High-precision map data processing method and device

Abstract

The invention discloses a high-precision map data processing method and device, and the method comprises the steps: carrying out the standardization of geographic information data through an adaptive standardization engine based on space-time semantic perception, and obtaining the standardized geographic information data; performing transverse classification and vertical classification on the standardized geographic information data to obtain a hierarchical data set; performing data encryption and desensitization on the hierarchical data set to obtain an encrypted and desensitized data set; according to the encrypted and desensitized data set, performing spatial range expansion auditing, relative height requirement auditing and auxiliary positioning element compliance expression auditing on the high-precision map to obtain a map auditing result; and under the condition that the image auditing result is that auditing is passed, optimizing the high-precision map data according to the encrypted and desensitized data set, and issuing the optimized high-precision map data to a map service platform. According to the method, the map checking efficiency and accuracy can be improved, and meanwhile, the integrity and safety of high-precision map data are effectively guaranteed.

Description

Technical Field

[0001] This application relates to the field of high-precision map review technology, and in particular to high-precision map data processing methods and apparatus. Background Technology

[0002] High-precision maps are primarily used in Level 4 and above advanced autonomous driving systems, where the driving force is the autonomous driving algorithm, not the human driver. These algorithms require safety redundancy. Therefore, compared to traditional navigation electronic maps and advanced driver assistance systems (ADAS) maps, the algorithms place higher demands on the spatial representation range of high-precision maps, various forms of relative altitude information, and auxiliary positioning elements, requiring a higher level of safety redundancy. These high-level safety redundancy requirements must be considered during map review. Traditional map review methods often focus on basic information such as geometric accuracy and road network connectivity. However, for high-precision maps, this information is insufficient to meet the high safety redundancy requirements of autonomous driving algorithms, resulting in low efficiency, poor accuracy, and difficulty in ensuring the integrity and security of high-precision map data. Summary of the Invention

[0003] The main objective of this application is to provide a high-precision map data processing method and apparatus, which aims to solve the technical problems of low efficiency, poor accuracy, and difficulty in ensuring the integrity and security of high-precision map data in traditional map review methods.

[0004] To achieve the above objectives, this application proposes a high-precision map data processing method, which includes: Acquire geographic information data for high-precision maps; An adaptive standardization engine based on spatiotemporal semantic awareness standardizes the geographic information data to obtain standardized geographic information data. The standardized geographic information data is horizontally classified based on an automatic clustering algorithm using multimodal feature extraction and vertically graded based on a knowledge graph reasoning algorithm using influence assessment to obtain a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data sets of different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data. The hierarchical data set is encrypted and desensitized to obtain the encrypted and desensitized data set. Based on the encrypted and desensitized data set, the high-precision map is reviewed for spatial range expansion, relative height requirements, and compliance of auxiliary positioning elements, resulting in the map review results. If the high-precision map is approved in the review, the high-precision map data is optimized based on the encrypted and desensitized data set, and the optimized high-precision map data is published to the map service platform.

[0005] In one embodiment, the spatiotemporal semantic awareness-based adaptive standardization engine standardizes the geographic information data to obtain standardized geographic information data, including: A unified spatiotemporal reference framework is constructed based on the global dynamic reference frame, the international Earth reference frame, and regional crustal deformation models. Based on the unified spatiotemporal reference framework, the geographic information data is spatiotemporally and semantically aligned to obtain spatiotemporally aligned data. The spatiotemporally aligned data is transformed using a coordinate system transformation algorithm to obtain the coordinate-transformed data. Construct a geographic element ontology knowledge base, and determine the entity standard attribute structure based on the geographic element ontology knowledge base; The data after coordinate system transformation is normalized according to the entity standard attribute structure to obtain the data after attribute semantic normalization. The data after the attribute semantics are normalized is encoded according to a preset hybrid encoding format to obtain standardized geographic information data.

[0006] In one embodiment, the standardized geographic information data is horizontally classified using an automatic clustering algorithm based on multimodal feature extraction and vertically hierarchically classified using a knowledge graph reasoning algorithm based on influence assessment to obtain a hierarchical data set, including: Feature extraction is performed on the standardized geographic information data based on a multimodal embedding model to obtain feature vectors for various types of data; The feature vectors of the various types of data are mapped to a unified semantic embedding space to obtain a unified semantic embedding vector; The unified semantic embedding vectors are clustered based on the HDBSCAN clustering algorithm to obtain data sets of different categories. Construct a high-precision map risk knowledge graph, wherein the nodes in the high-precision map risk knowledge graph include geographical elements, functional impacts, and safety levels; Based on the high-precision map risk knowledge graph, the impact assessment algorithm is used to assess the impact of data in the different categories of datasets to obtain decision influence scores. Based on the decision influence score, a knowledge graph reasoning path is constructed, and the knowledge graph reasoning path is used to vertically classify the different categories of data sets to obtain a hierarchical data set.

[0007] In one embodiment, the step of encrypting and desensitizing the hierarchical data set to obtain an encrypted and desensitized data set includes: The hierarchical data set is desensitized using a classification desensitization strategy to obtain different categories of data sets after desensitization. The classification desensitization strategies include a semantic segmentation-guided local fuzzy desensitization strategy for image data sets, an attribute differential privacy desensitization strategy for point cloud data sets, a spatiotemporal randomization desensitization strategy for trajectory data sets, a noise addition desensitization strategy for inertial navigation data sets, and a hash obfuscation desensitization strategy for mapping data sets. The data at different levels in the different categories of the de-identified data sets are encrypted using a hierarchical encryption strategy to obtain hierarchically encrypted data sets.

[0008] In one embodiment, the step of reviewing the high-precision map based on the encrypted and desensitized data set, including spatial range expansion review, relative height requirement review, and compliance expression review of auxiliary positioning elements, to obtain the high-precision map review result, includes: Based on the encrypted and desensitized data set, image data, point cloud data, trajectory data, inertial navigation data, and composition data are determined; Based on the image data, point cloud data, and inertial navigation data, a spatial range expansion review is performed to obtain the extended range around the road. Based on the point cloud data, trajectory data, and composition data, the relative height requirement is reviewed to obtain the relative height classification information of map elements; Based on the aforementioned compositional data, the compliance review of auxiliary positioning elements is conducted to obtain the relative height classification information of the auxiliary positioning elements; The high-precision map review result is obtained by comprehensively evaluating the surrounding extension range of the road, the relative height classification information of map elements, and the relative height classification information of auxiliary positioning elements.

[0009] In one embodiment, the step of performing spatial range expansion verification based on the image data, the point cloud data, and the inertial navigation data to obtain the road perimeter expansion range includes: Road images are extracted based on the image data, and the road images are subjected to inverse perspective transformation based on vanishing points to obtain corrected road images. The point cloud data is preprocessed using a radius filtering algorithm, and road surface point cloud data is extracted based on the preprocessed point cloud data. Road traffic markings are extracted based on the corrected road image and the road surface point cloud data; Vehicle speed data is extracted based on the inertial navigation data, and a speed-reaction time-environment regression model is constructed based on the vehicle speed data and the corresponding safety distance requirements. The reference distance matrix for the outward expansion of the road perimeter is determined based on the speed-reaction time-environment regression model. The extent of the road perimeter is determined based on the road traffic markings and the reference distance matrix extending outward from the road perimeter.

[0010] In one embodiment, the step of performing relative height requirement review based on the point cloud data, trajectory data, and composition data to obtain relative height classification information for map elements includes: A 3D scene model is constructed based on point cloud data, and the 3D scene model is then sliced ​​at different heights to obtain slice layers at different heights. The movement trajectory of the moving object is determined based on the trajectory data, and the movement trajectory is marked in the three-dimensional scene model; The location and shape of map elements are determined based on the composition data, and the map elements are marked in the three-dimensional scene model according to the location and shape. Analyze the vertical spatial relationship between map features and movement trajectories in the slice layer to obtain the relative height information between map features and movement trajectories; Based on the degree of continuity of the relative height information, a discrete object model and a continuous field model are established respectively; The relative height information of the discrete object model is subjected to a grading error analysis to obtain grading detail information; The relative height information of the continuous field model is decomposed by slope curvature to obtain spatial grid density information; The relative height classification information of map features is determined based on the classification detail information and spatial grid density information.

