as a disappeared element.

[0122] The second dimension is set position comparison. For elements whose existence comparison result is "existent," the difference between the currently observed position coordinates and the precise position recorded in the order database is further compared. Geometric position comparison is used to determine whether the element has been displaced or its position has changed. For example, if a straight arrow at an intersection has moved forward by 0.5 meters due to road reconstruction, and the order database records its original position, but the new position is detected in the current observation, the system determines that it is a position change.

[0123] The third dimension is attribute content comparison. For elements with identical existence, the attribute information of the elements is further compared, including lane number and arrow type. Attribute content comparison is used to determine whether the element has undergone semantic changes. For example, if an intersection originally had a straight arrow (Lane 2, Straight), and a combined straight and left-turn arrow (Lane 2, Straight and Left Turn) is detected in the current observation, the system determines that this is a content change.

[0124] Optionally, based on the above multi-dimensional comparison results, the system identifies three types of state changes according to preset rules: The first type of state change is a new addition. A certain element exists in the current observation, but does not exist in the historical records of the order database. Moreover, the element appears stably in multiple consecutive observations. After excluding temporary interference (such as temporary construction signs, vehicle shadows, etc.), it is determined to be a new element.

[0125] For example, in May of a certain city's main road, an intersection underwent renovation, adding a dedicated left-turn lane to the existing straight-ahead lane. The traffic record database only contained straight-ahead arrows in the historical data for this intersection. In June of that year, newly acquired images, after multi-path fusion, detected the newly added left-turn arrow. The system marked this element as "new" and recorded its location coordinates, lane number (lane 1), arrow type (left turn), and other information.

[0126] The second type of state change is a content change. The current observation and the historical records of the order database contain the same element (successful location matching), but its attribute content (lane number or arrow category) changes, and this change occurs consistently across multiple consecutive observations; this is determined to be a content change.

[0127] For example, at a certain intersection, the arrows originally displayed a combination of straight and left-turn arrows. In March of a certain year, the road management department optimized the traffic organization at the intersection, changing the combination arrows to straight-only arrows. The traffic order database recorded this arrow as "straight and left-turn"; after multi-process fusion of newly acquired images, the recognition result was "straight," and the results were consistent across three consecutive observations. The system determined that this element had undergone a "content change," updated the arrow category from "straight and left-turn" to "straight," and recorded the attributes before and after the change, as well as the time point of the change, in the status change log.

[0128] The third type of state change is disappearance. If an element exists in the historical records of the order database, but does not exist in the current observation, and the element has not appeared in multiple consecutive observations, after excluding temporary obstructions (such as temporarily parked vehicles, construction barriers, etc.), it is determined that the element has disappeared.

[0129] For example, at an intersection, there was originally a right-turn arrow. During road renovation in August of a certain year, the right-turn lane was converted into a non-motorized vehicle lane, and the original right-turn arrow was erased. The right-turn arrow was recorded in the order database. After multi-pass fusion of newly acquired images, the arrow was not detected in three consecutive observations, and there were no other temporary obstructions at the same location. The system determined that the element had "disappeared," marked its status as disappeared, and recorded the time of disappearance.

[0130] After completing the status change identification, the system generates a change detection result list, which includes the following information: the unique identifier of the element entity that changed, the element type (point element or relational element), geographical location information, change type (added / content change / disappearance), a description of the status before and after the change (applicable to content changes), and a timestamp of the change. This change detection result list serves as input for subsequent steps.

[0131] For example, taking the change of ground vehicle information elements at a certain intersection as an example, the change detection result can be described as follows: Element ID 01234, type is ground vehicle information, located in lane 2 of the east entrance of intersection A, change type is content change, before the change is "straight and left turn", after the change is "straight", change detection time is XX year Y month Z day, and the corresponding order database update version number is vXXYYZZ.

[0132] Step 307: Write the road element entities that have undergone state changes in the order database into the map master database, and establish a one-to-one correspondence between the road element entities in the order database and the map master database.

[0133] In this embodiment, after completing change detection and identifying road element entities that have undergone state changes, this step writes the road element entities with changed states from the order database into the map master database, and establishes a one-to-one correspondence between the road element entities in the order database and the map master database. Road element entities include point elements and relational elements.

