Map generation method, electronic device, vehicle, and program product

Abstract

The method comprises: generating a corresponding map element segment group according to a plurality of map element segments corresponding to a same type of map element; performing rasterization division on the plurality of map element segments in each map element segment group to obtain a plurality of grids, performing sorting on a plurality of element points in each grid to obtain a sorted grid, and performing sorting on each grid to obtain a target map element segment; performing segmented fitting processing on each target map element segment to obtain a corresponding target map segment; determining a key element point in a plurality of element points of the target map segment, and generating a target map element according to the key element point; and generating a target map according to target map elements corresponding to the same type of map element. In this way, complete and accurate map elements can be obtained for the same type of map element, and then a complete and accurate map can be obtained according to the map elements.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of map generation, and in particular relates to a map generation method, an electronic device, a vehicle and a program product. BACKGROUND

[0002] A map plays an important role in a navigation scenario and a vehicle auxiliary driving scenario. The accuracy and completeness of map elements such as lane lines, road edges, stop lines and ground markings in the map directly affect the accuracy of a navigation route, and further affect the safety of vehicle driving. Therefore, how to generate more accurate and complete map elements is an important research direction in the technical field of map generation. SUMMARY

[0003] The embodiments of the present application provide a map generation method, an electronic device, a vehicle and a program product, which can generate more accurate and complete map elements, and further generate a more accurate and complete map based on the more accurate and complete map elements.

[0004] To solve the above technical problem, in a first aspect, the embodiments of the present application provide a map generation method, which comprises: determining a plurality of sub-maps, and determining at least one map element in each sub-map to obtain a plurality of map elements corresponding to the plurality of sub-maps; determining map elements of a same type in the plurality of map elements corresponding to the plurality of sub-maps, and determining a plurality of map element segments corresponding to the map elements of the same type, generating a map element segment group corresponding to the map elements of the same type according to the plurality of map element segments corresponding to the map elements of the same type, to obtain at least one map element segment group corresponding to the plurality of sub-maps; performing rasterization division on a plurality of map element segments included in each map element segment group to obtain a plurality of grids corresponding to the plurality of map element segments, performing sorting on a plurality of element points included in each grid to obtain a sorted grid, and sorting a plurality of grids, generating a target map element segment corresponding to the map elements of the same type according to the sorted plurality of grids, performing segmented fitting processing on each target map element segment to obtain a target segment corresponding to the map elements of the same type, determining a key element point in a plurality of element points included in the target segment corresponding to the map elements of the same type, and generating a target map element corresponding to the map elements of the same type according to the key element point; and generating a target map according to the target map elements corresponding to the map elements of the same type.

[0005] According to the technical solution, the map element segments of the same type of map elements are rasterized and divided, a plurality of grids corresponding to the plurality of map element segments are obtained, a plurality of points in each grid are sorted to obtain a sorted grid, the plurality of grids are sorted to obtain a sorted target map element segment, the target map element segment is processed by segment fitting to obtain a target segment corresponding to the same type of map elements, key points in a plurality of element points of each target segment are determined, each target map element is generated according to the key points, and a map is generated according to each target map element. In this way, for the same type of map elements, raster division, sorting, segment fitting, element point screening and other operations are performed, and complete and accurate target map elements can be obtained, and then a complete and accurate target map can be obtained according to each complete and accurate target map element.

[0006] In a possible implementation of the first aspect, determining a plurality of map element segments corresponding to the same type of map elements, and generating a map element segment group corresponding to the same type of map elements according to the plurality of map element segments corresponding to the same type of map elements includes: converting coordinate systems of a plurality of sub-maps into a same coordinate system, determining map element segments corresponding to map elements included in the plurality of sub-maps in the same coordinate system, and determining distances between the map elements; based on an undirected graph matching method, classifying and grouping the plurality of map element segments according to the distances between the map elements, determining the plurality of map element segments corresponding to the same type of map elements, and generating the map element segment group corresponding to the same type of map elements according to the plurality of map element segments corresponding to the same type of map elements.