[0011] In one embodiment, the step of reviewing the compliance of auxiliary positioning elements based on the composition data to obtain the relative height classification information of the auxiliary positioning elements includes: Extracting road feature points, intersection feature points, and points of interest from graph-based data; Based on the location information of road feature points, intersection feature points, and points of interest, an auxiliary positioning element network is constructed; Relative height analysis is performed based on the aforementioned auxiliary positioning element network to obtain the relative height information of the auxiliary positioning elements; The relative height information of the auxiliary positioning elements is processed to obtain the relative height classification information of the auxiliary positioning elements.

[0012] In one embodiment, the step of comprehensively evaluating the high-precision map based on the road perimeter extension range, the relative height classification information of map features, and the relative height classification information of auxiliary positioning features to obtain the map review result includes: The deviation rate of the road extension range is determined based on the preset road extension range compliance standards; Obstacle recognition simulation and prediction are performed based on the relative height classification information of the map elements to obtain the obstacle recognition accuracy. The positioning accuracy is simulated and predicted based on the relative height classification information of the auxiliary positioning elements to obtain the positioning accuracy rate; A comprehensive evaluation score for the high-precision map is obtained by comprehensively evaluating the road extension range deviation rate, the obstacle recognition accuracy rate, and the positioning accuracy rate. The comprehensive evaluation score is compared with the preset map review passing score to obtain the map review result of the high-precision map.

[0013] Furthermore, to achieve the above objectives, this application also proposes a high-precision map data processing device, which includes: The acquisition module is used to acquire geographic information data from high-precision maps; The standardization module is used to standardize the geographic information data based on the spatiotemporal semantic perception adaptive standardization engine to obtain standardized geographic information data. The classification and grading module is used to perform horizontal classification of the standardized geographic information data based on an automatic clustering algorithm using multimodal feature extraction and vertical grading based on a knowledge graph reasoning algorithm using influence assessment, resulting in a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data sets of different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data. An encryption and desensitization module is used to encrypt and desensitize the hierarchical data set to obtain an encrypted and desensitized data set. The review module is used to review the spatial range expansion, relative height requirements, and compliance of auxiliary positioning elements of the high-precision map based on the encrypted and desensitized data set, and obtain the review result of the high-precision map. The publishing module is used to optimize the high-precision map data based on the encrypted and desensitized data set when the map review result is that the high-precision map has passed the review, and then publish the optimized high-precision map data to the map service platform.

[0014] This application proposes one or more technical solutions to acquire high-precision map geographic information data; standardize the geographic information data based on an adaptive standardization engine with spatiotemporal semantic awareness to obtain standardized geographic information data; perform horizontal classification on the standardized geographic information data based on an automatic clustering algorithm using multimodal feature extraction and vertical hierarchical classification based on a knowledge graph inference algorithm using influence assessment to obtain a hierarchical data set, wherein the hierarchical data set includes multiple data sets of different categories, each category of data set includes multiple data at different levels, and the different categories of data sets include image data sets, point cloud data sets, etc. The system comprises trajectory-based, inertial navigation-based, and mapping-based datasets, with different levels of data including core data, important data, and general data. These hierarchical datasets are encrypted and anonymized to obtain encrypted and anonymized datasets. Based on these encrypted and anonymized datasets, the high-precision map undergoes spatial range expansion review, relative height requirement review, and compliance review of auxiliary positioning element representation, resulting in a high-precision map review result. If the high-precision map passes the review, it is optimized based on the encrypted and anonymized datasets, and the optimized high-precision map data is then published to the map service platform. Through this approach, by introducing a classification and hierarchical adaptive processing mechanism and data encryption and anonymization technology, efficient utilization of high-precision map data is achieved. Furthermore, by conducting multiple reviews of the high-precision map, the efficiency and accuracy of the review process are improved, while effectively ensuring the integrity and security of the high-precision map data. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating an embodiment of the high-precision map data processing method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the high-precision map data processing method of this application. Figure 3 This is a schematic diagram of the module structure of the high-precision map data processing device according to an embodiment of this application.

[0018] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0020] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of this application embodiment is as follows: acquiring high-precision map geographic information data; standardizing the geographic information data based on an adaptive standardization engine with spatiotemporal semantic awareness to obtain standardized geographic information data; performing horizontal classification on the standardized geographic information data based on an automatic clustering algorithm using multimodal feature extraction and vertical hierarchical classification based on a knowledge graph reasoning algorithm using influence assessment to obtain a hierarchical data set, wherein the hierarchical data set includes multiple data sets of different categories, each category of data set includes multiple data at different levels, and the different categories of data sets include image data sets, point cloud data sets, and orbital data sets. The high-precision map is divided into three data sets: trace data, inertial navigation data, and mapping data. Data at different levels includes core data, important data, and general data. The hierarchical data sets are encrypted and anonymized to obtain encrypted and anonymized data sets. Based on these encrypted and anonymized data sets, the high-precision map undergoes spatial range expansion review, relative height requirement review, and compliance review of auxiliary positioning element expression to obtain the high-precision map review result. If the high-precision map review result is satisfactory, the high-precision map data is optimized based on the encrypted and anonymized data sets, and the optimized high-precision map data is published to the map service platform.

[0022] Traditional map review methods often focus on reviewing basic information such as the geometric accuracy of the map and the connectivity of the road network. However, for high-precision maps, this information can no longer meet the high requirements of autonomous driving algorithms for safety redundancy, resulting in low efficiency and poor accuracy in map review, and making it difficult to ensure the integrity and security of high-precision map data.

[0023] This application provides a solution that achieves efficient utilization of high-precision map data by introducing a classification and grading adaptive processing mechanism and data encryption and desensitization technology. Furthermore, by conducting multiple reviews of high-precision maps, the efficiency and accuracy of map review are improved, while effectively ensuring the integrity and security of high-precision map data.

[0024] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or high-precision map data processing device capable of performing the above functions. The following description uses a high-precision map data processing device as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, the embodiments of this application provide a high-precision map data processing method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the high-precision map data processing method of this application.

[0026] In this embodiment, the high-precision map data processing method includes steps S10 to S60: Step S10: Obtain geographic information data from a high-precision map.

[0027] It's important to note that high-precision maps, also known as autonomous driving maps, are maps used for high-precision positioning and environmental perception in autonomous driving systems. The geographic information data for high-precision maps is obtained through a multi-source heterogeneous data access layer that collects and integrates incoming data. This layer includes a unified data access middleware that supports access to data from vehicle-mounted LiDAR, drone aerial imagery, RTK-GNSS trajectories, IMU inertial navigation data, crowdsourced testing data, and road sign OCR recognition results. Access protocols support MQTT, HTTP / 3, and ROS2, adapting to real-time uploads from autonomous vehicles, thus achieving real-time updates and comprehensive coverage of high-precision map data. This data encompasses multiple categories, including imagery, point clouds, trajectories, inertial navigation data, and mapping, providing rich foundational information for subsequent data processing.

[0028] Step S20: The geographic information data is standardized by an adaptive standardization engine based on spatiotemporal semantic awareness to obtain standardized geographic information data.

[0029] It should be noted that standardization is to ensure that data from different sources and in different formats can be processed and analyzed uniformly. The spatiotemporal semantic-aware adaptive standardization engine used in this embodiment can perform spatiotemporal semantic alignment, intelligent coordinate system transformation, and attribute semantic normalization on geographic information data, ultimately outputting a standardized data format. This solves the problem of data heterogeneity and provides a unified data format and standard for subsequent data processing.