[0134] In some embodiments, road element entities in the order library are written into the map parent library to establish a one-to-one correspondence between road element entities in the order library and the map parent library. This includes: automatically writing the road element entities in the order library after multi-process fusion into the corresponding positions in the map parent library, and recording the association mapping between each road element entity in the order library and the road element entity at the corresponding position in the map parent library to form a bidirectional index.

[0135] Among them, the triggering mechanism for writing to the map master library is closely related to the type of change.

[0136] First, regarding newly added elements: When an element is identified as "new," the system automatically copies the element entity from the order database to the main map database, creating a corresponding element record in the main map database. The write location is determined based on the element's spatial coordinates, and is matched to the corresponding Link or intersection node in the main map database using spatial indexing.

[0137] For example, when a new left-turn arrow is added at an intersection, the system locates the corresponding road segment with Link ID L00123 in the map master database based on its coordinates, and creates a new feature record at the designated lane location of that road segment.

[0138] Secondly, regarding elements with changed content, when an element is determined to have "changed content," the system, based on the pre- and post-change states recorded in the order database, finds the corresponding existing record for that element in the map's parent database and performs an attribute update operation. The updated content includes attribute fields such as lane number, arrow type, and speed limit.

[0139] For example, if the arrow on the ground at an intersection changes from "straight and turn left" to "straight", the system will update the arrow category field of that element in the map master library from "straight and turn left" to "straight" and retain the history of changes.

[0140] Finally, regarding elements determined to be "disappeared," when an element is determined to be "disappeared," the system does not physically delete the element record from the map's master database. Instead, it marks its status as "removed" or logically deleted, while recording the removal time. This soft deletion mechanism preserves the element's historical information, facilitating data traceability and version rollback, while ensuring that the navigation engine does not use invalid elements during route planning.

[0141] It should be noted that road element entities are divided into two categories based on how they are represented in the map data model: point elements and relational elements.

[0142] For point features (such as speed limit signs, speed cameras, hazard signs, traffic lights, and zebra crossings), each point feature has a clearly defined spatial location, and its semantic information is directly attached to that location. When writing to the map master database, the system directly mounts the point features in the order database to the corresponding location of the corresponding Link in the map master database. Taking a speed limit sign as an example, the order database records that the sign is located at station K5+123 of Link L00123. After being written to the map master database, this speed limit information is directly associated with the attribute table of Link L00123, and the navigation engine can directly read the speed limit value of this Link during route planning.

[0143] The semantic essence of relational elements (such as ground vehicle information) is to describe the turning connectivity between lanes in different directions at an intersection, and they cannot be directly attached to a single link as point elements. Therefore, after relational elements are written into the map master database, they are not immediately used for navigation applications, but rather serve as intermediate results in the subsequent semantic transformation process. During the writing process, the system writes the spatial location, lane number, arrow type, and other attributes of the ground vehicle information elements into the data structure of the corresponding intersection node or lane group in the map master database, preserving complete geometric and attribute information to provide a data foundation for subsequent link-link transformations.

[0144] The following explains how to establish correspondences and bidirectional indexes.

[0145] First, an association mapping is created. When a feature entity is written from the order database to the map master database, the system automatically records the mapping relationship between the unique identifier of that feature in the order database and the unique identifier of the corresponding feature in the map master database. This mapping relationship is stored in a separate binding table, which includes fields such as the order database feature ID, the map master database feature ID, the binding time, and the binding status.

[0146] For example, taking a straight arrow on the ground as an example, the feature ID in the order library is ORD_12345. After being written into the map master library, the generated map master library feature ID is MDB_67890. The mapping relationship is recorded in the binding table and the status is marked as "bound".

[0147] Next, a bidirectional index is constructed. The system maintains index fields pointing to each other in both the order database and the parent map database. In the feature records of the order database, a "Parent Map Database Corresponding ID" field is added, storing the identifier of the record corresponding to that feature in the parent map database; in the feature records of the parent map database, an "Order Database Source ID" field is added, storing the identifier of the order database feature from which that feature originated. This bidirectional indexing mechanism allows for bidirectional queries of the relationship between the two databases: starting from the order database, the corresponding feature in the parent map database can be quickly located; starting from the parent map database, the historical records and status changes of that feature in the order database can be traced.