[0007] According to the technical solution, the map element segments corresponding to the same type of map elements are classified into a map element segment group, which facilitates generation of a complete segment corresponding to the type of map elements according to the map element segments in the map element segment group.

[0008] In a possible implementation of the first aspect, rasterizing and dividing a plurality of map element segments included in each map element segment group to obtain a plurality of grids corresponding to the plurality of map element segments, and sorting a plurality of element points included in each grid to obtain a sorted grid includes: rasterizing and dividing each map element segment included in each map element segment group based on a divide-and-conquer method to obtain a grid corresponding to each map element segment; clustering the plurality of element points included in each grid into different classes based on a density clustering algorithm; and obtaining shortest paths between the element points in the classes based on a shortest path algorithm, sorting the element points included in each grid according to the shortest paths to obtain a sorted grid.

[0009] According to the technical scheme, the map element segments are rasterized and divided, the element points in each raster are sorted to generate an ordered point set, and all the rasters are sorted to generate accurate and ordered map element segments.

[0010] In a possible implementation of the first aspect, the segment fitting processing on the target map element segments is performed to obtain the target map segment corresponding to the same type of map element, including: performing segment processing on the target map element segments to obtain a plurality of map element blocks corresponding to each target map element segment, wherein each map element block has a repeated area with another map element block; performing fitting processing on the plurality of map element blocks to obtain a fitting result; performing clustering processing on the fitting result to obtain a repeated area confidence of the plurality of map element blocks corresponding to each target map element segment; and obtaining the target map segment according to the map element block with the highest repeated area confidence.

[0011] According to the technical scheme, the target map segment is obtained based on the repeated area confidence, and complete and accurate map segments corresponding to map elements of each type can be generated.

[0012] In a possible implementation of the first aspect, determining the key element points in the target map segment corresponding to the same type of map element includes: determining a direction vector of the plurality of element points based on a dynamic programming method, and determining the element points deviating from the direction vector by a preset distance as the key element points.

[0013] According to the technical scheme, the key element points are screened based on the direction vector of the plurality of element points, the element points deviating from the direction vector are determined, the key element points used to describe the curve shape of the map element are obtained, and the accuracy of the generated target map element is ensured.

[0014] In a possible implementation of the first aspect, generating a target map according to the target map elements corresponding to the same type of map element includes: performing splicing processing on the target map elements corresponding to the same type of map element to obtain the target map.

[0015] According to the technical scheme, the accurate and complete target map elements are obtained by rearranging the map elements and obtaining the key element points, and the complete and accurate target map is obtained by splicing the target map elements.

[0016] In a possible implementation of the first aspect, generating the target map element corresponding to the same type of map element according to the key element points includes: generating the target map element corresponding to the same type of map element according to the key element points and target association information, and the target association information is lane line related information.

[0017] By adopting the technical scheme, the key element points and the lane line related information are associated, so as to obtain the relationship between each target map element and the lane line, and obtain a map that can accurately reflect the route.

[0018] In a second aspect, the implementation of the present application further discloses a vehicle for implementing the map generation method provided by any one of the implementation manners of the first aspect.

[0019] In a third aspect, the implementation of the present application further discloses an electronic device, comprising a processor and a memory connected with the processor in communication; the memory stores computer execution instructions; and the processor executes the computer execution instructions stored in the memory, so that the electronic device implements the map generation method provided by any one of the implementation manners of the first aspect.

[0020] In a fourth aspect, the implementation of the present application further discloses a computer readable storage medium, which stores a computer program, and the computer program can be executed by an electronic device to implement the map generation method provided by any one of the implementation manners of the first aspect.

[0021] In a fifth aspect, the implementation of the present application further discloses a computer program product, comprising a computer program, and the computer program is executed by an electronic device to implement the map generation method provided by any one of the implementation manners of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the present application, the drawings used in the implementation description will be briefly introduced.