[0030] In one feasible implementation, step S20 may include: constructing a unified spatiotemporal reference framework based on the global dynamic reference frame, the international Earth reference frame, and a regional crustal deformation model; performing spatiotemporal semantic alignment on the geographic information data according to the unified spatiotemporal reference framework to obtain spatiotemporally aligned data; performing coordinate system transformation on the spatiotemporally aligned data using a coordinate system transformation algorithm to obtain coordinate system transformed data; constructing a geographic feature ontology knowledge base, and determining the entity standard attribute structure based on the geographic feature ontology knowledge base; performing attribute semantic normalization processing on the coordinate system transformed data according to the entity standard attribute structure to obtain attribute semantic normalized data; and encoding the attribute semantic normalized data according to a preset hybrid encoding format to obtain standardized geographic information data.

[0031] It should be noted that the Global Dynamic Baseline Frame (GDBF) is a high-precision geospatial reference frame used to provide dynamically updated baseline data, ensuring the accuracy and reliability of global positioning and measurement. The International Geodetic Reference Frame (ITRF) is a global geographic coordinate reference system used to provide an accurate global geodetic reference frame. A regional crustal deformation model is a mathematical model describing the deformation of the Earth's crust over time within a specific region.

[0032] Understandably, by introducing a global dynamic benchmark frame and combining it with the International Earth Reference Frame and regional crustal deformation models, a unified spatiotemporal benchmark framework is constructed, enabling centimeter-level dynamic coordinate correction. Using this unified spatiotemporal benchmark framework to perform spatiotemporal semantic alignment of geographic information data effectively improves the spatiotemporal accuracy and consistency of the data.

[0033] Understandably, the Transformer-based GeoAlignNet model automatically identifies the original coordinate system, timestamp format, and unit system in the spatiotemporally aligned data and intelligently matches them to the target standard, such as WGS84+EGM2008 elevation+UTC time, to achieve intelligent conversion between different coordinate systems, thereby ensuring the accuracy and consistency of geographic information.

[0034] Attribute semantic normalization, based on a geographic feature ontology knowledge base, provides a unified description and standardization of data attribute information, resolving the issue of inconsistent attribute information. Encoding data using a hybrid encoding format not only achieves standardized data storage but also facilitates rapid data retrieval and efficient utilization.

[0035] It is worth noting that the geographic feature ontology knowledge base is a database used to store and manage geographic feature ontology knowledge. Based on the geographic feature ontology knowledge base, standard attribute structures for entities such as "lane lines," "traffic lights," and "roadside guardrails" can be defined. Then, the standard attribute structures of these entities are parsed and normalized to ensure that the attribute information of each geographic feature conforms to unified standards and specifications.

[0036] The default hybrid encoding format is eoJSON-Lite + Protobuf, balancing readability and transmission efficiency. Fields strictly adhere to the ISO 19157 Geographic Information Quality Standard, ensuring efficient data transmission and accuracy. Data with attribute semantic normalization is efficiently compressed and stored using the eoJSON-Lite + Protobuf hybrid encoding format, reducing storage space usage, accelerating data retrieval, and improving overall data processing efficiency.

[0037] Step S30: The standardized geographic information data is horizontally classified based on an automatic clustering algorithm using multimodal feature extraction and vertically graded based on a knowledge graph reasoning algorithm using influence assessment to obtain a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data at different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data.

[0038] It should be noted that classifying and grading the standardized geographic information data is to improve data utilization efficiency. The automatic clustering algorithm using multimodal feature extraction employed in this embodiment can automatically identify data features and cluster them, thereby classifying the data into different categories, such as imagery, point clouds, trajectories, inertial navigation systems, and mapping. This automatic clustering method not only improves data processing efficiency but also ensures the accuracy and objectivity of the classification.

[0039] Meanwhile, the knowledge graph reasoning algorithm based on impact assessment can vertically classify data. By analyzing the correlations and impact between data points, the algorithm categorizes data into different levels, such as core data, important data, and general data. This hierarchical processing approach helps to adopt different processing strategies and priorities for different levels of data during subsequent data processing and review, thereby improving overall processing efficiency and accuracy.

[0040] Through classification and grading, a hierarchical dataset was obtained. This dataset not only includes multiple datasets of different categories, such as image datasets, point cloud datasets, trajectory datasets, inertial navigation datasets, and mapping datasets, but also includes different levels of data within each category, such as core data, important data, and general data.

[0041] It is worth noting that image data refers to images or videos collected by spatiotemporal data sensors or drone aerial photography; point cloud data includes 3D point cloud data acquired through LiDAR sensors; trajectory data is vehicle trajectory data acquired through sensors such as RTK-GNSS and IMU, which is data that can determine the vehicle's absolute coordinates on Earth; inertial navigation data is vehicle attitude and motion state data acquired through IMU inertial navigation systems, such as vehicle attitude angles (or angular rates), acceleration, and their derived data; and mapping data is vector data containing absolute coordinates generated based on the above data.

[0042] It is worth noting that data levels are determined according to the following rules: 1. Intelligent vehicle basic map data that meets any of the following conditions should be identified as core data: Important data covering the entire country and possessing a large scale; meeting the core data level determination rules stipulated in GB / T 43697. 2. Intelligent vehicle basic map data that meets any of the following conditions should be identified as important data: Geographic information reflecting important and sensitive areas such as military management zones, national defense science and technology units, and party and government organs at the county level and above; national strategic reserve capacity; relevant attributes of important infrastructure such as transportation and energy; imagery data, point cloud data, trajectory data, inertial navigation data, mapping data, map product data, and other data that reach a certain scale and involve important targets and areas; meeting the important data level determination rules stipulated in GB / T 43697. 3. Other data not identified as core data or important data is identified as general data.

[0043] In one feasible implementation, step S30 may include: extracting features from the standardized geographic information data based on a multimodal embedding model to obtain feature vectors for various types of data; mapping the feature vectors of various types of data to a unified semantic embedding space to obtain a unified semantic embedding vector; clustering the unified semantic embedding vector based on the HDBSCAN clustering algorithm to obtain data sets of different categories; constructing a high-precision map risk knowledge graph, wherein the nodes in the high-precision map risk knowledge graph include geographic elements, functional impacts, and safety and traffic levels; based on the high-precision map risk knowledge graph, using an impact assessment algorithm to assess the impact of the data in the data sets of different categories to obtain a decision influence score; constructing a knowledge graph reasoning path based on the decision influence score, and using the knowledge graph reasoning path to vertically classify the data sets of different categories to obtain a hierarchical data set.

[0044] It should be noted that the multimodal embedding model MM-Embedder is used to extract feature vectors for various types of data. For example, for image data, CNN is used to extract texture, color, and object distribution features; for point cloud data, PointNet++ is used to extract geometric density, curvature, and reflection intensity distribution; for trajectory data, LSTM is used to extract motion patterns and acceleration distribution; for inertial navigation data, wavelet analysis is used to extract vibration spectrum features; and for graph data, graph neural networks are used to extract topological connectivity.

[0045] Mapping feature vectors of various types of data to a unified semantic embedding space can eliminate the semantic gap between different modalities, making the data comparable and consistent at the semantic level, thus facilitating clustering.

[0046] Clustering unified semantic embedding vectors using the HDBSCAN clustering algorithm can automatically identify the inherent structure and distribution patterns of data, classifying it into different categories such as image, point cloud, trajectory, inertial navigation, and mapping. This automatic clustering method not only improves data processing efficiency but also ensures the accuracy and objectivity of classification.

[0047] In the specific implementation, since the embedding vector is a high-dimensional vector in the semantic space, cosine distance is used to measure similarity, as shown in the following formula: in, Let z be the cosine distance. i To represent the unified semantic embedding vector of the i-th geographic information data, z j This represents the semantic embedding vector of the j-th geographic information data.