[0148] Finally, the above binding relationships are maintained and synchronized. When the status of a feature in the order repository changes, the system automatically locates the corresponding feature in the parent map repository through bidirectional indexing and performs a synchronization update.

[0149] For example, if a ground arrow is marked as "content changed" in the order database, the system will find the corresponding record in the map parent database through the "map parent database corresponding ID" field and automatically perform attribute update; after the update is completed, the "order database source ID" field in the map parent database will keep pointing to the updated record in the order database to ensure that the two databases are consistent.

[0150] For example, see Figure 7 , Figure 7 This is a schematic diagram illustrating a fully automated writing of an order library to a parent library according to an embodiment of this application. For example... Figure 7 As shown, Figure 7 (a) in the example shows a schematic diagram after the establishment of the order library. Figure 7 (a) includes point element 701, relational element 702, and element status page 703 displayed after selecting any element. This element status page 703 is used to display the current status and historical change records of the element. Figure 7 (b) in the example shows a schematic diagram after writing to the parent library. Figure 7 (b) includes point element 704 and relational element 705.

[0151] Step 308: For relational features written into the map master library, automatically convert them into road segment connectivity relationships for navigation route planning based on the meaning of the arrows and the road network topology.

[0152] In this embodiment, after the changed elements are written into the master database and the two databases are bound together, the relational elements written into the master map database are automatically converted into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology. That is, relational elements such as ground vehicle information are transformed from static geometric location and attribute information into dynamic, computable topological relationships, enabling them to directly serve intelligent navigation and route planning applications.

[0153] The semantic essence of relational elements such as ground vehicle information is not an attribute expression of a single location point, but rather a description of the turning connectivity between lanes in different directions at an intersection. Taking a standard intersection as an example, the ground arrows on an entry link actually specify which exit links a vehicle can access after entering the intersection from that entry link. A straight arrow indicates that a vehicle can access the oncoming road segment; a left-turn arrow indicates that a vehicle can access the intersecting road segment on the left; a right-turn arrow indicates that a vehicle can access the intersecting road segment on the right; and a U-turn arrow indicates that a vehicle can access the reverse lane of the oncoming road segment.

[0154] This kind of turning relationship cannot be expressed by simply attaching point features. If the ground arrow is directly attached as a point feature to the entry link, the navigation engine can only know "there is an arrow at this location" during route planning, but cannot know "which links can be reached after entering the intersection from this link". Only by transforming the meaning of the arrow into the connectivity relationship between links can the navigation engine, when determining the route, query the set of subsequent links that are passable based on the current link at the intersection node, thereby generating a complete navigation path.

[0155] In some embodiments, an automatic conversion method based on topology lookup can be used. This involves finding the corresponding intersection through the forward topological road network, and then automatically determining all relevant entry and exit links based on the arrow meanings to generate road segment connectivity relationships. Correspondingly, for relational features written into the map master database, based on the arrow meanings and road network topology of the relational features, they are automatically converted into road segment connectivity relationships for navigation route planning, including: First, locate the intersection. For relational features written into the map master database, find the corresponding intersection by forward-looking the road network based on the location of the relational feature. That is, locate the intersection node corresponding to the feature. An intersection node is the intersection point of multiple road segments in the road network and is the anchor point for turning relationships.

[0156] For example, if the ground arrow is located at the end of Link L00123, the system searches forward along the Link direction and finds the node N1001 connected to it. This node is the intersection corresponding to the arrow.

[0157] Then, based on the meaning of each arrow in the relational element, all relevant entry and exit links of the intersection are automatically determined, generating the road segment connectivity relationship.

[0158] First, identify the entry link. The entry link is the road segment where the vehicle is approaching the intersection. For the ground arrow, the link it is located on is the entry link. In the example above, the entry link is L00123.

[0159] Next, determine the exit link. Based on the arrow meaning, filter out the exit links that meet the turning requirements from all outgoing links at intersection node N1001.