[0023] Figure 1 A flowchart of a map generation method provided by an embodiment of the present application;

[0024] Figure 2 A flowchart of obtaining a map element segment group provided by an embodiment of the present application;

[0025] Figure 3 A flowchart of obtaining a sorted grid provided by an embodiment of the present application;

[0026] Figure 4 A flowchart of obtaining a target map segment provided by an embodiment of the present application;

[0027] Figure 5 A principle diagram of a map generation method provided by an embodiment of the present application;

[0028] Figure 6 A function relationship diagram of generating a map provided by an embodiment of the present application;

[0029] Figure 7 FIG. 1 shows a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] As mentioned above, the map plays an important role in the navigation scene and the vehicle auxiliary driving scene. In the map generation process, a plurality of map subgraphs (i.e., subgraphs) are predicted based on a map prediction model, each of which contains lane lines, road edges, stop lines, ground markings and other static targets of multiple categories. The complete map information can be obtained by splicing the plurality of map subgraphs. Therefore, the accuracy and completeness of the positions of the lane lines, road edges, stop lines, ground markings and other map elements in the map directly affect the accuracy of the navigation route, and further affect the safety of the vehicle driving. Therefore, how to generate a more accurate and complete map is an important research direction in the field of map generation technology.

[0031] Based on this, the implementation manner of the present application proposes a map generation method. The same type of map element segments in each subgraph are grouped to obtain a plurality of map element segment groups. Then, the map element segments in each map element segment group are rasterized and divided, and the raster is sorted to obtain a complete target map element segment corresponding to the same type of map element. Then, the target map element segment is subjected to a piecewise fitting process to obtain a target map segment corresponding to the same type of map element. Then, a key element point extraction process is performed to obtain an accurate and complete target map element corresponding to the same type of map element. Finally, a target map is generated according to each target map element.

[0032] Next, the map generation method provided by the implementation manner of the present application will be described in detail.

[0033] The map generation method provided by the implementation manner of the present application, as shown in Figure 1 includes the following steps.

[0034] S100, a plurality of subgraphs are determined, and at least one map element in each subgraph is determined to obtain a plurality of map elements corresponding to the plurality of subgraphs.

[0035] S200, the same type of map elements in the plurality of map elements corresponding to the plurality of subgraphs is determined, and a plurality of map element segments corresponding to the same type of map elements is determined. A map element segment group corresponding to the same type of map elements is generated according to the plurality of map element segments corresponding to the same type of map elements, and at least one map element segment group corresponding to the plurality of subgraphs is obtained.

[0036] S300, rasterize and divide the plurality of map element segments included in each map element segment group to obtain a plurality of grids corresponding to the plurality of map element segments, sort a plurality of element points included in each grid to obtain a sorted grid, and sort the plurality of grids, and generate a target map element segment corresponding to a map element of the same type according to the sorted plurality of grids.

[0037] S400, segment fitting processing is performed on each target map element segment to obtain a target map segment corresponding to a map element of the same type, determine a key element point in the plurality of element points included in the target map segment, and generate a target map element corresponding to a map element of the same type according to the key element point.

[0038] S500, generate a target map according to each target map element corresponding to a map element of the same type.

[0039] The map generation method provided by the implementation manner of the application rasterizes and divides each map element segment of a map element of the same type to obtain a plurality of grids corresponding to a plurality of map element segments, sorts a plurality of points in each grid to obtain a sorted grid, sorts the plurality of grids to obtain a sorted target map element segment, performs segment fitting processing on the target map element segment, determines a key point in a plurality of element points of each target map segment, generates each target map element according to the key point, and generates a map according to each target map element. In this way, for a map element of the same type, by performing grid division, sorting, segment fitting, element point screening and other operations, a complete and accurate target map element can be obtained, and then a complete and accurate target map can be obtained according to each complete and accurate target map element.

[0040] First, step S100 is performed to determine a plurality of sub-maps, and at least one map element in each sub-map is determined to obtain a plurality of map elements corresponding to the plurality of sub-maps.

[0041] For example, the map prediction model generates a plurality of sub-maps according to the collected map images, and detects map elements in each sub-map based on two-dimensional detection or three-dimensional detection to obtain a plurality of map elements corresponding to the plurality of sub-maps.