[0048] HDBSCAN defines clustering based on local density, with core metric being core distance. It assumes a cluster contains at least k points, and for each point z... i Its core distance is: core_distance k (z i ) = the k-th smallest distance to other points, i.e., the distance from all points in dist(z) i ,z j In the equation, the k-th smallest value is taken. The core distance is used to measure the density of the region where the point is located. The higher the density, the smaller the core distance.

[0049] If both points are very close together, meaning the core distance is small, then we introduce the reachability distance, which is taken as z. i core distance k (z i ), z j core distance k (z j ) and z i and z j The maximum value of the pre-defined distance between points. In this case, the reachability distance is close to the original distance. If a point is sparse, the reachability distance is increased, preventing it from becoming a cluster center.

[0050] Then, a complete graph is constructed using a cross-distance matrix, where each node in the complete graph represents all N embedding vectors, and the edge weights are the cross-distances. Next, Kruskal's algorithm is run on the complete graph to generate a minimum spanning tree. The minimum spanning tree is then pruned according to edge weights from largest to smallest to simulate clustering states under different density thresholds. Generate a hierarchical clustering structure, i.e., multiple candidate clusters; for each candidate cluster, calculate its existence time at different density thresholds, i.e., cluster survival time cst, as follows:

[0051] in, Let i be the cluster lifetime of data point i. Let C be a cluster containing point i, and λ = 1 / distance, representing the density intensity. For the cluster's birth strength parameter, This is the cluster death intensity parameter.

[0052] We select the clustering partition that maximizes the total CST, obtaining K valid clusters and cluster labels for each point. We then map these cluster labels to semantic categories, such as image categories. Specifically, this includes: for each cluster C... k The original data sources of its members are statistically analyzed as follows:

[0053] in, For a given cluster C k The probability that its members come from a certain category, C k For the k-th cluster, The is a counting symbol that indicates the number of elements in a set. For vector z i The source function.

[0054] Set a threshold for automatic labeling, if If so, it is labeled as an image; Labeled as point cloud type; if It is labeled as a trajectory / inertial navigation type; It is labeled as a composition type.

[0055] After data classification, a high-precision map risk knowledge graph is constructed. Nodes in the high-precision map risk knowledge graph represent key information such as geographic features, functional impacts, and security levels, while edges represent the relationships between these nodes. For each geographic feature *e*, all its propagation paths in the knowledge graph are found, and the weight of each path is:

[0056] in, The weights of the propagation path p, The normalization coefficient is... The confidence level of the edge is 0 to 1, derived from expert annotations or historical data statistics. This is the attenuation factor, usually set to 0.9. Score the severity of the functional impact. This is the security level mapping value.

[0057] The influence score of an element can be further determined based on the weight of the path:

[0058] in, The influence score for geographic element e. Let P(e) be the weight of the propagation path p, and let P(e) be all paths originating from e.

[0059] Geographic elements are categorized by data source, and then the decision influence score for each data category is determined based on the element influence score. Based on the decision influence score, an inference path is constructed to achieve vertical hierarchical management of data, i.e., core > important > general.

[0060] Step S40: Encrypt and desensitize the hierarchical data set to obtain the encrypted and desensitized data set.

[0061] It should be noted that encrypting and de-identifying data can effectively prevent data leakage and misuse, thus ensuring data security. In this embodiment, hierarchical data sets are classified, de-identified, and encrypted hierarchically.

[0062] Category-based desensitization refers to processing data using different desensitization strategies based on the data category. For example, for image data, desensitization can be achieved through pixelation, blurring, or adding noise. For point cloud data, desensitization can be achieved through attribute differential privacy processing, random offsetting, or removing some features. For trajectory data and inertial navigation data, desensitization can be achieved through adding noise, temporal blurring, or spatial generalization. For mapping data, desensitization can be achieved by adjusting the topology, changing node attributes, or adding dummy nodes. This embodiment does not impose specific limitations on these methods.

[0063] Tiered encryption uses different encryption algorithms and key lengths based on the sensitivity level of the data to ensure security during transmission and storage. For example, core data can be encrypted with high-strength algorithms and long keys to ensure high security; important data can be encrypted with medium-strength algorithms and key lengths; and general data can be protected with lower-strength algorithms. This categorized desensitization and tiered encryption effectively improves data security and prevents data leakage and misuse.

[0064] In one feasible implementation, step S40 may include: applying a classification desensitization strategy to the hierarchical data set to obtain desensitized data sets of different categories, wherein the classification desensitization strategy includes a semantic segmentation-guided local fuzzing desensitization strategy for image data sets, an attribute differential privacy desensitization strategy for point cloud data sets, a spatiotemporal randomization desensitization strategy for trajectory data sets, a noise addition desensitization strategy for inertial navigation data sets, and a hash obfuscation desensitization strategy for mapping data sets; and encrypting data at different levels in the desensitized data sets of different categories using a hierarchical encryption strategy to obtain hierarchically encrypted data sets.

[0065] It should be noted that, in this embodiment, corresponding desensitization strategies are adopted for different types of data sets. For image data sets, a semantic segmentation-guided local fuzziness desensitization strategy is adopted. For example, only sensitive areas such as faces, license plates, and house numbers are replaced using GAN generative replacement to preserve the clarity of road structures. For point cloud data sets, a perturbation and attribute differential privacy desensitization strategy is adopted. For example, point clouds 1.5 meters above the ground are perturbed along the Z-axis with a perturbation height of ±0.3m to maintain the ground geometry. Attribute differential privacy is used to add Laplace noise to the reflection intensity. For trajectory data sets, a spatiotemporal randomization desensitization strategy is adopted, that is, through spatiotemporal k-anonymity + continuous privacy protection, it is ensured that at least k trajectories are indistinguishable within any time period. For inertial navigation data sets, a noise addition desensitization strategy is adopted. For example, after performing Fourier transform on the acceleration and angular velocity sequences, noise is added in the frequency domain, and then inverse transform is performed to restore the motion trend. For graph data sets, a hash obfuscation desensitization strategy is adopted. For example, the node IDs in the graph are hash obfuscated, the edge relationships are preserved, but the attributes are desensitized.

[0066] Understandably, after anonymization, different levels of data in different categories of data sets are encrypted hierarchically according to data level. For example, core data uses quantum-resistant encryption algorithms, such as CRYSTALS-Kyber, for end-to-end encryption, with the key managed by a Trusted Execution Environment (TEE); important data uses AES-256-GCM encryption with additional integrity checks; and general data uses lightweight ChaCha20 encryption, suitable for rapid processing by edge devices.

[0067] In practical implementation, the specific implementation methods of classification-based desensitization and hierarchical encryption strategies can be adjusted and optimized according to actual needs. For example, in the classification-based desensitization stage, deep learning models can be used for automated desensitization processing, automatically selecting the most suitable desensitization strategy based on the characteristics of the data; in the hierarchical encryption stage, the encryption algorithm and key length can be dynamically adjusted according to the data's access permissions and usage scenarios to achieve more flexible and fine-grained data protection. Furthermore, to ensure the efficiency and accuracy of data processing, parallel processing and distributed computing technologies can be employed to efficiently process large-scale datasets.

[0068] Step S50: Based on the encrypted and desensitized data set, the high-precision map is reviewed for spatial range expansion, relative height requirements, and compliance of auxiliary positioning elements, to obtain the review result of the high-precision map.

[0069] It should be noted that the spatial range expansion is necessary in complex traffic scenarios with blind spots or weak signal areas. To compensate for the vehicle's insufficient timely perception and effectively handle emergency braking situations such as sudden pedestrian appearances, advanced autonomous driving algorithms need to use high-precision maps to "prioritize" the traffic environment within a certain buffer zone on both sides of the road. This trades map space for braking time to ensure driving safety at certain speeds. However, the expanded area may involve rich geographical features around the road, potentially including sensitive locations, high-voltage power lines, substations, and other important public service facilities, posing potential geographic information security risks. Therefore, it is necessary to study the distance of the expandable buffer zone. Understandably, the requirement for relative height is due to the fact that advanced autonomous driving algorithms need to understand important information such as relative height, height levels, and near-surface undulations (relative elevation differences). For example, they need the relative height of traffic lights to distinguish them from the taillights of cars ahead; they need the relative height hierarchy of elevated roads to ensure that the road level can be quickly identified when starting on an elevated road; and they need to understand the changes in near-surface relative height undulations to distinguish near-surface road protrusions from low-lying objects that temporarily appear on the roadside, such as people sitting on the roadside or small animals that suddenly rush out. However, contiguous relative height information is related to geographic information security, and current high-altitude auxiliary maps do not allow direct expression of elevation information. Therefore, the review of relative height needs to use indirect and ambiguous methods to express relative height information to ensure the safety and compliance of map use.

