Automatic driving global position estimation method and system based on standard definition electronic map static ground object association

Abstract

The invention discloses an automatic driving global position estimation method and system based on static ground object association of a standard-definition electronic map. The method comprises the following steps: acquiring the standard-definition electronic map in an automatic driving city operation scene; analyzing and preprocessing to extract a vehicle position candidate node set during automatic driving; static ground feature boundaries of the vehicle position candidate node set within a preset range are judged and screened, and urban structured static ground feature description information on a standard definition electronic map is extracted after screening to construct a prior static ground feature environment description database; the method comprises the following steps: sensing and inputting three-dimensional point cloud data in real time through a vehicle, and extracting urban structured static ground feature description to obtain real-time static ground feature environment description data; and performing similarity correlation on the static ground feature data in the standard definition electronic map and real-time static ground feature environment description data sensed by the vehicle end in real time to obtain a result with the highest similarity with the real-time sensed by the vehicle end so as to realize global position estimation of the automatic driving vehicle.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving global position estimation method and system based on static object association of a standard definition electronic map. Background Art

[0002] The navigation, decision-making and planning of autonomous driving rely on the accurate estimation of the vehicle's position on the map, so positioning is one of the key issues in the field of autonomous driving. In urban scenarios, in the face of positioning inaccuracies caused by satellite signal obstruction, multipath effects, etc., autonomous positioning methods based on pre-built high-precision maps are the current mainstream solution. However, in scenarios where pre-built high-precision map coverage is missing or poorly maintained, the classic "map first-position later" technical route faces the challenge of extreme performance degradation. Using standard-definition electronic maps to achieve global vehicle position estimation will provide a new route for existing "light map" autonomous driving technology, thereby improving the scope of application and functional integrity of autonomous driving vehicles.

[0003] First, standard-definition electronic maps contain road network information of urban structured scenes, which is suitable for the operation scenarios of most autonomous vehicles; second, standard-definition electronic maps contain static landform information of urban structured scenes, which has commonality and similarity with vehicle-side perception; finally, standard-definition electronic maps have a wide coverage, are easy to obtain, and have a high degree of standardization maturity. Therefore, standard-definition electronic maps can provide "map-positioning" technology for autonomous vehicles in the absence of pre-built high-precision maps. However, the accuracy and richness of information in standard-definition electronic maps are very different from the real-time perception modality of autonomous driving vehicles. How to solve cross-modal information association is the main problem facing this technology; in addition, there is a significant inconsistency between the freshness of information in standard-definition electronic maps and the real-time perception of the vehicle. How to avoid the impact of real-time dynamic information on the accuracy of vehicle-side perception-standard-definition electronic map information association is also the main problem facing this technology. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for estimating the global position of an autonomous driving system based on the association of static objects in a standard-definition electronic map. By parsing the environmental information of static objects in a standard-definition electronic map, a virtual point cloud with the same mode as that perceived by the vehicle is simulated and generated in the map position candidate. It is further proposed to use the directional change trend characteristics of static objects in the standard-definition electronic map as the basis for environmental description and matching association. By extracting the directional trend change characteristics of corresponding static objects in the real-time perception of the vehicle, the problem of vehicle global position estimation under conditions of significant differences in data accuracy, dimension, and mode faced by electronic standard-definition maps when used for autonomous driving vehicle positioning is solved.

[0005] The present invention provides a method for estimating the global position of an autonomous driving system based on static object association of a standard definition electronic map, comprising: Obtain standard-definition electronic maps for autonomous driving city operation scenarios; Parsing and preprocessing the standard definition electronic map to extract a set of candidate vehicle position nodes during autonomous driving; The static object boundaries within a preset range are judged and screened for the candidate node set of the vehicle position, and after the screening, the description information of the structured static objects in the city on the standard definition electronic map is extracted to construct a priori static object environment description database; Through real-time vehicle perception and real-time input of 3D point cloud data, the description of urban structured static objects is extracted to obtain real-time static object environment description data; The static object data in the standard-definition electronic map is correlated with the real-time static object environment description data perceived by the vehicle in real time, and the result with the highest similarity to the real-time perception of the vehicle is obtained to realize the global position estimation of the autonomous driving vehicle.

[0006] Preferably, the parsing and preprocessing of the standard definition electronic map includes: Obtaining road network data corresponding to the standard definition electronic map, and analyzing the connection relationship of nodes in the standard definition electronic map; By selecting road attributes to perform node filtering, roads that are passable by autonomous driving vehicles are selected as a vehicle drivable domain road set P', and the vehicle drivable domain road set P' includes several original road network node coordinates p' (x, y); For two adjacent road nodes p1′ (x1, y1) and p2′ (x2, y2) belonging to the same road, the plane geometric distance and the horizontal and vertical coordinate difference between the two road nodes are calculated respectively, and the number of nodes to be interpolated is determined according to the preset node interval; According to the difference between the horizontal coordinates of two adjacent road nodes X 差值 , the difference between the vertical coordinates Y 差值 and the number of nodes to be interpolated, calculate the coordinate increment p3′ (X 增量， Y 增量 ); According to the coordinate difference between the two road nodes, the number of nodes to be interpolated and the corresponding coordinate increment, the interpolation points are equally spaced horizontally and vertically between the two road nodes, and each original road network node in the drivable road set P' in the standard definition electronic map is traversed and interpolated to form a set of candidate vehicle position nodes in the electronic map. ; Among them, X 差值 =x2-x1,Y 差值 =y2-y1,X 增量 =X 差值 / Number of nodes to be interpolated, Y 增量 = Y 差值 / Number of nodes to be interpolated, N represents the number of nodes in the road network after interpolation; p1, p2, ...p N Represents the plane node coordinates corresponding to the 1st, 2nd, ...Nth road nodes.

[0007] Preferably, the parsing and preprocessing of the standard definition electronic map includes: Obtaining the road network data corresponding to the standard definition electronic map, parsing the connection relationship of the nodes in the standard definition electronic map, and extracting all the static feature boundary information nodes in the parsed standard definition electronic map; Filter nodes by selecting boundary attributes to select target static objects with boundary, wherein the boundary attributes include buildings, fences, barriers and road-related boundary objects; Divide according to the same attributes of the boundary and construct a static object set composed of static ground feature boundary points ; Where K is the number of static feature boundaries contained in the standard definition electronic map, and the element Bj in the set B represents the set of nodes of static features corresponding to the same attribute, j=1, 2, ..., K; ; , ,... Represents the two-dimensional coordinates of the 1st, 2nd, ...nth nodes in the static feature in the electronic map.