[0160] Optionally, the rules are determined as follows: Straight arrow: Filters out links that are in the same direction as the entry link (angle less than 30 degrees); Left arrow: Filters out links that are at a left-hand angle (usually 60-120 degrees) to the direction of entry; Right arrow: Filters out links that are at a right angle (usually 240-300 degrees) from the direction of entry to exit links; U-turn arrow: Filters out exit links that are in the opposite direction to the entry link (approximately 180 degrees apart); Combined arrows (e.g., straight ahead and left turn): Filter the set of exit links that simultaneously meet the conditions of going straight ahead and turning left.

[0161] Finally, the above connectivity relationships are generated. Each "Enter Link - Exit Link" combination is recorded as a road segment connectivity relationship.

[0162] For example, entering Link L00123 and exiting Link L00124 (straight direction) creates one connection, and exiting Link L00125 (left turn direction) creates another connection.

[0163] In some embodiments, a mapping library-based approach can be used for conversion. Accordingly, for relational features written into the master map library, based on the arrow meanings and road network topology of the relational features, they are automatically converted into road segment connectivity relationships for navigation route planning, including: First, perform arrow mapping. Establish a pre-defined arrow-steering mapping library, mapping common arrow categories to standard steering types.

[0164] Optionally, the turning type of the straight arrow is STRAIGHT; the turning type of the left turn arrow is LEFT; the turning type of the right turn arrow is RIGHT; the turning type of the U-turn arrow is U_TURN; the turning type of the straight-and-left-turn combination is STRAIGHT_LEFT; the turning type of the straight-and-right-turn combination is STRAIGHT_RIGHT; and the turning type of the left-and-right-turn combination is LEFT_RIGHT.

[0165] Next, the steering type is matched. Based on the meaning of the arrows in the relational elements, the corresponding steering type is matched from the preset arrow-steering map library.

[0166] For example, if a ground arrow is identified as "go straight and turn left", the system matches the steering type as STRAIGHT_LEFT from the mapping library.

[0167] Then, based on the location of relational elements and the road network topology, the entry and exit links corresponding to the turning type are determined, and the road segment connectivity is generated.

[0168] First, determine the exit links. Based on the turning type and road network topology, identify all exit links that match that turning type. For the STRAIGHT_LEFT type, the system filters exit links for both straight and left turns; for the LEFT_RIGHT type, the system filters exit links for both left and right turns.

[0169] In generating connectivity relationships, each entering link is combined with a defined exit link to generate a set of road segment connectivity relationships.

[0170] It should be noted that the two methods described above can complement each other. The topology-based approach is suitable for scenarios with clear intersection topologies and well-defined turning rules; the mapping library-based approach is suitable for scenarios requiring complex combinations of arrows and rapid response. In practical applications, the appropriate method can be selected based on the specific scenario, or a combination of both methods can be used to improve conversion accuracy.

[0171] It should be noted that the generated connectivity relationships are stored in the parent database in the form of "Enter Link - Exit Link" tuples, forming the core components of the road network topology. Each connectivity relationship represents a legal turning path and comes with relevant attribute information, including turning type, turning angle, lane number, and traffic restrictions (such as time restrictions and vehicle type restrictions).

[0172] For example, taking a standard intersection as an example, if the ground arrow indicates "straight or left turn", the generated connectivity may include: Connectivity 1, entering Link L10023 and exiting Link L10024 (straight direction), with a turn type of STRAIGHT; Connectivity 2, entering Link L10023 and exiting Link L10025 (left turn direction), with a turn type of LEFT.

[0173] See Figure 8 As shown, Figure 8 This is a schematic diagram illustrating the conversion of entity data from the parent database to semantic data, according to an embodiment of this application. The map parent database constructs a road network with Links as the basic unit to represent a database storing the locations and attributes of map road features in the real world. It includes point features 801 (e.g., speed limit signs) and relational features 802 (e.g., ground vehicle signals). Point features are directly applied to the corresponding positions in the parent database links. For the meaning of each entity arrow, all relevant entry links 803 and exit links 804 of the intersection are automatically calculated, achieving fully automatic conversion of parent database entities to parent database semantics.