[0042] Each sub-map includes one or more map elements, such as a plurality of static map elements including lane lines, road edges, ground markings, zebra crossings, traffic lights, stop lines, etc.

[0043] Next, step S200 is performed to determine a plurality of map element segments corresponding to a map element of the same type, and generate a map element segment group corresponding to a map element of the same type according to the plurality of map element segments corresponding to the map element of the same type.

[0044] For example, map elements can be categorized, with map elements of the same type (i.e., the same map element) grouped together to obtain multiple types of map elements.

[0045] Furthermore, such as Figure 2 As shown, generating a group of map element fragments corresponding to the same type of map element from multiple map element fragments of the same type includes the following steps.

[0046] S210, convert the coordinate systems of multiple sub-maps to the same coordinate system, determine the map element fragments corresponding to each map element included in the multiple sub-maps under the same coordinate system, and determine the distance between each map element.

[0047] Since each submap has an arbitrary orientation and contains any complete or incomplete map elements, it is necessary to determine the relationships between different map elements and group map elements of the same type in multiple submaps.

[0048] For example, the camera coordinates of all sub-maps are converted to the same coordinate system, the distance from a point to a curve is calculated based on the triangulation of the points, and this is extended to the distance between map elements.

[0049] Specifically, it detects common feature points (such as traffic signs and special landmarks) between sub-maps.

[0050] Calculate the similarity transformation matrix based on the rotation angle, translation amount, and scaling factor of each sub-map:

[0051]

[0052] in, For the rotation angle, ( ) is the translation amount, and s is the scaling factor, which is usually 1.

[0053] All submaps are aligned to a global coordinate system (such as the Universal Transversor Grid System (UTM) or a local reference coordinate system) based on a transformation matrix, eliminating rotation / evaluation differences and ensuring that all submaps have the same orientation.

[0054] Furthermore, for any two map elements, determine a point in one map element and a curve in the other map element, and calculate the distance between the areas of the smallest circumscribed triangle.

[0055] Then, for two curves in two map elements, the key points are sampled to calculate the bidirectional Hausdorff distance between the two curves, thus obtaining the distance between the two map elements.

[0056] S220, according to the distance between the map elements, the plurality of map element segments are classified and grouped based on the undirected graph matching method to determine the plurality of map element segments corresponding to the same map element.

[0057] S230, according to the plurality of map element segments corresponding to the same type of map element, a map element segment group corresponding to the same type of map element is generated.

[0058] Exemplarily, each map element segment is taken as a node, the distance between the map elements is taken as a variable weight, the similarity of each map element segment is calculated by using a spectral clustering or Louvain algorithm, the intra-group similarity is maximized, the collinear map element segments with similar similarity are grouped into the same group, and the map element segment group corresponding to the same type of map element is obtained.

[0059] Next, step S300 is performed, the plurality of map element segments in each map element segment group are rasterized and divided to obtain a plurality of grids corresponding to the plurality of map element segments, the plurality of element points in each grid are sorted to obtain a sorted grid, and each grid is sorted to generate a target map element segment corresponding to the same type of map element according to the plurality of sorted grids.

[0060] The element point set of one map element is in order, the point set of the same map element (for example, a lane line) is dispersed in a plurality of sub-maps, the local order is lost, and a long-distance element may have a non-continuous gap, so that direct global sorting is easy to produce distortion, and the same type of map element segment in all sub-maps needs to be found and the element segment set is sorted according to the curve extension rule.

[0061] In the implementation manner of the present application, as shown in the figure, Figure 3 The plurality of map element segments included in each map element segment group are rasterized and divided to obtain a plurality of grids corresponding to the plurality of map element segments, the plurality of element points included in each grid are sorted to obtain a sorted grid, and the target map element segment corresponding to the same type of map element is generated according to the plurality of sorted grids.

[0062] S310, based on the divide-and-conquer method, the plurality of map element segments included in each map element segment group are rasterized and divided to obtain a grid corresponding to each map element segment.