[0008] Preferably, the determining and screening of the static object boundaries within a preset range of the vehicle position candidate node set includes: Traverse the set of candidate vehicle position nodes P, take the current candidate vehicle position node as the center of the circle in the geographic coordinate system of the standard definition electronic map, and use the preset prior visual range distance parameter R P , with R P The static features within the radius are screened and truncated; Calculate the distance between the current vehicle position candidate node and all the static object corresponding nodes to obtain a distance matrix, wherein each row of the distance matrix represents a vehicle position candidate node, each column represents a static object corresponding node, and the elements in the distance matrix represent the target distance from the vehicle position candidate node to the static object corresponding node; Determine whether the target distance in the distance matrix is ​​greater than the static object node within the preset prior visible range. When the target distance is greater than the static object node within the preset prior visible range, screen the static object corresponding nodes corresponding to the target distance and perform temporary filtering in the current state to form a local prior candidate set formed by a single vehicle global position candidate node and a boundary within the prior visible range.

[0009] Preferably, extracting the city structured static land object description information on the standard definition electronic map after screening to construct a priori static land object environment description database includes: In the geographic coordinate system of the standard definition electronic map, the candidate node of the current vehicle global position is taken as the center of the circle, and the same angle interval is taken. The ray of the current vehicle global position candidate node is segmented to form a plurality of segmented areas arranged in a counter-clockwise order, and the area close to the boundary of the current vehicle global position candidate node is scanned; Calculate the points where each angle ray intersects with the boundary line of the static object in turn, and take the plane coordinates corresponding to the point as the discrete point cloud coordinates in the segmented area to generate a discrete static object boundary point cloud set , and the candidate coordinates of the vehicle global position Have a unique corresponding relationship; Calculate the direction vectors between discrete static object boundary point clouds in sequence And the preset observation direction vector , and calculate the angle between the current boundary direction vector and the preset observation direction vector The one-dimensional vector formed is used as a priori static feature boundary representation To achieve the extraction of the boundary direction trend of static objects in the standard definition electronic map scene; Among them, the current direction vector is calculated by the following method: ; in, Indicates the number of columns of the vector, i.e. the number of plane splitting rays, and Respectively represent the position coordinates of the nth and n-1th discrete static object boundary point clouds, Indicates the direction vector from the n-1th discrete static object boundary point cloud to the nth discrete static object boundary point cloud, x i and i For vector The starting point coordinates, x n and n For vector The end point coordinates are also the position coordinates of the nth discrete static feature boundary point cloud. Preferably, the real-time perception and real-time input of three-dimensional point cloud data by the vehicle to extract the description of the structured static features in the city and obtain the real-time static feature environment description data include: Taking the current vehicle observation point as the center, based on the 3D point cloud data input by the vehicle-side real-time perception, the data is screened according to the same attribute categories as the static objects in the prior standard definition electronic map, and a real-time static object boundary 3D point cloud set S with the same attributes is obtained. The 3D point cloud set S is converted into 2D coordinates by XOY plane projection in the vehicle body coordinate system, and an unordered plane point cloud set S′ containing only x-coordinates and y-coordinates is obtained. Centered on the current vehicle observation point, take the same angle interval The ray divides the XOY plane of the current vehicle coordinate system into multiple divided areas arranged in a counter-clockwise order; For each point cloud in the segmented area, the plane average center of the point cloud set in the segmented area is solved, and the coordinates of the average center are used as the equivalent point cloud in the area, and finally the real-time static feature boundary plane equivalent input point cloud set is obtained. ; Calculate the direction vectors between the equivalent input point clouds of the real-time static object boundary plane in sequence And the corresponding observation direction vector , and calculate the angle between the current boundary direction vector and the observation direction vector , forming a one-dimensional vector As a real-time static object boundary environment representation, it can realize the extraction of static object boundary direction trend in the real-time perception scenario on the vehicle side; Among them, the current direction vector is calculated by the following method: ; in, Indicates the number of columns of the vector, i.e. the number of plane splitting rays, and Respectively represent the position vector coordinates of the nth and n-1th real-time point clouds, Indicates the direction vector from the n-1th real-time static object boundary equivalent input point cloud to the nth real-time static object boundary equivalent input point cloud, x n and n For vector The vector coordinates in the XOY plane of the vehicle coordinate system centered on the current vehicle observation point are also the position coordinates of the equivalent input point cloud of the static object boundary at the nth time.

[0010] Preferably, the method of associating the static object data in the standard definition electronic map with the real-time static object environment description data perceived by the vehicle in real time to obtain the result with the highest similarity to the real-time perception of the vehicle to achieve global position estimation of the autonomous driving vehicle includes: By traversing each element in the prior static object environment description database, the most similar result to the real-time perception is queried in the prior static object environment description database, and the similarity calculation is implemented using the n-norm method as follows: , Where D(i) represents the similarity between the input static object environment description vector perceived in real time and the i-th candidate in the prior static object environment description database. Represents the reference vector corresponding to the real-time perceived static feature description, represents the i-th static object description vector corresponding to each vehicle global position candidate in the database, and N is the total number of static object environment description vectors in the prior static object environment description database; The calculated similarity results are compared to select the candidate result with the highest similarity to the input perception vector, and the vehicle's global position estimate is obtained by accessing the candidate coordinates of the vehicle's global position corresponding to the optimal candidate result.