[0070] The need for auxiliary positioning elements stems from the fact that the spatial positioning accuracy of high-level autonomous driving is generally within 20cm. To compensate for the inaccuracy caused by weak positioning signals, it is necessary to use high-precision maps to provide prior spatial feature positioning information. This involves selecting spatial feature positioning elements, such as point, line, and surface geometric elements, in the road and surrounding scene to construct a description of the spatial scene. Then, by matching the position with the spatial scene perceived by the vehicle in real time, accurate spatial positioning can be quickly achieved.

[0071] Therefore, by comprehensively reviewing multiple aspects, including spatial extent expansion, relative height requirement, and compliance of auxiliary positioning element representation, the safety, compliance, and accuracy of high-precision maps in autonomous driving applications can be ensured. Spatial extent expansion review primarily focuses on whether the expanded area involves sensitive geographical locations and whether the expansion distance is reasonable, avoiding the leakage of important geographical information. Relative height requirement review focuses on checking whether the expression of relative height information is indirect or ambiguous to prevent the leakage of contiguous relative height information. Compliance of auxiliary positioning element representation review focuses on whether the selected spatial feature positioning elements are accurate and compliant, and whether they can effectively support precise spatial positioning for autonomous driving.

[0072] Step S60: If the map review result is that the high-precision map has passed the review, optimize the high-precision map data according to the encrypted and desensitized data set, and publish the optimized high-precision map data to the map service platform.

[0073] It's important to note that optimizing high-precision map data ensures its efficient and accurate utilization by autonomous driving systems on the map service platform. The map service platform can be a cloud server or a distributed storage system used to store, manage, and provide high-precision map data services. Before publishing the optimized high-precision map data to the map service platform, further format conversion and compression operations can be performed to adapt to the needs of different autonomous driving systems and devices.

[0074] In practical implementation, optimization steps may include, but are not limited to, improving the accuracy of map data, removing data redundancy, and optimizing path planning algorithms. Improving accuracy can be achieved by further refining the encrypted and anonymized dataset, such as by depicting road boundaries, traffic signs, and obstacles in greater detail, thereby enhancing the autonomous vehicle's ability to recognize the road environment. Removing data redundancy aims to reduce the size of the map data and improve data transmission and loading efficiency; this can be achieved through methods such as compressing the map data and merging similar elements. Optimizing path planning algorithms addresses the driving needs of autonomous vehicles by intelligently analyzing the map data and optimizing the path planning algorithm to improve the efficiency and safety of autonomous driving.

[0075] This embodiment provides a high-precision map data processing method to acquire high-precision map geographic information data; the geographic information data is standardized based on an adaptive standardization engine with spatiotemporal semantic awareness to obtain standardized geographic information data; the standardized geographic information data is horizontally classified based on an automatic clustering algorithm using multimodal feature extraction and vertically graded based on a knowledge graph inference algorithm using influence assessment to obtain a hierarchical data set, wherein the hierarchical data set includes multiple data sets of different categories, each category of data set includes multiple data at different levels, and the different categories of data sets include image data sets, point cloud datasets, etc. The system comprises three data sets: a combination dataset, a trajectory dataset, an inertial navigation dataset, and a mapping dataset. These datasets are categorized into different levels, including core data, important data, and general data. The hierarchical datasets are then encrypted and anonymized to obtain encrypted and anonymized data sets. Based on these encrypted and anonymized data sets, the high-precision map undergoes spatial range expansion review, relative height requirement review, and compliance review of auxiliary positioning element representation, resulting in a high-precision map review result. If the high-precision map passes the review, it is optimized using the encrypted and anonymized data sets, and the optimized high-precision map data is then published to the map service platform. This approach, by introducing a classification and hierarchical adaptive processing mechanism and data encryption and anonymization technology, achieves efficient utilization of high-precision map data. Furthermore, by conducting multiple reviews of the high-precision map, the efficiency and accuracy of the review process are improved, while effectively ensuring the integrity and security of the high-precision map data.

[0076] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S10 includes steps S501 to S504: Step S501: Determine image data, point cloud data, trajectory data, inertial navigation data, and mapping data based on the encrypted and desensitized data set.

[0077] It should be noted that the security and privacy protection of the data in the encrypted and anonymized dataset have been enhanced. Image data typically comes from satellite remote sensing, drone aerial photography, or vehicle-mounted cameras, providing rich information on roads and the surrounding environment; point cloud data is acquired through devices such as LiDAR, which can accurately depict the three-dimensional shape of objects; trajectory data records the motion paths of vehicles or other moving objects; inertial navigation data contains inertial navigation information such as acceleration and angular velocity, which helps determine the position and attitude of vehicles; and mapping data is vector data containing absolute coordinates obtained by integrating the above types of data.

[0078] Step S502: Perform spatial range expansion verification based on the image data, point cloud data, and inertial navigation data to obtain the extended range around the road.

[0079] It should be noted that during the spatial range expansion review process, the macroscopic road and surrounding environment overview provided by imagery data, the precise three-dimensional shape information of objects depicted by point cloud data, and the vehicle position and attitude information determined by inertial navigation data will be comprehensively considered. These information will work together to determine the expansion range around the road, ensuring that the expansion area can cover all areas that autonomous vehicles may travel in, while avoiding overly sensitive geographical locations. By analyzing this data through algorithms, a reasonable expansion distance can be obtained, which can meet the safety requirements of autonomous driving while avoiding the leakage of critical geographical information.

[0080] In one feasible implementation, step S502 may include: extracting a road image based on the image data, and performing an inverse perspective transformation based on the vanishing point on the road image to obtain a corrected road image; preprocessing the point cloud data using a radius filtering algorithm, and extracting road surface point cloud data based on the preprocessed point cloud data; extracting road traffic markings based on the corrected road image and the road surface point cloud data; extracting vehicle speed data based on the inertial navigation data, and constructing a speed-reaction time-environment regression model based on the vehicle speed data and the corresponding safety distance requirements; determining a reference distance matrix for the outward expansion of the road perimeter based on the speed-reaction time-environment regression model; and determining the outward expansion range of the road perimeter based on the road traffic markings and the reference distance matrix for the outward expansion of the road perimeter.

[0081] It should be noted that the vanishing point-based inverse perspective transform recovers spatial information from the image by identifying the vanishing points—the points where all parallel lines intersect at a distance—thus correcting road images distorted by perspective effects. The radius filtering algorithm preprocesses the point cloud data, retaining points within a certain radius around each point, removing noise and outliers, and improving the accuracy and reliability of the point cloud data. Extracting road surface point cloud data from the preprocessed point cloud data involves analyzing the spatial distribution characteristics of the point cloud data to identify point cloud data belonging to the road surface.

[0082] Understandably, by combining corrected road images and road surface point cloud data to extract road markings, the complementary advantages of image and point cloud data can be fully utilized to improve the accuracy and robustness of road marking extraction. Road markings are an important basis for autonomous vehicles to perform path planning and driving decisions. Accurate road marking information helps autonomous vehicles correctly identify road rules, improving driving safety and efficiency.