[0063] Exemplarily, the map element segments are divided into overlapping grids according to a fixed size to ensure repeated coverage of the boundary points, so that the complexity of long-distance sorting can be reduced.

[0064] S320, based on the density clustering algorithm, the plurality of element points included in each grid are clustered into different classes.

[0065] Exemplarily, the points in each grid are quickly queried using a spatial index, and the multiple element points in each grid are clustered based on a DBSCAN clustering algorithm to obtain multiple different classes.

[0066] Specifically, a neighborhood radius and a minimum point number are set, and a clustering label is output for the points in each grid based on the DBSCAN clustering algorithm. The points with the same label belong to the same curve branch and can be classified into a class. In this way, multiple different classes are obtained. The points in the same class have the same label, and the points in different classes have different labels.

[0067] In S330, the shortest paths between the element points in the class are obtained based on a shortest path algorithm, and the element points included in each grid are sorted according to the shortest paths to obtain a sorted grid.

[0068] Exemplarily, for each class, all the element points in the class are taken as nodes, edges are formed by connecting the expected nearest 5 adjacent points of each point, the Euclidean distance between two points is calculated, and the main direction of all the element points is determined. Principal component analysis (PCA) is performed on the element point set in the class, the points on the outermost side along the first principal component are selected as the start / endpoint of the path, the Dijkstra algorithm is run with the start point as the source node, and the point set is output in the order of the shortest path. If the distance between the endpoint and the start point is less than a threshold, a closed loop is forcibly connected. In this way, the shortest path of each element point is obtained.

[0069] The element points are sorted based on the shortest paths to obtain a sorted grid.

[0070] Further, all the grids are sorted again to obtain a complete target map element segment.

[0071] Exemplarily, the overlapping regions of adjacent grids are aligned through Iterative Closest Point Transform (ICP registration) or least squares fitting to ensure geometric continuity. The direction consistency and spacing of the connection are checked, and unreasonable connections are removed. The sorting results of all the grids are polynomial smoothed to eliminate local jitter. In this way, the sorted grids are connected based on the overlapping regions to realize secondary sorting, and all the grids are spliced to obtain a complete map element segment with a connection order.

[0072] Next, step S400 is performed to perform segmentation fitting processing on the target map element segments to obtain target map segments corresponding to map elements of the same type.

[0073] In the implementation of the present application, as Figure 4As shown, the segmentation fitting processing is performed on each target map element segment to obtain a target map segment corresponding to the same type of map element, including the following steps.

[0074] S410, segmenting each target map element segment to obtain a plurality of map element blocks corresponding to each target map element segment, wherein each map element block has a repeated area with another map element block.

[0075] For example, the segmentation parameters (block length and overlap area) are determined, and the target map element segment is segmented based on the block length and the overlap area to obtain a plurality of map element blocks.

[0076] S420, fitting processing is performed on the plurality of map element blocks to obtain a fitting result.

[0077] For example, the plurality of map element blocks are fitted using a fitting method to obtain a fitting result. The fitting result includes a fitting curve.

[0078] The fitting method can be a 3rd order B-spline fitting.

[0079] S430, clustering processing is performed on the fitting result to obtain a repeated area confidence of the plurality of map element blocks corresponding to each target map element segment.

[0080] For example, 10 points are uniformly sampled in the overlap area of each two map element blocks, the mean square error of the two fitting curves is calculated, and the first confidence is obtained according to the mean square error. And, the curvature difference of both ends is calculated, and the second confidence is obtained according to the curvature difference. And check whether the connection of adjacent map element blocks is broken, to obtain the second confidence.

[0081] Based on the preset weight, the first confidence, the second confidence, and the third confidence are weighted and summed to obtain the repeated area confidence.

[0082] S440, obtaining a target map segment according to the map element block with the highest repeated area confidence.

[0083] Determine the map element block corresponding to the overlap area with the highest confidence as the true value to obtain the target map segment.