[0011] The present invention also provides an automatic driving global position estimation system based on static object association of a standard definition electronic map, comprising: Data acquisition module, used to obtain standard-definition electronic maps in the scenario of autonomous driving city operation; A parsing and processing module, used for parsing and preprocessing the standard definition electronic map to extract a set of candidate vehicle position nodes during automatic driving; An electronic map extraction module is used to determine and screen the static feature boundaries within a preset range for the vehicle position candidate node set, and after screening, extract the city structured static feature description information on the standard definition electronic map to construct a priori static feature environment description database; A real-time perception and extraction module is used to extract the description of urban structured static objects through real-time perception of vehicles and input three-dimensional point cloud data in real time, so as to obtain real-time static object environment description data; The position estimation module is used to associate the static object data in the standard-definition electronic map with the real-time static object environment description data perceived by the vehicle side in real time, and obtain the position candidate result with the highest similarity to the real-time perception of the vehicle side to realize the global position estimation of the autonomous driving vehicle.

[0012] The present invention also provides an electronic device, comprising: A memory, the memory being used to store a processing program; A processor, when executing the processing program, implements the automatic driving global position estimation method based on standard definition electronic map static object association as described in an embodiment of the present invention.

[0013] The present invention also provides a readable storage medium, on which a processing program is stored. When the processing program is executed by a processor, the method for estimating the global position of an autonomous driving system based on the association of static objects in a standard-definition electronic map as described in an embodiment of the present invention is implemented.

[0014] With respect to the prior art, the present invention has the following beneficial effects: The present invention proposes a method for estimating the global position of an autonomous driving system based on the association of static objects in a standard-definition electronic map. The present invention analyzes the environmental information of static objects in a standard-definition electronic map, simulates and generates a virtual point cloud in the same mode as the vehicle-side perception in the map position candidate, and further proposes to use the directional change trend characteristics of static objects in the standard-definition electronic map as the basis for environmental description and matching association. By extracting the directional trend change characteristics of the corresponding static objects in the real-time perception of the vehicle, the problem of estimating the global position of the vehicle under the conditions of significant differences in data accuracy, dimension, and mode faced by the electronic standard-definition map for the positioning of autonomous driving vehicles is solved. The invention has the following significant advantages and effects: 1. Get rid of the reliance on pre-built high-precision maps The existing "mapping-positioning" technology route for autonomous driving relies on the accurate description of the static terrain environment provided in the pre-built high-precision map. The key to achieving positioning lies in the matching and registration of real-time perception and high-precision map elements. The method of the present invention directly obtains static terrain environment information from the standard-definition electronic map, and can be associated with the real-time perception of the vehicle to achieve global position estimation, without the need for pre-built high-precision maps.

[0015] 2. Wide range of applications The existing "mapping-positioning" technology route for autonomous driving relies on pre-built high-precision maps, but the coverage of pre-built high-precision maps is limited. Standard-definition electronic maps have a wide coverage and are easy to obtain. For areas where pre-built maps are not fully covered and maintenance is lagging, the method of the present invention can effectively support the global positioning needs of autonomous driving in areas where pre-built maps are missing.

[0016] 3. Reduce costs The positioning route based on pre-built high-precision maps needs to maintain the freshness of map information at a high frequency to avoid positioning performance degradation. However, the existing map maintenance requires a high-precision map data collection vehicle and relies on manual annotation, with a low degree of automation. The positioning technology route of the present invention does not rely on pre-built high-precision maps, thereby reducing the cost of using the positioning solution.

[0017] In summary, the present invention proves through theoretical analysis and experimental data that standard-definition electronic maps have the potential and technical feasibility to replace pre-built high-precision maps in urban structured scenarios in terms of positioning. At the same time, it significantly improves the accuracy of vehicle global position estimation provided by using standard-definition electronic maps, reduces the promotion cost of the "mapping-positioning" technical route of autonomous driving, and has important practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of the steps of a method for estimating a global position of an autonomous driving system based on association of static objects on a standard definition electronic map according to an embodiment of the present invention; Figure 2 It is a principle block diagram of an automatic driving global position estimation system based on standard definition electronic map static object association in one embodiment of the present invention; Figure 3 A schematic diagram of a priori static terrain environment description in one embodiment of the present invention; Figure 4 It is a schematic diagram of real-time static terrain environment description in one embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0021] It should be noted that the concepts such as "once" and "second" mentioned in the present application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0022] It should be noted that the modifications of "one" and "plurality" mentioned in the disclosure of this application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0023] Embodiment 1 like Figure 1-4 As shown, the present invention also provides an automatic driving global position estimation method based on standard definition electronic map static object association, which is applicable to an automatic driving global position estimation system based on standard definition electronic map static object association. The system consists of the following main modules: standard definition electronic map parsing and preprocessing module, urban structured static object description extraction module for standard definition electronic map, and urban structured static object description extraction module for real-time perception. The method includes the following steps: S1: Obtain a standard definition electronic map for the autonomous driving city operation scenario; S2: parsing and preprocessing the standard definition electronic map to extract a set of candidate vehicle position nodes during autonomous driving, including three steps: electronic map road network parsing, electronic map road network interpolation, and static object boundary parsing, as detailed below: 1. Electronic map road network analysis: Obtain the road network data corresponding to the standard definition electronic map, and analyze the connection relationship of the nodes in the standard definition electronic map. The information in the standard definition electronic map is composed of points, lines and polygons. The east direction in the electronic map is the positive direction of the x-axis, and the north direction is the positive direction of the y-axis. The drivable road set P′ is composed of several original road network node coordinates p′ (x, y); By selecting road attributes to filter nodes, i.e., using attributes such as road grade and type in the map, roads that are passable by autonomous driving vehicles are selected as a set of vehicle drivable road domains P′, wherein the set of vehicle drivable road domains P′ includes several original road network node coordinates p′ (x, y); 2. Electronic map road network interpolation: Since the original road network nodes are relatively sparse, for example, the nodes in the electronic map are very unevenly distributed. For example, if a road is defined as 100 meters, there are only 3 nodes, and if a road is defined as 10 meters, there are 5 nodes. Interpolation operations are used to ensure the uniformity of node distribution on each road. Specifically, by analyzing the set of drivable roads with vehicle-accessible vehicle domains, for two adjacent road nodes p1′ (x1, y1) and p2′ (x2, y2) belonging to the same road, the plane geometric distance and the horizontal and vertical coordinate difference between the two road nodes are calculated respectively, and the number of nodes to be interpolated is determined according to the preset node interval and rounded; According to the difference between the horizontal coordinates of two adjacent road nodes X 差值 , the difference between the vertical coordinates Y 差值 and the number of nodes to be interpolated, calculate the coordinate increment p3′ (X 增量， Y 增量 ); According to the coordinate difference between the two road nodes, the number of nodes to be interpolated and the corresponding coordinate increment, the interpolation points are equally spaced horizontally and vertically between the two road nodes. The above method is used to traverse and interpolate each original road network node in the set P' of all drivable roads in the standard definition electronic map to form a set of candidate nodes for the global position of the vehicle in the electronic map. ; Among them, X 差值 =x2-x1,Y 差值 =y2-y1,X 增量 =X 差值 / Number of nodes to be interpolated, Y 增量 = Y 差值 / Number of nodes to be interpolated, N represents the number of road nodes after interpolation in the electronic map; p1, p2, ... p N Represents the plane node coordinates corresponding to the 1st, 2nd, ...Nth road nodes. The above method can effectively extract and process road network data from standard definition electronic maps to provide accurate road information for autonomous driving vehicles. In the face of different position estimation needs, the interval can be determined on demand. In addition, the parameters of the interval can be determined according to the size of the scene. The smaller the interval, the larger the number of databases to be matched. By interpolating the original road network nodes based on road curvature, the density of the road network can be effectively increased, thereby providing more accurate and rich navigation information for autonomous driving vehicles.