[0174] It should be noted that, to make the map data update scheme provided in this application easier to understand, please refer to [link / reference needed]. Figure 9 , Figure 9 This is a flowchart illustrating a map data update scheme provided according to an embodiment of this application. Figure 9 As shown, the process includes the following steps: 901. Image acquisition. 902. Automated road data update algorithm. 902 includes: 9021. Building a multi-path image database for the order database; 9022. Automatically writing entities from the order database to the parent database; 9023. Automatically converting parent database entities to parent database semantics; 903. Automated data update by binding the two databases.

[0175] See Figure 10 , Figure 10This is an overall architecture diagram of a map data update scheme provided according to an embodiment of this application. For example... Figure 10 As shown, the core process includes the following: Starting with multi-path image database construction 1001. An order database 1002 is constructed through multi-path acquisition and fusion, forming accurate road element entities. Starting with the quality of the parent database 1003, the parent database entity 1004 serves as the storage carrier, receiving the element information output from the order database. Automatic semantic conversion of relational elements (e.g., ground vehicle information) automatically converts relational elements such as ground vehicle information into road segment connectivity relationships for navigation route planning, i.e., converting them into parent database semantics 1005. Automatic lifecycle writing records the addition, content changes, and disappearance of elements on the timeline to the order database, constructing an element lifecycle archive. A one-to-one correspondence between element entities in the order database and the parent database is established through two-database binding 1006, forming a bidirectional indexing mechanism to achieve updates of relationships and timeliness. Finally, automated map data updates are achieved 1007. Lifecycle updates ensure that the status changes of each element are traceable and verifiable, providing a reliable guarantee for the continuous freshness of map data updates.

[0176] This application provides a map data update scheme. By constructing an order database, it aligns and records the status of ground vehicle information elements collected from multiple routes, and establishes a one-to-one correspondence between element entities in the order database and the parent map database, achieving automated identification and fusion of relational elements. Simultaneously, for relational elements such as ground vehicle information, it automatically converts them into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology, eliminating the need for manual intervention in topology construction for turning relationships. When changes occur in the real world, the binding mechanism between the order database and the parent map database automatically updates the parent map database by updating the order database, forming a fully automated closed loop from change detection to data update. This scheme significantly reduces manual operation costs and substantially improves the efficiency of map data updates.

[0177] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0178] Please refer to Figure 11 , Figure 11 This is a block diagram of a map data updating device according to an embodiment of this application. The device includes: an acquisition module 1101, a generation module 1102, a relationship module 1103, and a conversion module 1104.

[0179] The acquisition module 1101 is used to acquire multiple image sequences of the same geographical location through a multi-pass acquisition method; The generation module 1102 is used to align the positions of the same ground vehicle signal element in different images in multiple image sequences, and generate an ordered library that records the status information of the ground vehicle signal element at different acquisition time points based on the alignment results. Relationship module 1103 is used to write road element entities from the order library into the map master library and establish a one-to-one correspondence between road element entities between the order library and the map master library. Road element entities include point elements and relational elements. The conversion module 1104 is used to automatically convert relational features written into the map master library into road segment connectivity relationships for navigation route planning, based on the meaning of the arrows and the road network topology of the relational features.

[0180] In some embodiments, the generation module 1102 is used to select ground feature matching points and determine a homography transformation matrix; use the homography transformation matrix to determine the projection box of the ground vehicle signal element recognition box in the first image onto the second image; when the projection box hits the ground vehicle signal element recognition box in the second image, confirm that the ground vehicle signal element is aligned in the first image and the second image.

[0181] In some embodiments, the generation module 1102 is used to fuse the recognition results of the same ground vehicle signal element at different time points in multiple image sequences based on the alignment result to obtain a fusion result; compare the fusion result with the historical records of the same geographical location in the order database to identify the state changes of the ground vehicle signal element, including addition, content change or disappearance; update the state information of the ground vehicle signal element, the state information is used to record the state changes of the ground vehicle signal element on the time axis.

[0182] In some embodiments, the generation module 1102 is used to correct the trajectory drift of any ground vehicle information element by multi-path position weighting; and to determine the lane number and arrow category of the ground vehicle information element by multi-path content fusion.