[0084] Further, determining the key element points in the plurality of element points included in the target map segment corresponding to the same type of map element includes: determining the direction vector of the plurality of element points based on the dynamic programming method, and determining the element points deviating from the direction vector by a preset distance in the plurality of element points as the key element points.

[0085] For example, the plurality of element points in the target map picture segment are an ordered point set based on the foregoing ordering. For each element point, a local motion direction of the element point is calculated, a direction vector of a curve formed by all the element points is obtained, and for a curve, the greater the distance between an element segment and the direction vector, the more necessary the element point is to describe the shape of the curve. Therefore, by calculating the geometric distance between each element point and the direction vector, the element points deviating from the direction vector are determined, that is, the element points greater than the preset distance of the direction vector are reserved as key element points to represent the curve shape of the target map element.

[0086] Further, the target map element segment composed of the key element points is taken as the expression of the corresponding map element to obtain the target map picture segment. In this way, the element points are screened.

[0087] In the implementation manner of the present application, the target map element corresponding to the same type of map element is generated according to the key element points, including: generating the corresponding target map element according to each key element point and target association information, and the target association information is lane line related information.

[0088] For example, the lane line element is determined from the plurality of target map elements, and the other target map elements and the lane line element are associated to judge the breakpoint and branch line information, so as to obtain a target map element set accurately divided according to the intersection lane line.

[0089] The target map is generated according to the target map elements corresponding to each same type of map element, including: performing splicing processing on the target map elements corresponding to each same type of map element to obtain the target map.

[0090] For example, the target map elements corresponding to the plurality of different types of map elements are spliced to obtain the target map.

[0091] Further, as shown in Figure 5 The map generation method provided by the implementation manner of the present application performs merging processing on the map subgraphs (i.e., sub-maps) A, B, C, … containing lane lines, road edges, stop lines, map identifiers, zebra crossings and other static elements, for example, merging the map subgraph A and the map subgraph B to obtain the map subgraph AB, and then merging the map subgraph AB, the map subgraph C and other map subgraphs to obtain a complete map element.

[0092] Further, as shown in Figure 6 In the implementation manner of the present application, the generation of the map is realized based on a preset code script.

[0093] Specifically, the basic instance hierarchy is defined. Among them, the instance class (Instance) is the base class of all map elements, which defines common attributes and methods (such as ID, confidence, geometric type), the point instance class (PointInstance) is used to store key point data (such as stop line end point, lane line turning point), which contains coordinates and topological relationship, the line instance class (LineInstance) is used to represent linear elements (lane line, road edge), which stores ordered point set, fitting parameters and connection relationship, and the polygon instance class (PolygonInstance) is used to represent surface elements (ground sign, custom area), which supports complex polygon and hole processing.

[0094] Further, the processing logic of the subgraph level is defined. Among them, the frame instance container function (FrameInstancesContainer) is a temporary container for single frame data, which stores original detection instances (such as storing the results of a single map subgraph), the frame container function (FrameContainer) is an extended frame container, which supports frame-level operations (such as verification, serialization), and the map container function (MapContainer) is a global map container, which is a function for judging the relative relationship between subgraphs, managing instances after multi-frame fusion, supporting cross-frame topological construction and persistent storage. Among them, FrameContainer is the inherited class of FrameInstancesContainer and MapContainer.

[0095] Further, the verification function (check) is used for basic verification (such as coordinate range, data type) to eliminate unqualified instances, for example, preliminary fitting of lane lines to eliminate unqualified fitting information. The classification function (class_nms) is used for non-maximum suppression between instances of the same class, filtering repeated instances (according to confidence and IoU threshold), and classifying map element fragments between instances of the same class. The frame verification function (frame_check) is used to implement map element verification and audit between different classes, for example, to implement verification and audit between lane line type map elements and road edge type map elements. The key point binding function (bind_keypoint) is used to bind key points to parent instances (such as binding key points and lane lines), and key points are used to generate target map elements.