[0024] In this embodiment, the static feature boundary analysis also includes the analysis of the static feature boundary of the map, which is described in detail as follows: Obtaining the road network data corresponding to the standard definition electronic map, parsing the connection relationship of the nodes in the standard definition electronic map, and extracting all the static feature boundary information nodes in the parsed standard definition electronic map; Filter nodes by selecting boundary attributes to select target static objects with boundary, wherein the boundary attributes include buildings, fences, barriers and road-related boundary objects; Divide according to the same attributes of the boundary and construct a static object set composed of static ground feature boundary points ; Where K is the number of static feature boundaries contained in the standard definition electronic map, and the element Bj in the set B represents the set of nodes of static features corresponding to the same attribute, j=1, 2, ..., K; ; , ,... Represents the plane two-dimensional coordinates of the 1st, 2nd, ...nth nodes in the static feature in the electronic map. By accurately parsing the node connection relationship in the standard definition electronic map, it can ensure that the extracted static feature boundary information is accurate and provide a reliable basis for subsequent processing; using boundary attributes to filter and divide nodes can effectively distinguish different types of static feature boundaries and improve the efficiency and accuracy of data processing; when constructing a static object set, dividing them according to the same attributes of the boundary helps to simplify the data structure and facilitate subsequent analysis and processing; by clearly representing the plane two-dimensional coordinates of each node in the static feature in the electronic map, it can provide accurate environmental perception information for the automatic driving system and improve driving safety and stability.

[0025] S3: The static feature boundaries of the vehicle position candidate node set within a preset range are judged and screened, and after screening, the city structured static feature description information on the standard definition electronic map is extracted to construct a priori static feature environment description database; this step includes a visible static feature boundary screening step, which is described in detail as follows: Traverse the set of candidate nodes P for the global position of the vehicle, and in the geographic coordinate system of the standard definition electronic map, such as the UTM / WGS84 coordinate system, take the current candidate node of the global position of the vehicle as the center of the circle, and use the preset prior visual range distance parameter R P , with R P The static object boundary within the radius is screened and truncated; it can be understood that the visible static object boundary screening step uses the vehicle global position candidate node set P, and introduces the prior visual range distance parameter R P , with R P Screening and truncating the boundaries of static objects within the radius to form a local priori candidate set formed by a single candidate position node and the boundaries within the visible range; or, preferably, screening out the nearest boundary point in each direction by simulating light occlusion characteristics; In order to remove the static object boundary nodes outside the visible range, the distance between the current vehicle global position candidate node and all static object corresponding nodes is calculated to obtain a distance matrix, in which each row of the distance matrix represents a vehicle global position candidate node, each column represents a static object corresponding node, and the elements in the distance matrix represent the target distance from the vehicle global position candidate node to the static object corresponding node; Determine whether the target distance in the distance matrix is ​​greater than the static object node within the preset prior visible range. When the target distance is greater than the static object node within the preset prior visible range, screen the static object corresponding nodes corresponding to the target distance and perform temporary filtering in the current state to form a local prior candidate set formed by a single vehicle global position candidate node and a boundary within the prior visible range. The temporary filtering is to generate a static object environment description corresponding to the position candidate node.

[0026] In this embodiment, the step S3 extracts the description information of the structured static features on the standard definition electronic map after screening to construct a priori static feature environment description database, further comprising: generating a discrete point cloud of the plane static feature boundary and extracting the direction trend of the static feature boundary, which are described in detail as follows: In the geographic coordinate system, take the candidate node of the current vehicle global position as the center and take the same angle interval The plane where the candidate node of the current vehicle global position is located is divided into multiple segmented areas arranged in a counter-clockwise order by the ray, and the area near the boundary of the candidate node of the current vehicle global position is scanned; for example, after the visible boundary is screened, the candidate node is taken as the center and the same angle interval is taken. The plane is segmented by the ray, and the boundary near the candidate point is scanned 360° counterclockwise, with the candidate position as the dot and radiating outward. The intersection point when the ray intersects with the boundary of the static object for the first time is the visible observation point. The segmentation operation in the above and below embodiments is to simulate the scanning mode of the laser radar. The static objects such as roads and buildings in the standard definition electronic map are described by polygons composed of discrete points, which is very different from the point cloud mode. In order to unify the representation method of the vehicle laser radar point cloud, the plane needs to be segmented and the corresponding calculation is performed.