[0183] In some embodiments, the relationship module 1103 is used to automatically write the road element entities in the order library after multi-process fusion into the corresponding positions in the parent library, and record the association mapping between each road element entity in the order library and the road element entity at the corresponding position in the parent library to form a bidirectional index.

[0184] In some embodiments, the conversion module 1104 is used to find the corresponding intersection in the forward topological road network based on the location of the relational features after they are written into the map master library; and to automatically determine all relevant entry and exit links of the intersection based on the meaning of each arrow in the relational features, thereby generating road segment connectivity relationships.

[0185] In some embodiments, the conversion module 1104 is used to match the corresponding turning type from a preset arrow-turning mapping library according to the meaning of the arrow of the relational element; and to determine the entry link and exit link corresponding to the turning type according to the location of the relational element and the road network topology, thereby generating the road segment connectivity relationship.

[0186] In some embodiments, the acquisition module 1101 is used to capture images of the road scene ahead by repeatedly passing through the same road segment and taking photos at different times using an in-vehicle camera device, thereby obtaining a sequence of multiple images with time information.

[0187] In some embodiments, the acquisition module 1101 is used to acquire images of a road scene containing the same road elements each time the vehicle passes through any road segment, and the same road element in the acquired images presents different shooting angles due to the difference in the lateral position of the vehicle in the lane at different times of passing through.

[0188] In some embodiments, the generation module 1102 is used to perform image quality evaluation on the newly acquired multiple image sequences to obtain the image quality evaluation result of the newly acquired image; if the image quality evaluation result is lower than a preset threshold, the multi-stage historical record of the corresponding geographical location in the order library is called as a reference benchmark to correct the change detection result, so as to filter out false detection results caused by image quality degradation.

[0189] In some embodiments, the generation module 1102 is used to perform change detection on the newly acquired multiple image sequences and the historical records in the order database, compare the state information of ground vehicle information elements at the same geographical location at different acquisition time points, and identify the element entities whose state has changed.

[0190] This application provides a map data updating device. By constructing an order database, it aligns and records the status of ground vehicle information elements collected from multiple routes, and establishes a one-to-one correspondence between the order database and the main map database, achieving automated identification and fusion of relational elements. Simultaneously, for relational elements such as ground vehicle information, it automatically converts them into road segment connectivity relationships for navigation route planning based on their arrow meanings and road network topology, eliminating the need for manual intervention in topology construction for turning relationships. When changes occur in the real world, the binding mechanism between the order database and the main map database allows for automatic synchronous updates to the main map database by updating the order database, forming a fully automated closed loop from change detection to data update. This solution significantly reduces manual operation costs and substantially improves the efficiency of map data updates.

[0191] It should be noted that the apparatus provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the content structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0192] Please refer to Figure 12 , Figure 12 This is a structural block diagram of a computer device according to an embodiment of this application. The computer device can be implemented as described above. Figure 1 The terminal or server in the system. For example... Figure 12 As shown, computer device 1200 includes a central processing unit (CPU) 1201, a system memory 1204 including random access memory (RAM) 1202 and read-only memory (ROM) 1203, and a system bus 1205 connecting the system memory 1204 and the CPU 1201. Computer device 1200 also includes a basic input / output system (I / O system) 1206 that facilitates information transfer between various devices within the computer, and a mass storage device 1207 for storing the operating system 1213, application programs 1214, and other program modules 1215.

[0193] The basic input / output system 1206 includes a display 1208 for displaying information and an input device 1209 for user input, such as a mouse or keyboard. Both the display 1208 and the input device 1209 are connected to the central processing unit 1201 via an input / output controller 1210 connected to the system bus 1205. The basic input / output system 1206 may also include the input / output controller 1210 for receiving and processing input from multiple other devices such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controller 1210 also provides output to a display screen, printer, or other types of output devices.

[0194] Mass storage device 1207 is connected to central processing unit 1201 via a mass storage controller (not shown) connected to system bus 1205. Mass storage device 1207 and its associated computer-readable media provide non-volatile storage for computer device 1200. That is, mass storage device 1207 may include computer-readable media (not shown) such as hard disk or CD-ROM (Compact Disc Read-Only Memory) drive.