[0083] In practical implementation, different vehicle speeds are extracted based on inertial navigation data, and safety distance requirements are extended under different braking reaction times, weather conditions, road conditions, etc., to establish a speed-reaction time-environment regression model. This model shows that the buffer distance is positively correlated with vehicle speed and reaction braking time, and strongly correlated with influencing factors such as weather and road surface smoothness. The speed-reaction time-environment regression model is as follows:

[0084] in, The required safe buffer distance, or safe distance requirement, is given by v, where v is the vehicle speed. For braking reaction time, The coefficient of friction is the overall friction factor, which is affected by weather conditions (w) and road conditions (r), and g is the acceleration due to gravity.

[0085] Based on the speed-reaction time-environment regression model, statistical analysis was conducted to form a reference distance matrix for the expansion of the road perimeter under different conditions. Several road-based auxiliary positioning elements were selected as sample objects based on road traffic markings, and their distances from the roadside were statistically analyzed to ultimately determine the reference range for road perimeter expansion. Taking Guangzhou as an example, considering the construction status of Guangzhou's municipal roads, surrounding traffic facilities, and spatial relationships of geographical locations, 36 road-based auxiliary positioning elements, including Xingdao Huanbei Road (simple urban road), University Town Outer Ring West Road (complex urban road), and Nansha Port Expressway (high-speed road), were selected as sample objects. Their distances from the roadside were statistically analyzed, and a range of 15 meters was ultimately selected as the reference range for the expansion of the road perimeter in Guangzhou. Regarding construction status, the selection rule was that the road must not be under construction; regarding surrounding facilities, the selection rule was that there were temporary parking spaces and roadside features such as support poles, tree trunks, and lampposts around the road, and the expansion range should include these roadside features.

[0086] As shown in Table 1, this table contains data on roadside buffer reference distances. It includes roadside buffer reference distances under different speed limits and vehicle sensor response time limits, along with corresponding suggested buffer distances. A suggested reference distance is 0.5 seconds as the standard upper limit for response time. For example, with a speed limit of 80 km / h and a vehicle sensor response time limit of 0.5 seconds, the roadside buffer reference distance is 15 meters, and the suggested buffer distance is 16 meters. A 15-meter buffer zone can encompass most auxiliary positioning feature characteristics; therefore, 15 meters is selected as the roadside perimeter extension range for Guangzhou.

[0087] Table 1

[0088] Step S503: Based on the point cloud data, trajectory data, and composition data, perform relative height requirement review to obtain the relative height classification information of map elements.

[0089] It's important to note that the relative height requirement review process comprehensively considers the 3D shape information of objects provided by point cloud data, the vehicle movement path information recorded by trajectory data, and the absolute coordinate information integrated from mapping data. This information works together to determine the relative height of map elements. Relative height information ensures the safe operation of autonomous vehicles under different road conditions while avoiding data redundancy and privacy leaks caused by overly detailed height information. Map elements are categorized into different levels based on their relative height to determine their obstacle recognition accuracy, thus enabling the relative height requirement review process.

[0090] In one feasible implementation, step S503 may include: constructing a three-dimensional scene model based on point cloud data, and performing height slicing on the three-dimensional scene model to obtain slice layers of different heights; determining the movement trajectory of a moving object based on the trajectory data, and marking the movement trajectory in the three-dimensional scene model; determining the position and shape of map elements based on the composition data, and marking the map elements in the three-dimensional scene model according to the position and shape; analyzing the vertical spatial relationship between map elements and movement trajectories in the slice layers to obtain relative height information between map elements and movement trajectories; establishing discrete object models and continuous field models respectively according to the continuity of the relative height information; performing grading error analysis on the relative height information of the discrete object models to obtain grading fineness information; performing slope curvature decomposition on the relative height information of the continuous field models to obtain spatial grid density information; and determining the relative height grading information of map elements according to the grading fineness information and spatial grid density information.

[0091] It's important to note that 3D scene models built from point cloud data can realistically reflect the spatial characteristics of roads and their surrounding environment. Height slicing involves dividing the 3D scene model into layers of 2D slices at different heights, which display spatial information at different elevations. Trajectory data, recording the movement trajectory and marking it within the 3D scene model, helps in understanding the movement paths of moving objects, such as autonomous vehicles. Mapping data, providing the location and shape information of map features, marks key elements such as roads, buildings, and trees within the 3D scene model. By analyzing the vertical spatial relationship between map features and movement trajectories in the slice layers, their relative height information can be obtained, which is crucial for autonomous vehicles in path planning and obstacle avoidance decisions.

[0092] Discrete object models are suitable for representing spatially independent map features with discontinuous height information, such as traffic lights and signs, represented by points, lines, and polygons at specific locations. By performing grading error analysis on the relative height information of these features, an appropriate grading granularity can be determined—the difference between adjacent height gradations—ensuring that autonomous vehicles can accurately identify and avoid these obstacles. Continuous field models, on the other hand, are suitable for representing spatially continuous map features with relatively smooth height information, such as roads and bridges. By performing slope curvature decomposition on the relative height information of these features, spatial grid density information can be obtained—a parameter describing the rate of change of height information in space. This helps autonomous vehicles maintain safe driving under different road conditions.

[0093] In the specific implementation, data models such as discrete object models and continuous field models are established according to the degree of continuity of relative height. The relative height of the discrete object model is expressed in a graded and hierarchical manner, and its main review content is the graded fineness. The continuous field model mainly adopts the slope and curvature decomposition to the height difference, and its review content is mainly the fineness of the two-dimensional plane grid unit and the decomposition based on slope and curvature.

[0094] For elements such as traffic lights and road signs that belong to discrete object models, error analysis was conducted at different levels of 0.2 meters, 0.3 meters, 0.4 meters, 0.5 meters, 0.6 meters, and 1.0 meters. The simulated sample size was 288 samples. The accuracy of the recognition was verified through field testing. Finally, the fineness of the classification was determined to be 0.2 meters and 0.5 meters.

[0095] As shown in Table 2, Table 2 is a schematic table of relative height classification of traffic lights. The table includes the model of traffic lights, the number of samples, and the number and percentage of misjudgments under different classifications. The number of samples for the 400 type traffic light is 126, and the number of samples for the 300 type traffic light is 90. The number and percentage of misjudgments for both are 0 under the 0.2-meter classification.

[0096] Table 2

[0097] As shown in Table 3, Table 3 is a schematic table of relative height classification of traffic signs. The table shows the shape of the traffic signs, the number of samples, and the number and percentage of misjudgments under different classifications. The number of samples for round traffic signs is 104, and the number of samples for square traffic signs is 112. The number and percentage of misjudgments for both are 0 under the 0.5-meter classification.

[0098] Table 3

[0099] To address the decomposition and representation of relative height information in continuous field models, slope curvature decomposition is employed. Spatial grids are used to represent height information, and the optimal grid density that meets both compliance and autonomous driving requirements is validated. Testing and verification are primarily conducted on two finenesses: 0.5m × 0.5m and 1m × 1m. Test points were selected at road shoulders and within 10 meters of the road shoulders for multi-case verification. The results showed that the 0.5m × 0.5m grid accurately reproduces the relative undulations around the vehicle, meeting the safety requirements of autonomous driving. The 1m × 1m grid, however, is not precise enough, affecting obstacle recognition in autonomous driving and consequently impacting vehicle and road traffic safety. Therefore, the 0.5m × 0.5m unit grid is adopted.

[0100] As shown in Table 4, Table 4 contains the error verification data after the grid sparsity processing. The table includes the test range, the total number of sampling points, and the number and proportion of points with different error ranges. For example, the total number of sampling points in the shoulder is 58,461,238, the number of points with an error exceeding 20cm is 89,072, the proportion is 0.15%, and the number of points with an error exceeding 30cm is 69,107, the proportion is 0.12%.

[0101] Table 4

[0102] Step S504: Based on the composition data, review the compliance of the auxiliary positioning elements to obtain the relative height classification information of the auxiliary positioning elements.