[0027] Planar static feature boundary discrete point cloud generation: Calculate the points where each angle ray intersects with the static feature boundary line in turn, and take the plane coordinates corresponding to the point as the discrete point cloud coordinates within the segmented area to generate a discrete static feature boundary point cloud set , and the candidate coordinates of the vehicle global position There is a unique correspondence relationship, that is, each global position candidate coordinate has a unique corresponding discrete static object boundary point cloud set; Static object boundary direction trend extraction: For a given plane virtual point cloud set s′ and its corresponding vehicle global position candidate , at equal angles , counterclockwise is positive, starting from the positive direction of the x-axis, and sequentially calculate the direction vectors between the discrete static object boundary point clouds And the preset observation direction vector , and calculate the angle between the boundary direction vector and the preset observation direction vector The preset observation direction here is the ray direction, that is, the observation direction of the laser in the process of simulating laser point cloud scanning. The one-dimensional vector formed is used as the prior static feature boundary expression ; The direction vector is calculated as follows: ; in, The number of columns representing the vector is the number of plane segmentation rays, that is, the number of sectors into which the plane is segmented = 360 degrees of the circumference divided by the ray angle interval And round up to the integer, and Respectively represent the position coordinates of the nth and n-1th discrete static object boundary point clouds, Indicates the direction vector from the n-1th discrete static feature boundary point cloud to the nth discrete static feature boundary point cloud, that is, the boundary direction vector corresponding to the nth ray. The direction vector is calculated from the point cloud coordinates corresponding to the n-1th ray to the point cloud coordinates corresponding to the nth ray. i and i For vector The starting point coordinates, x n and n For vector The end point coordinates are also the position coordinates of the nth discrete static feature boundary point cloud, where the number of rays = the number of plane point clouds = the number of direction vectors.

[0028] The principle of this embodiment is to use polygons and lines in the standard definition electronic map to generate a virtual point cloud with the same mode as the vehicle-side perception, and use the boundary direction extracted from the virtual point cloud to solve the problem of cross-modal information association. The vehicle-side perception itself is a point cloud, and the boundary direction can be directly extracted. The specific implementation steps are as follows: S4: Through real-time vehicle perception and real-time input of 3D point cloud data, the description of urban structured static objects is extracted to obtain real-time static object environment description data, including: 1. Plane projection of real-time static feature boundary point cloud: With the current vehicle observation point as the center, the vehicle forward direction is defined as the positive direction of the y-axis, the right side of the vehicle is defined as the positive direction of the x-axis, and the top of the vehicle is defined as the positive direction of the z-axis to construct the vehicle body coordinate system. The dense 3D point cloud input by the vehicle-side real-time perception is filtered according to the same attribute categories as the static features in the prior standard definition electronic map to obtain the real-time static feature boundary 3D point cloud set. Through XOY plane projection, the elevation information of the point cloud is filtered to obtain an unordered plane point cloud set without elevation. ; 2. Real-time static ground plane equivalent point cloud generation: With the current vehicle observation point as the center, counterclockwise as positive, take the same angle interval The ray of the vehicle divides the plane where the candidate node of the current vehicle global position is located to form multiple segmented areas arranged in a counter-clockwise order; For each point cloud in the segmented area, the plane average center of the point cloud set in the segmented area is solved, and the coordinates of the average center are used as the equivalent point cloud in the area, and finally the real-time static feature boundary plane equivalent input point cloud set is obtained. ; 3. Real-time static feature boundary direction trend extraction: at the same angle interval , taking counterclockwise as positive, calculate the direction vectors between the equivalent input point clouds of the real-time static object boundary plane in sequence And the corresponding observation direction vector , and calculate the angle between the current boundary direction vector and the observation direction vector The observation direction here is the measurement coordinate returned by the laser sensor, which is the direction, forming a one-dimensional vector As a real-time static feature boundary environment representation; Among them, the current direction vector is calculated by the following method: ; in, Indicates the number of columns of the vector, i.e. the number of plane splitting rays, and Respectively represent the position vector coordinates of the nth and n-1th real-time point clouds, vector Indicates the direction vector from the n-1th real-time static object boundary equivalent input point cloud to the nth real-time static object boundary equivalent input point cloud, x n and n For vector The vector coordinates in the XOY plane of the vehicle coordinate system centered on the current vehicle observation point are also the position coordinates of the equivalent input point cloud of the static object boundary at the nth time.

[0029] S5: Perform similarity correlation between the static object data in the standard definition electronic map and the real-time static object environment description data perceived by the vehicle side, and obtain the result with the highest similarity to the real-time perception of the vehicle side to achieve global position estimation of the autonomous driving vehicle. This step is implemented by the vehicle global position estimation system, which includes a priori static object environment description database generated by the standard definition electronic map, in which each element is composed of 1x dimensional row vectors, corresponding to each candidate vehicle position in the standard definition electronic map; the real-time static terrain environment description is generated by the 1x The row vectors of dimension By traversing each element in the prior static object environment description database, the most similar result to the real-time perception is queried in the prior static object environment description database, and the similarity calculation is implemented using the n-norm method as follows: , Where D(i) represents the similarity between the input vector and the i-th candidate in the prior static feature environment description database. Represents the reference vector corresponding to the real-time perceived static feature description, represents the i-th static feature description vector corresponding to each vehicle global position candidate in the database, N is the total number of vectors in the database. Generally speaking, n=1 or n=2 is taken, that is, the 1-norm or 2-norm is used to calculate the vector similarity. The method of this embodiment is effective under different values ​​and only has a slight impact on the calculation efficiency.

[0030] The calculated similarity results are compared to select the candidate result with the highest similarity to the input perception vector, and the vehicle's global position estimate is obtained by accessing the candidate coordinates of the vehicle's global position corresponding to the optimal candidate result. Through the above specific implementation, the present invention can use the static object information in the standard definition electronic map to associate with the vehicle's real-time perception to achieve global position estimation of the autonomous driving vehicle, and get rid of the dependence on pre-built high-precision maps.

[0031] The embodiment of the present invention analyzes the environmental information of static objects in the standard-definition electronic map, simulates and generates a virtual point cloud with the same mode as the vehicle-side perception in the map position candidate, and further proposes to use the directional change trend characteristics of static objects in the standard-definition electronic map for environmental description and matching association basis, and extracts the directional trend change characteristics of the corresponding static objects in the real-time perception of the vehicle, thereby solving the problem of vehicle global position estimation under the conditions of significant differences in data accuracy, dimension, and mode faced by the electronic standard-definition map for autonomous driving vehicle positioning. In the experiment, the technical solution proposed in the embodiment of the present invention was compared with the existing method on the autonomous driving data set of urban structured scenes. The results show that in the urban scene sequence of KITTI, the present invention has achieved better performance in both the positioning success rate within 2 meters and the positioning success rate within 5 meters. And under the conditions of different confidence levels in the acquisition of real-time boundary point clouds, the embodiment of the present invention shows higher robustness and accuracy.