[0195] Without loss of generality, computer-readable media can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented using any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media includes RAM (Random Access Memory), ROM (Read-Only Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory or other solid-state storage technologies, CD-ROM, DVD (Digital Video Disc), or other optical storage, magnetic tape cassettes, magnetic tape, disk storage, or other magnetic storage devices. Of course, those skilled in the art will recognize that computer storage media are not limited to the above-mentioned types. The system memory 1204 and mass storage device 1207 described above can be collectively referred to as memory.

[0196] Computer device 1200 can be connected to the Internet or other network devices via network interface unit 1211 connected to system bus 1205.

[0197] The memory also includes one or more programs / instructions, which are stored in the memory. The central processing unit 1201 executes the one or more programs / instructions to implement all or part of the steps in the methods shown in the above embodiments of this application.

[0198] In an exemplary embodiment, a computer-readable storage medium is also provided, storing a computer program that, when executed by a processor, implements the aforementioned map data update method. Optionally, the computer-readable storage medium may include ROM (Read-Only Memory), RAM (Random Access Memory), SSD (Solid State Drives), or optical disc, etc. The random access memory may include ReRAM (Resistance Random Access Memory) and DRAM (Dynamic Random Access Memory).

[0199] In an exemplary embodiment, a computer program product is also provided, the computer program product including a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, causing the computer device to perform the map data update method described above.

[0200] It should be noted that the collection and processing of relevant data (such as image sequences) in this application should strictly comply with the requirements of relevant national laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0201] It should be understood that "multiple" as used herein refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. Furthermore, the step numbers described herein are merely illustrative of one possible execution order. In some other embodiments, the steps may not be executed in numerical order, such as two steps with different numbers being executed simultaneously, or two steps with different numbers being executed in the reverse order of the illustration. This application does not limit this.

[0202] The above description is merely an exemplary embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A map data update method, characterized in that, The method includes: Multiple image sequences of the same geographical location are acquired through multi-pass acquisition. The positions of the same ground vehicle signal element in different images in the multiple image sequences are aligned. Based on the alignment results, an ordered database is generated that records the state information of the ground vehicle signal element at different acquisition time points. The state information of the ground vehicle signal element includes: the geometric position and attribute information of the element, timestamp information, and state change log. The timestamp information is used to record the time of each observation and the time point when the state changes. The state change log is used to record every addition, content change, or disappearance event experienced by the element. The road element entities in the order database are written into the map master database, establishing a one-to-one correspondence between the road element entities in the order database and the map master database. This one-to-one correspondence is achieved by recording the association mapping between each road element entity in the order database and the corresponding road element entity in the map master database, forming a bidirectional index. The road element entities include point elements and relational elements, and the relational elements include the ground vehicle information elements. When writing the relational elements, the spatial location, lane number, and arrow category of the relational elements are written into the data structure of the corresponding intersection node or lane group in the map master database to retain complete geometric and attribute information. For relational features written into the map master library, the corresponding turning type is matched from the preset arrow-turning mapping library according to the meaning of the arrows in the relational features; Based on the location of the relational elements and the road network topology, the entry and exit links corresponding to the turning type are determined, and road segment connectivity is generated. Each set of connectivity represents a legal turning path and is accompanied by relevant attribute information, including turning type, turning angle, lane number and traffic restrictions. When the status information of road element entities in the order database changes, the corresponding element in the map parent database is automatically located through the bidirectional index, and a synchronous update is performed. When querying any road feature entity in the map master database, the historical records and status changes of the corresponding road feature entity in the order database are traced through the bidirectional index.

2. The method according to claim 1, characterized in that, The alignment process for the positions of the same ground vehicle information feature in different images within the multiple image sequences includes: Select ground feature matching points and determine the homography transformation matrix; The homography transformation matrix is ​​used to determine the projection box of the ground vehicle information feature identification box in the first image onto the projection box in the second image; When the projection frame hits the ground vehicle information feature recognition frame in the second image, it confirms that the ground vehicle information feature is aligned in the first image and the second image.