[0096] Further, a result class (LoadResult class) is defined for data conversion processing, such as an image processing function (to_image) for rendering instances into a raster image (for visualization or deep learning training), a labeling processing function (to_annotation) for generating annotation tool compatible formats (such as COCO, LabelMe), and a format processing function (to_json) for serializing data into JSON format, supporting nested structures (such as control points, attribute dictionaries).

[0097] Further, a combination frame container function (combine frame_container) combines multiple FrameContainer instances for incremental map updating.

[0098] Further, different processes are defined for map element processing, including a stop line processing function (stopline_process) for stop line map element corresponding map element segment generation processing (such as pair verification, perpendicularity constraint with lane lines), a lane line processing function (lane_process) for lane line map element corresponding map element segment generation processing (such as direction consistency, branch connection), a key point processing function (keypoint_process) for dynamic programming to extract key element points, integrate key element points, and output PointInstance for subsequent association with target associated information (such as association with lane line associated information), a curb processing function (curb_process) for road edge map element corresponding map element segment generation processing (geometric repair for dealing with occlusion caused by breakage), an arrow processing function (arrow_process) for ground sign map element corresponding map element segment generation processing (such as vectorization of turning arrows and text symbols), a polygon processing function (polygon_process) for surface map element corresponding map element segment generation processing (such as hierarchical filling of parking areas). A lane refining function (lane_refine) is used to determine breakpoints and branch lines based on the relationship between key points and lane lines, to perform multi-lane line fusion and smoothing processing to eliminate jitter, and to obtain a global topology based on geometric and semantic similarity.

[0099] The map generation method provided by the implementation of the application is actually a large-scale map element splicing method based on dynamic matching and adaptive fusion. Different types of map elements are obtained by classifying map elements of multiple sub-maps, map element segments of the same type of map elements are rasterized, and element points are sorted to generate accurate target map element segments. Each target map element segment is subjected to piecewise fitting processing to obtain a complete target map segment. The target map segment is further subjected to key element point extraction to obtain an accurate and complete target map element. The target map element is spliced to generate an accurate and complete target map.

[0100] The map generation method provided by the implementation of the application is applied to an electronic device.

[0101] Please refer to Figure 7 , Figure 7 The structure of the electronic device provided by the embodiment of the application is shown in FIG. 1. As shown in FIG. 1, the electronic device can include a transceiver 121, a processor 122, and a memory 123. Figure 7

[0102] The processor 122 executes computer execution instructions stored in the memory, so that the processor 122 executes the technical solutions of the map generation method in the above embodiments. The processor 122 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0103] The memory 123 is connected with the processor 122 through a system bus and completes mutual communication, and the memory 123 is used for storing computer program instructions.

[0104] ​By way of example, and without limitation, memory 123 can include a hard disk drive (HDD), a floppy drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Storage 123 can include removable or non-removable (or fixed) media, where appropriate. Storage 123 can be internal or external to integrated gateway device, where appropriate. In particular embodiments, storage 123 is nonvolatile, solid-state memory. In particular embodiments, storage 123 includes read-only memory (ROM). Where appropriate, this ROM can be mask-programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically alterable ROM (EAROM), or flash memory or a combination of two or more of these. Transceiver 121 can be used to obtain tasks to be run and configuration information for tasks to be run.

[0105] The system bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The system bus can be implemented as a single bus or a combination of buses, such as an address bus, a data bus, and a control bus, using newer technologies. The transceiver is used to implement communication between the database access device and other computers, such as clients, read-write libraries, and read-only libraries. The memory can include random access memory (RAM) and can also include non-volatile memory.

[0106] Further, the electronic device may, for example, be a computer, a mobile phone, a server, or the like.

[0107] Further, the implementation manner of the present application also discloses a vehicle, and the vehicle performs the map generation method to generate a map.

[0108] The embodiment of the present application further provides a chip for running instructions, which is used for executing the technical solution of the map generation method in the above embodiment.

[0109] The embodiment of the present application further provides a computer readable storage medium, which stores computer instructions, and when the computer instructions run on a processor of an electronic device, the processor of the electronic device executes the technical solution of the map generation method in the above embodiment.