[0032] The specific experimental data are as follows: In the typical sequence of KITTI urban structured scenes, when using a large-scale standard-definition electronic map (including 11,612 position candidates), the 2-meter positioning success rate is 43.27%, and the 5-meter positioning success rate is 53.20%; when using a small-scale standard-definition electronic map (including 4,541 position candidates), the 5-meter positioning success rate is 57.8%. It is significantly better than other existing positioning methods.

[0033] In summary, the embodiments of the present invention have proved through theoretical analysis and experimental data that standard-definition electronic maps have the potential and technical feasibility to replace pre-built high-precision maps in urban structured scenarios in terms of positioning. At the same time, they significantly improve the accuracy of vehicle global position estimation provided by standard-definition electronic maps, reduce the promotion cost of the "mapping-positioning" technical route of autonomous driving, and have important practical application value.

[0034] Embodiment 2 Based on the same concept, the present invention provides an automatic driving global position estimation system based on static object association of standard definition electronic map, comprising: Data acquisition module, used to obtain standard-definition electronic maps in the scenario of autonomous driving city operation; The parsing and processing module is a standard definition electronic map parsing and preprocessing module, which is used to parse and preprocess the standard definition electronic map to extract a set of candidate vehicle position nodes during automatic driving; The electronic map extraction module is a city structured static feature description extraction module for standard definition electronic maps, which is used to judge and screen the static feature boundaries within a preset range of the vehicle position candidate node set, and after screening, extract the city structured static feature description information on the standard definition electronic map to construct a priori static feature environment description database; The real-time perception extraction module is a city structured static feature description extraction module for real-time perception, which is used to extract the city structured static feature description by real-time input of three-dimensional point cloud data through real-time perception of the vehicle, and obtain real-time static feature environment description data; The position estimation module is used to associate the static object data in the standard definition electronic map with the real-time static object environment description data perceived by the vehicle side in real time, and obtain the position candidate result with the highest similarity to the real-time perception of the vehicle side to achieve the global position estimation of the autonomous driving vehicle. The system is described as follows: 1. Standard definition electronic map analysis and preprocessing module 1.1 Electronic map road network analysis: The information in the standard-definition electronic map is composed of points, lines and polygons. By analyzing the connection relationship of the nodes in the standard-definition electronic map and filtering using the attributes such as road grade and type in the map, roads that are theoretically passable for autonomous vehicles are retained as the set of roads in the vehicle's drivable domain.

[0035] 1.2 Electronic map road network interpolation: By parsing the set of drivable roads in the vehicle's drivable domain that are passable by vehicles, for two adjacent road nodes belonging to the same road, the plane geometric distances and the horizontal and vertical coordinate interpolation of the two nodes are calculated respectively, the number of nodes to be interpolated is calculated according to the preset road node interval and rounded, and the interpolation points to be interpolated are distributed at equal distances in the horizontal and vertical directions between the two road nodes to form a set of candidate nodes for the global position of the vehicle in the electronic map.

[0036] 1.3 Analysis of static feature boundaries: By parsing all the static feature boundary information nodes in the standard definition electronic map, filtering by boundary attributes (such as buildings, fences, barriers, etc.), retaining the static feature boundaries with long-term validity; secondly, further dividing according to the homogeneous properties of the boundaries, constructing a static object set composed of static feature boundary points.

[0037] 2. Urban structured static feature description extraction module for standard definition electronic maps 2.1 Visual boundary screening: Using the vehicle global position candidate node set generated by module 1.2, the static object boundaries within the visible range of the candidate node are filtered out by presetting the prior visible range distance parameter, forming a local prior candidate set consisting of a single candidate position node and the boundaries within the visible range; 2.2 Plane boundary discrete point cloud generation: After the visible boundary screening, with the candidate position node as the center, the map east direction as the x-axis, the map north direction as the y-axis, and the counterclockwise direction as the positive, the plane is segmented by taking the rays with the same preset angle interval, and the boundary near the candidate point is scanned 360° counterclockwise, and the plane coordinates of the points where the scanning rays intersect the static feature boundary line are calculated in sequence as the point cloud coordinates in the segmented area, and then a priori local boundary discrete point cloud set is generated, and the corresponding local plane point cloud virtual observation data set is formed with the vehicle candidate position coordinates; 2.3 Extraction of static feature boundary direction trend: Based on the plane virtual point cloud observation generated by module 2.2, the direction vectors between the discrete static object boundary point clouds and the preset observation direction vector are calculated in a counterclockwise order starting from the positive direction of the x-axis at the same angular intervals as in 2.2, and the angle between the boundary direction vector and the preset observation direction vector is further calculated. The formed one-dimensional vector is used as the prior static object environment description.

[0038] 3. Urban structured static feature description extraction module for real-time perception 3.1 Real-time static object boundary point cloud plane projection: The dense 3D point cloud input by the vehicle-side real-time perception is filtered according to the categories with the same attributes as the static feature boundaries in the prior map, such as buildings, fences, retaining walls, etc., to obtain a real-time boundary 3D point cloud set. With the vehicle observation point as the center, the front of the vehicle is defined as the positive y-axis, the right side of the vehicle is defined as the positive x-axis, and the top of the vehicle is defined as the positive z-axis. Through XOY plane projection, the Z-axis information of the point cloud is filtered to obtain an unordered plane point cloud set without elevation; 3.2 Real-time static ground plane equivalent point cloud generation: With the vehicle observation point as the center, the vehicle front is defined as the positive direction of the y-axis, the right side of the vehicle is defined as the positive direction of the x-axis, and the counterclockwise direction is positive. The plane area is segmented at the same angle interval. For each plane point cloud in the segmented area, the center point is determined by taking the average value of the coordinates of the point cloud set in the area in the x-axis and y-axis directions, and the coordinates of the center point are used as the equivalent point cloud in the area. The number of real-time plane segmentations and the angle interval are the same as those in module 2.2; 3.3 Real-time static feature boundary direction trend extraction: The direction vectors between discrete boundary point clouds and the corresponding observation direction vectors are calculated in sequence with the same angle interval and counterclockwise as positive, and the angle between the boundary direction vector and the observation direction vector is further calculated. The formed one-dimensional vector is used as a real-time static ground object environment representation.