3. The method according to claim 1 or 2, characterized in that, The ordered database generated based on the alignment results, which records the status information of the ground vehicle information elements at different acquisition time points, includes: Based on the alignment result, the recognition results of the same ground vehicle information element in the multiple image sequences at different time points are fused to obtain the fusion result; The fusion result is compared with the historical records of the same geographical location in the order database to identify the state changes of the ground vehicle information elements, including additions, content changes, or disappearances. Update the status information of the ground vehicle information element, which is used to record the status changes of the ground vehicle information element on the time axis.

4. The method according to claim 3, characterized in that, The step of fusing the recognition results of the same ground vehicle information element at different time points in the multiple image sequences based on the alignment result includes: For any ground vehicle information element, the trajectory drift of the ground vehicle information element is corrected by multi-path position weighting; The lane number and arrow category of the ground vehicle information elements are determined by multi-process content fusion.

5. The method according to claim 1, characterized in that, The method of acquiring multiple image sequences of the same geographical location through multi-pass acquisition includes: By repeatedly passing through the same road segment and taking photos at different times using an in-vehicle camera, images of the road scene ahead are collected, resulting in a sequence of multiple images with time-series information.

6. The method according to claim 5, characterized in that, The process of capturing images of the road scene ahead by repeatedly passing through the same road segment and taking photos at different times using an in-vehicle camera includes: For any road segment, each time the vehicle passes through the road segment, the on-board camera captures images of the road scene containing the same road elements. Due to the difference in the lateral position of the vehicle in the lane, the same road elements in the captured images present different shooting angles at different times of passing through.

7. The method according to claim 1, characterized in that, The method further includes: The newly acquired image sequences are subjected to image quality evaluation to obtain the image quality evaluation results of the newly acquired images; If the image quality evaluation result is lower than a preset threshold, the multi-stage historical records of the corresponding geographical location in the order database are used as a reference benchmark to correct the change detection result.

8. The method according to claim 1, characterized in that, The method further includes: The newly acquired multiple image sequences are compared with the historical records in the order database to detect changes. The state information of ground vehicle information elements at the same geographical location at different acquisition time points is compared to identify the element entities whose state has changed.

9. A map data updating device, characterized in that, The device includes: The acquisition module is used to acquire multiple image sequences of the same geographical location through multi-pass acquisition. The generation module is used to align the positions of the same ground vehicle signal element in different images in the multiple image sequences, and generate an ordered library based on the alignment results, which records the state information of the ground vehicle signal element at different acquisition time points. The state information of the ground vehicle signal element includes: the geometric position and attribute information of the element, timestamp information, and state change log. The timestamp information is used to record the time of each observation and the time point when the state changes. The state change log is used to record every addition, content change, or disappearance event experienced by the element. The relationship module is used to write road element entities from the order database into the map master database, establishing a one-to-one correspondence between the road element entities in the order database and the map master database. This one-to-one correspondence is achieved by recording the association mapping between each road element entity in the order database and the corresponding road element entity in the map master database, forming a bidirectional index. The road element entities include point elements and relational elements, and the relational elements include the ground vehicle information elements. When writing the relational elements, the spatial location, lane number, and arrow type of the relational elements are written into the data structure of the corresponding intersection node or lane group in the map master database to retain complete geometric and attribute information. The conversion module is used to match the corresponding turning type from a preset arrow-turning mapping library according to the meaning of the arrow of the relational element after it is written into the map master library; and to determine the entry link and exit link corresponding to the turning type according to the location of the relational element and the road network topology, and generate road segment connectivity. Each set of connectivity represents a legal turning path and is accompanied by relevant attribute information, including turning type, turning angle, lane number and traffic restrictions. The relationship module is also used to automatically locate the corresponding element in the map parent library through the bidirectional index and perform synchronous updates when the status information of the road element entity in the order library changes; when querying any road element entity in the map parent library, the historical records and status changes of the corresponding road element entity in the order library are traced through the bidirectional index.

10. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing a computer program that is loaded and executed by the processor to implement the map data update method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which is loaded and executed by a processor to implement the map data update method as described in any one of claims 1 to 8.

12. A computer program product, characterized in that, The computer program product includes a computer program executed by a processor to implement the map data update method as described in any one of claims 1 to 8.

Images