[0103] It should be noted that the compliance review of auxiliary positioning elements refers to classifying the relative height of auxiliary positioning elements to ensure their accurate representation and compliance in autonomous driving maps. The positioning accuracy can be determined based on the relative height classification information of the auxiliary positioning elements, thereby achieving the compliance review of their representation.

[0104] In one feasible implementation, step S504 may include: extracting road feature points, intersection feature points, and points of interest based on the mapping data; constructing an auxiliary positioning element network based on the location information of the road feature points, intersection feature points, and points of interest; performing relative height analysis based on the auxiliary positioning element network to obtain the relative height information of the auxiliary positioning elements; and performing grading processing based on the relative height information of the auxiliary positioning elements to obtain the relative height grading information of the auxiliary positioning elements.

[0105] It should be noted that the composition - related data usually contains key information such as high - precision road geometric information, intersection layouts, and the locations of points of interest (POIs). Road feature points mainly reflect geometric features such as the road's orientation and curvature, and are the basis for constructing the road network. Intersection feature points describe in detail the layout at road intersections, including intersection angles, lane allocations, etc., which are crucial for the navigation of autonomous vehicles at intersections. Points of interest cover important landmarks, service facilities, etc. along the way, providing rich navigation information for autonomous vehicles.

[0106] The construction of the auxiliary positioning element network aims to organically integrate these discrete road feature points, intersection feature points, and points of interest to form a coherent and accurate positioning reference framework. This network not only improves the accuracy and reliability of positioning information but also provides a more detailed description of the road environment for autonomous vehicles, helping to enhance the vehicle's path planning and decision - making capabilities. The relative height analysis is based on this network to further extract the relative height information of each auxiliary positioning element.

[0107] Finally, according to the relative height information of the auxiliary positioning elements, a grading process is carried out, which is the same as the height grading of the discrete correspondence model, and this embodiment will not elaborate here. Finally, according to the test statistical results, 0.2 meters is also selected as the grading interval. The auxiliary positioning elements only express the set - form information and shall not represent semantic content.

[0108] Step S505: Based on the road - surrounding expansion range, the relative height grading information of map elements, and the relative height grading information of auxiliary positioning elements, a comprehensive evaluation is carried out to obtain the review result of the high - precision map.

[0109] It should be noted that the road - surrounding expansion range can provide an adequate safety buffer area for autonomous vehicles, ensuring that the vehicle can timely identify and respond to obstacles around the road during driving. The relative height grading information of map elements is closely related to the obstacle recognition accuracy rate. Through reasonable grading, the obstacle recognition ability of autonomous vehicles can be improved. The relative height grading information of auxiliary positioning elements helps to ensure that autonomous vehicles can still accurately position in complex road environments. By integrating these information, a comprehensive evaluation of the high - precision map can be carried out to ensure that it meets the safety and accuracy requirements of autonomous driving. During the evaluation process, it is considered whether the road - surrounding expansion range is sufficient, whether the relative height grading of map elements is reasonable, and whether the relative height grading of auxiliary positioning elements is accurate. Only when these elements all meet the safety and accuracy requirements of autonomous driving can the high - precision map be determined to be qualified.

[0110] In one feasible implementation, step S505 may include: determining the road extension range deviation rate based on a preset road extension range compliance standard; performing obstacle recognition simulation and prediction based on the relative height classification information of the map elements to obtain the obstacle recognition accuracy rate; performing positioning accuracy simulation and prediction based on the relative height classification information of the auxiliary positioning elements to obtain the positioning accuracy rate; performing a comprehensive evaluation based on the road extension range deviation rate, the obstacle recognition accuracy rate, and the positioning accuracy rate to obtain a comprehensive evaluation score for the high-precision map; and comparing the comprehensive evaluation score with a preset map review passing score to obtain the map review result for the high-precision map.

[0111] It should be noted that the preset road extension compliance standards are typically determined based on traffic regulations, road design guidelines, and the safety requirements of autonomous vehicles. The road extension deviation rate measures the degree of difference between the actual road extension and the preset standard. This metric helps assess the accuracy of high-precision maps in representing road safety buffer zones.

[0112] Obstacle recognition simulation and prediction assesses a vehicle's ability to recognize obstacles by simulating various obstacle scenarios that autonomous vehicles may encounter during operation, using the relative height classification information of map features. This process can reveal the rationality of map feature classification and its impact on the safety of autonomous driving.

[0113] Positioning accuracy simulation and prediction, based on the relative height classification information of auxiliary positioning features, simulates the positioning process of autonomous vehicles in complex road environments to evaluate the vehicle's positioning accuracy. This step helps ensure that auxiliary positioning features in high-precision maps can provide reliable positioning references in practical applications.

[0114] The comprehensive evaluation score is derived by taking into account the deviation rate of road extension range, the accuracy rate of obstacle recognition, and the accuracy rate of positioning. It reflects the overall performance of the high-precision map in terms of safety, accuracy, and compliance. Comparing the comprehensive evaluation score with the preset passing score for map review allows for a direct assessment of whether the high-precision map meets the application requirements of autonomous driving, thus determining the map review result.

[0115] This embodiment comprehensively and accurately assesses the quality and compliance of high-precision maps by conducting multiple reviews, including comprehensive spatial range expansion review, relative height requirement review, and compliance expression review of auxiliary positioning elements. It then compares the comprehensive evaluation score with the preset map review passing score, thereby effectively improving the efficiency and accuracy of map review.

[0116] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the high-precision map data processing method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0117] This application also provides a high-precision map data processing device; please refer to [reference needed]. Figure 3 The high-precision map data processing device includes: The acquisition module 10 is used to acquire geographic information data from high-precision maps.

[0118] The standardization module 20 is used to standardize the geographic information data based on the spatiotemporal semantic perception adaptive standardization engine to obtain standardized geographic information data.

[0119] The classification and grading module 30 is used to perform horizontal classification of the standardized geographic information data based on an automatic clustering algorithm using multimodal feature extraction and vertical grading based on a knowledge graph reasoning algorithm using influence assessment, resulting in a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data at different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data.

[0120] The encryption and desensitization module 40 is used to encrypt and desensitize the hierarchical data set to obtain the encrypted and desensitized data set.

[0121] The review module 50 is used to review the spatial range expansion, relative height requirements, and compliance expression of auxiliary positioning elements of the high-precision map based on the encrypted and desensitized data set, and obtain the review result of the high-precision map.

[0122] The publishing module 60 is used to optimize the high-precision map data based on the encrypted and desensitized data set when the map review result is that the high-precision map has passed the review, and to publish the optimized high-precision map data to the map service platform.

[0123] The high-precision map data processing apparatus provided in this application, employing the high-precision map data processing method described in the above embodiments, can solve the technical problems of low efficiency, poor accuracy, and difficulty in ensuring the integrity and security of high-precision map data in traditional map review methods. Compared with the prior art, the beneficial effects of the high-precision map data processing apparatus provided in this application are the same as those of the high-precision map data processing method provided in the above embodiments, and other technical features in the high-precision map data processing apparatus are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0124] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A high-precision map data processing method, characterized in that, The method includes: Acquire geographic information data for high-precision maps; An adaptive standardization engine based on spatiotemporal semantic awareness standardizes the geographic information data to obtain standardized geographic information data. The standardized geographic information data is horizontally classified based on an automatic clustering algorithm using multimodal feature extraction and vertically graded based on a knowledge graph reasoning algorithm using influence assessment to obtain a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data sets of different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data. The hierarchical data set is encrypted and desensitized to obtain the encrypted and desensitized data set. Based on the encrypted and desensitized data set, the high-precision map is reviewed for spatial range expansion, relative height requirements, and compliance of auxiliary positioning elements, resulting in the map review results. If the high-precision map is approved in the review, the high-precision map data is optimized based on the encrypted and desensitized data set, and the optimized high-precision map data is published to the map service platform.