[0039] 4. Implementation of vehicle global position estimation system The system includes a priori static feature description database generated by a standard definition electronic map, wherein each element in the database is composed of a number of row vectors corresponding to each candidate vehicle position in the standard definition electronic map, and the number of database elements depends on the size of the standard definition electronic map used, and there is no specific number limit; the real-time static feature environment description is composed of static feature description vectors generated corresponding to each frame of point cloud input; by traversing each element in the priori static feature description database, a result that is most similar to real-time perception can be found in the priori static feature description database, and by accessing the position coordinates corresponding to the result, the global position estimate of the vehicle can be obtained.

[0040] It should be noted that the division of the modules in the embodiment of the device / system is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or physically separated. Moreover, these modules can be implemented in the form of software calling through processing elements; or in the form of hardware; or some units can be implemented in the form of software calling through processing elements, and some units can be implemented in the form of hardware.

[0041] The implementation principles of the above modules have been described in the above embodiments, so they will not be repeated here.

[0042] Embodiment 3 Based on the same concept, in some embodiments of the present application, an electronic device is also provided. The electronic device includes a memory and a processor, wherein the memory is used to store a processing program, and the processor executes the processing program according to instructions. When the processor executes the processing program, the automatic driving global position estimation method based on the association of static objects in the standard definition electronic map in the above embodiment is implemented.

[0043] In some embodiments of the present application, a readable storage medium is also provided, which may be a non-volatile readable storage medium or a volatile readable storage medium. The readable storage medium stores instructions, and when the instructions are executed on a computer, an electronic device including the readable storage medium executes the aforementioned automatic driving global position estimation method based on static object association of a standard definition electronic map.

[0044] It is understandable that, for the aforementioned automatic driving global position estimation method based on the association of static objects on standard definition electronic maps, if they are all implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., and other media that can store program codes.

[0045] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0046] The program code involved in executing the technical solution disclosed in this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Python, C++, etc., and also conventional procedural programming languages ​​such as C language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for global position estimation of autonomous driving based on static object association of standard definition electronic map, characterized in that: include: Obtain standard-definition electronic maps for autonomous driving city operation scenarios; Parsing and preprocessing the standard definition electronic map to extract a set of candidate vehicle position nodes during autonomous driving; The static object boundaries within a preset range are judged and screened for the candidate node set of the vehicle position, and after the screening, the description information of the structured static objects in the city on the standard definition electronic map is extracted to construct a priori static object environment description database; Through real-time vehicle perception and real-time input of 3D point cloud data, the description of urban structured static objects is extracted to obtain real-time static object environment description data; The static feature data in the standard-definition electronic map is similarly associated with the real-time static feature environment description data perceived by the vehicle side, and the position candidate result with the highest similarity to the real-time perception of the vehicle side is obtained to realize the global position estimation of the autonomous driving vehicle.

2. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The parsing and preprocessing of the standard definition electronic map includes: Obtaining road network data corresponding to the standard definition electronic map, and analyzing the connection relationship of nodes in the standard definition electronic map; By selecting road attributes to perform node filtering, roads that are passable by autonomous driving vehicles are selected as a vehicle drivable domain road set P', and the vehicle drivable domain road set P' includes several original road network node coordinates p' (x, y); For two adjacent road nodes p1′ (x1, y1) and p2′ (x2, y2) belonging to the same road, the plane geometric distance and the horizontal and vertical coordinate difference between the two road nodes are calculated respectively, and the number of nodes to be interpolated is determined according to the preset node interval; According to the difference between the horizontal coordinates of two adjacent road nodes X 差值 , the difference between the vertical coordinates Y 差值 and the number of nodes to be interpolated, calculate the coordinate increment p3′ (X 增量， Y 增量 ); The interpolation points are distributed at equal distances in the horizontal and vertical directions between the two road nodes according to the coordinate difference between the two road nodes, the number of nodes to be interpolated and the corresponding coordinate increments; According to the above method, each original road network node in the set P' of all drivable domain roads in the standard definition electronic map is traversed and interpolated to form a set of candidate vehicle position nodes in the electronic map. ; Among them, X 差值 =x2-x1,Y 差值 =y2-y1,X 增量 =X 差值 / Number of nodes to be interpolated, Y 增量 = Y 差值 / Number of nodes to be interpolated, N represents the number of road nodes after interpolation in the electronic map; p1, p2, ... p N Represents the plane node coordinates corresponding to the 1st, 2nd, ...Nth road nodes.

3. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The parsing and preprocessing of the standard definition electronic map includes: Obtaining the road network data corresponding to the standard definition electronic map, parsing the connection relationship of the nodes in the standard definition electronic map, and extracting all the static feature boundary information nodes in the parsed standard definition electronic map; Filter nodes by selecting boundary attributes to select target static objects with boundary, wherein the boundary attributes include buildings, fences, barriers and road-related boundary objects; Divide according to the same attributes of the boundary and construct a static object set composed of static ground feature boundary points ; Where K is the number of static feature boundaries contained in the standard definition electronic map, and the element Bj in the set B represents the set of nodes of static features corresponding to the same attribute, j=1, 2, ..., K; ; , ,... Represents the two-dimensional coordinates of the 1st, 2nd, ...nth nodes in the static feature in the electronic map.

4. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The determining and screening of the static object boundary within a preset range of the vehicle position candidate node set includes: Traverse the set of candidate vehicle position nodes P, take the current vehicle global position candidate node as the center in the geographic coordinate system of the standard definition electronic map, and use the preset prior visual range distance parameter R P , with R P The static features within the radius are screened and truncated; Calculate the distance between the current vehicle position candidate node and all the static object corresponding nodes to obtain a distance matrix, wherein each row of the distance matrix represents a vehicle position candidate node, each column represents a static object corresponding node, and the elements in the distance matrix represent the target distance from the vehicle position candidate node to the static object corresponding node; Determine whether the target distance in the distance matrix is ​​greater than the static object node within the preset prior visible range. When the target distance is greater than the static object node within the preset prior visible range, screen the static object corresponding nodes corresponding to the target distance and perform temporary filtering in the current state to form a local prior candidate set formed by a single vehicle global position candidate node and a boundary within the prior visible range.

5. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The method of extracting the city structured static land object description information on the standard definition electronic map after screening to construct a priori static land object environment description database includes: In the geographic coordinate system of the standard definition electronic map, the candidate node of the current vehicle global position is taken as the center of the circle, and the same angle interval is taken. The ray of the current vehicle global position candidate node is segmented to form a plurality of segmented areas arranged in a counter-clockwise order, and the area close to the boundary of the current vehicle global position candidate node is scanned; Calculate the points where each angle ray intersects with the boundary line of the static object in turn, and take the plane coordinates corresponding to the point as the discrete point cloud coordinates in the segmented area to generate a discrete static object boundary point cloud set , and the candidate coordinates of the vehicle global position Have a unique corresponding relationship; Calculate the direction vectors between discrete static object boundary point clouds in sequence And the preset observation direction vector , and calculate the angle between the current boundary direction vector and the preset observation direction vector The one-dimensional vector formed is used as a priori static feature boundary representation , in order to realize the extraction of the boundary direction trend of static objects in the standard definition electronic map scene; Among them, the current direction vector is calculated by the following method: ； in, Indicates the number of columns of the vector, i.e. the number of plane splitting rays, and Respectively represent the position coordinates of the nth and n-1th discrete static object boundary point clouds, Indicates the direction vector from the n-1th discrete static object boundary point cloud to the nth discrete static object boundary point cloud, x i and i For vector The starting point coordinates, x n and n For vector The end point coordinates are also the position coordinates of the nth discrete static feature boundary point cloud.

6. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The real-time input of three-dimensional point cloud data by vehicle real-time perception to extract the description of urban structured static objects and obtain real-time static object environment description data includes: Taking the current vehicle observation point as the center, based on the 3D point cloud data input by the vehicle-side real-time perception, the data is screened according to the same attribute categories as the static objects in the prior standard definition electronic map, and a real-time static object boundary 3D point cloud set S with the same attributes is obtained. The 3D point cloud set S is converted into 2D coordinates by XOY plane projection in the vehicle body coordinate system, and an unordered plane point cloud set S′ containing only x-coordinates and y-coordinates is obtained. Centered on the current vehicle observation point, take the same angle interval The ray divides the XOY plane of the current vehicle coordinate system into multiple divided areas arranged in a counter-clockwise order; For each point cloud in the segmented area, the plane average center of the point cloud set in the segmented area is solved, and the coordinates of the average center are used as the equivalent point cloud in the area, and finally the real-time static feature boundary plane equivalent input point cloud set is obtained. ; Calculate the direction vectors between the equivalent input point clouds of the real-time static object boundary plane in sequence And the corresponding observation direction vector , and calculate the angle between the current boundary direction vector and the observation direction vector , forming a one-dimensional vector As a real-time static object boundary environment representation, it can realize the extraction of static object boundary direction trend in the real-time perception scenario on the vehicle side; Among them, the current direction vector is calculated by the following method: ； in, Indicates the number of columns of the vector, i.e. the number of plane splitting rays, and They represent the position coordinates of the nth and n-1th real-time point clouds respectively. Indicates the direction vector from the n-1th real-time static object boundary equivalent input point cloud to the nth real-time static object boundary equivalent input point cloud, x n and n For vector The vector coordinates in the XOY plane of the vehicle coordinate system centered on the current vehicle observation point are also the position coordinates of the equivalent input point cloud of the static object boundary at the nth time.

7. The method for global position estimation of autonomous driving based on static object association of standard definition electronic map according to claim 1, characterized in that: The method of performing similarity correlation between the static object data in the standard definition electronic map and the real-time static object environment description data perceived by the vehicle in real time, and obtaining the result with the highest similarity to the real-time perception of the vehicle to achieve global position estimation of the autonomous driving vehicle includes: By traversing each element in the prior static object environment description database, the most similar result to the real-time perception is queried in the prior static object environment description database, and the similarity calculation is implemented using the n-norm method as follows: ； Where D(i) represents the similarity between the input static object environment description vector perceived in real time and the i-th candidate in the prior static object environment description database. Represents the reference vector corresponding to the real-time perceived static feature description, represents the i-th static object description vector corresponding to each vehicle global position candidate in the database, and N is the total number of static object environment description vectors in the prior static object environment description database; The calculated similarity results are compared to select the candidate result with the highest similarity to the input perception vector, and the vehicle's global position estimate is obtained by accessing the candidate coordinates of the vehicle's global position corresponding to the optimal candidate result.

8. An automatic driving global position estimation system based on static object association of standard definition electronic map, characterized in that: include: Data acquisition module, used to obtain standard-definition electronic maps in the scenario of autonomous driving city operation; A parsing and processing module, used for parsing and preprocessing the standard definition electronic map to extract a set of candidate vehicle position nodes during automatic driving; An electronic map extraction module is used to determine and screen the static feature boundaries within a preset range for the vehicle position candidate node set, and after screening, extract the city structured static feature description information on the standard definition electronic map to construct a priori static feature environment description database; A real-time perception and extraction module is used to extract the description of urban structured static objects through real-time perception of vehicles and input three-dimensional point cloud data in real time, so as to obtain real-time static object environment description data; The position estimation module is used to associate the static object data in the standard-definition electronic map with the real-time static object environment description data perceived by the vehicle side in real time, and obtain the position candidate result with the highest similarity to the real-time perception of the vehicle side to realize the global position estimation of the autonomous driving vehicle.

9. An electronic device, characterized in that: include: A memory, the memory being used to store a processing program; A processor, wherein when executing the processing program, the processor implements the method for global position estimation of autonomous driving based on static object association of standard definition electronic map as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a processing program, and when the processing program is executed by the processor, the method for estimating the global position of an autonomous driving based on the association of static objects in a standard-definition electronic map is implemented as described in any one of claims 1 to 7.

Images