2. The method as described in claim 1, characterized in that, The adaptive standardization engine based on spatiotemporal semantic awareness standardizes the geographic information data to obtain standardized geographic information data, including: A unified spatiotemporal reference framework is constructed based on the global dynamic reference frame, the international Earth reference frame, and regional crustal deformation models. Based on the unified spatiotemporal reference framework, the geographic information data is spatiotemporally and semantically aligned to obtain spatiotemporally aligned data. The spatiotemporally aligned data is transformed using a coordinate system transformation algorithm to obtain the coordinate-transformed data. Construct a geographic element ontology knowledge base, and determine the entity standard attribute structure based on the geographic element ontology knowledge base; The data after coordinate system transformation is normalized according to the entity standard attribute structure to obtain the data after attribute semantic normalization. The data after the attribute semantics are normalized is encoded according to a preset hybrid encoding format to obtain standardized geographic information data.

3. The method as described in claim 1, characterized in that, The standardized geographic information data is horizontally classified using an automatic clustering algorithm based on multimodal feature extraction and vertically hierarchically classified using a knowledge graph reasoning algorithm based on influence assessment, resulting in a hierarchical data set, including: Feature extraction is performed on the standardized geographic information data based on a multimodal embedding model to obtain feature vectors for various types of data; The feature vectors of the various types of data are mapped to a unified semantic embedding space to obtain a unified semantic embedding vector; The unified semantic embedding vectors are clustered based on the HDBSCAN clustering algorithm to obtain data sets of different categories. Construct a high-precision map risk knowledge graph, wherein the nodes in the high-precision map risk knowledge graph include geographic elements, functional impacts, and security levels; Based on the high-precision map risk knowledge graph, the impact assessment algorithm is used to assess the impact of data in the different categories of datasets to obtain decision influence scores. Based on the decision influence score, a knowledge graph reasoning path is constructed, and the knowledge graph reasoning path is used to vertically classify the different categories of data sets to obtain a hierarchical data set.

4. The method as described in claim 1, characterized in that, The step of encrypting and de-identifying the hierarchical data set to obtain the encrypted and de-identified data set includes: The hierarchical data set is desensitized using a classification desensitization strategy to obtain different categories of data sets after desensitization. The classification desensitization strategies include a semantic segmentation-guided local fuzzy desensitization strategy for image data sets, an attribute differential privacy desensitization strategy for point cloud data sets, a spatiotemporal randomization desensitization strategy for trajectory data sets, a noise addition desensitization strategy for inertial navigation data sets, and a hash obfuscation desensitization strategy for mapping data sets. The data at different levels in the different categories of the de-identified data sets are encrypted using a hierarchical encryption strategy to obtain hierarchically encrypted data sets.

5. The method as described in claim 1, characterized in that, The process of reviewing the high-precision map based on the encrypted and desensitized data set, including spatial extent expansion review, relative height requirement review, and compliance review of auxiliary positioning element expression, yields the high-precision map review result, including: Based on the encrypted and desensitized data set, image data, point cloud data, trajectory data, inertial navigation data, and composition data are determined; Based on the image data, point cloud data, and inertial navigation data, a spatial range expansion review is performed to obtain the extended range around the road. Based on the point cloud data, trajectory data, and composition data, the relative height requirement is reviewed to obtain the relative height classification information of map elements; Based on the aforementioned compositional data, the compliance review of auxiliary positioning elements is conducted to obtain the relative height classification information of the auxiliary positioning elements; The high-precision map review result is obtained by comprehensively evaluating the surrounding extension range of the road, the relative height classification information of map elements, and the relative height classification information of auxiliary positioning elements.

6. The method as described in claim 5, characterized in that, The step of performing spatial range expansion verification based on the image data, point cloud data, and inertial navigation data to obtain the extended range around the road includes: Road images are extracted based on the image data, and the road images are subjected to inverse perspective transformation based on vanishing points to obtain corrected road images. The point cloud data is preprocessed using a radius filtering algorithm, and road surface point cloud data is extracted based on the preprocessed point cloud data. Road traffic markings are extracted based on the corrected road image and the road surface point cloud data; Vehicle speed data is extracted based on the inertial navigation data, and a speed-reaction time-environment regression model is constructed based on the vehicle speed data and the corresponding safety distance requirements. The reference distance matrix for the outward expansion of the road perimeter is determined based on the speed-reaction time-environment regression model. The extent of the road perimeter extension is determined based on the road traffic markings and the reference distance matrix of the road perimeter extension.

7. The method as described in claim 5, characterized in that, The step of reviewing relative height requirements based on the point cloud data, trajectory data, and composition data to obtain relative height classification information for map elements includes: A 3D scene model is constructed based on point cloud data, and the 3D scene model is then sliced ​​at different heights to obtain slice layers at different heights. The movement trajectory of the moving object is determined based on the trajectory data, and the movement trajectory is marked in the three-dimensional scene model; The location and shape of map elements are determined based on the composition data, and the map elements are marked in the three-dimensional scene model according to the location and shape. Analyze the vertical spatial relationship between map features and movement trajectories in the slice layer to obtain the relative height information between map features and movement trajectories; Based on the degree of continuity of the relative height information, a discrete object model and a continuous field model are established respectively; The relative height information of the discrete object model is subjected to a grading error analysis to obtain grading detail information; The relative height information of the continuous field model is decomposed by slope curvature to obtain spatial grid density information; The relative height classification information of map features is determined based on the classification detail information and spatial grid density information.

8. The method as described in claim 5, characterized in that, The step of reviewing the compliance of auxiliary positioning elements based on the composition data to obtain the relative height classification information of the auxiliary positioning elements includes: Extracting road feature points, intersection feature points, and points of interest from graph-based data; Based on the location information of road feature points, intersection feature points, and points of interest, an auxiliary positioning element network is constructed; Relative height analysis is performed based on the aforementioned auxiliary positioning element network to obtain the relative height information of the auxiliary positioning elements; The relative height information of the auxiliary positioning elements is processed to obtain the relative height classification information of the auxiliary positioning elements.

9. The method as described in claim 5, characterized in that, The process of comprehensively evaluating the road's surrounding extension range, the relative height classification information of map features, and the relative height classification information of auxiliary positioning features to obtain the high-precision map review result includes: The deviation rate of the road extension range is determined based on the preset road extension range compliance standards; Obstacle recognition simulation and prediction are performed based on the relative height classification information of the map elements to obtain the obstacle recognition accuracy. The positioning accuracy is simulated and predicted based on the relative height classification information of the auxiliary positioning elements to obtain the positioning accuracy rate; A comprehensive evaluation score for the high-precision map is obtained by comprehensively evaluating the road extension range deviation rate, the obstacle recognition accuracy rate, and the positioning accuracy rate. The comprehensive evaluation score is compared with the preset map review passing score to obtain the map review result of the high-precision map.

10. A high-precision map data processing device, characterized in that, The high-precision map data processing device includes: The acquisition module is used to acquire geographic information data from high-precision maps; The standardization module is used to standardize the geographic information data based on the spatiotemporal semantic awareness adaptive standardization engine to obtain standardized geographic information data. The classification and grading module is used to perform horizontal classification of the standardized geographic information data based on an automatic clustering algorithm using multimodal feature extraction and vertical grading based on a knowledge graph reasoning algorithm using influence assessment, resulting in a hierarchical data set. The hierarchical data set includes multiple data sets of different categories, and each category of data set includes multiple data sets of different levels. The different categories of data sets include image data sets, point cloud data sets, trajectory data sets, inertial navigation data sets, and mapping data sets. The different levels of data include core data, important data, and general data. An encryption and desensitization module is used to encrypt and desensitize the hierarchical data set to obtain an encrypted and desensitized data set. The review module is used to review the spatial range expansion, relative height requirements, and compliance of auxiliary positioning elements of the high-precision map based on the encrypted and desensitized data set, and obtain the review result of the high-precision map. The publishing module is used to optimize the high-precision map data based on the encrypted and desensitized data set when the map review result is that the high-precision map has passed the review, and then publish the optimized high-precision map data to the map service platform.

Images