[0110] In some possible implementation manners, various aspects of the method provided by the present application can also be implemented in the form of a program product, which includes program codes, and when the program product runs on a processor of an electronic device, the program codes are used to make the processor of the electronic device execute the steps in the method according to various exemplary implementation manners of the present application described above in the specification, for example, the electronic device can execute the map generation method described in the embodiment of the present application.

[0111] The program product can adopt any combination of one or more readable media. The readable medium can be a readable data medium or a readable storage medium. The readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include an electrical connection having one or more wires, a portable disc, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CDROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0112] The implementation manner of the present application further provides a computer program product, which includes a computer program stored in a computer readable storage medium, at least one processor can read the computer program from the computer readable storage medium, and when the at least one processor executes the computer program, the technical solution of the map generation method in the above embodiment can be implemented.

[0113] It is to be understood that the above description is intended to be illustrative and not restrictive. Many other implementations will be apparent to those of skill in the art upon reading and understanding the above description. Although the application has been described with reference to certain implementations, it is to be understood that these are intended in an illustrative sense and not a restrictive sense. The scope of the application should be interpreted in accordance with the following claims and their equivalents. The benefits provided by the features and attributes of the application will be apparent from all of the descriptions and / or examples discussed above and / or upon further pat of the following claims. As used throughout this application, the following terms have the following meanings.

[0114] It should be noted that in this specification, like reference numerals and letters indicate like items that are alike in various figures, and thus once an item is defined in one figure, it is not necessary to further define and explain it in a subsequent figure.

[0115] It should be noted that the terms "first", "second", and so on do not necessarily indicate any relative importance.

[0116] It should be noted that in the drawings, some structural or methodical features can be shown in a particular arrangement and / or order. However, it should be understood that such a particular arrangement and / or order can not be required. Rather, in some implementations, these features can be arranged in a different manner and / or order than shown in the illustrative drawings. Additionally, inclusion of structural or methodical features in a particular figure is not meant to imply that such features are required in all implementations, and in some implementations, these features can not be included or can be combined with other features.

[0117] While the application has been illustrated and described in detail in the drawings and foregoing description, the same is to be considered as illustrative and not restrictive in character, it being understood that only the preferred implementations have been shown and described and that all changes and modifications that come within the spirit of the application are desired to be protected. There are many alternative ways of implementing the application. The described embodiments are to be considered in all respects only as illustrative and not restrictive.

Claims

1. A map generation method, characterized in that, The method includes: Determine multiple sub-maps and at least one map element in each of the sub-maps to obtain multiple map elements corresponding to the multiple sub-maps; Determine the map elements of the same type among the map elements corresponding to the multiple sub-maps, and determine the multiple map element fragments corresponding to the map elements of the same type. Generate a group of map element fragments corresponding to the map elements of the same type based on the multiple map element fragments corresponding to the map elements of the same type, and obtain at least one group of map element fragments corresponding to the multiple sub-maps. The map element fragments included in each map element fragment group are rasterized to obtain multiple grids corresponding to the multiple map element fragments. The multiple element points included in each grid are sorted to obtain sorted grids. The multiple grids are then sorted, and target map element fragments corresponding to the same type of map elements are generated based on the sorted multiple grids. The target map element fragments are segmented and fitted to obtain target map fragments corresponding to the same type of map elements. Key element points among the multiple element points included in the target map fragments are determined, and target map elements corresponding to the same type of map elements are generated based on the key element points. Generate a target map based on the target map elements corresponding to each map element of the same type.

2. The map generation method according to claim 1, characterized in that, Identify multiple map element fragments corresponding to the same type of map element, and generate a group of map element fragments corresponding to the same type of map element based on the multiple map element fragments corresponding to the same type of map element, including: The coordinate systems of the multiple sub-maps are converted to the same coordinate system, the map element segments corresponding to each map element included in the multiple sub-maps under the same coordinate system are determined, and the distances between each map element are determined. Based on the undirected graph matching method, the multiple map element fragments are classified and grouped according to the distance between the map elements, multiple map element fragments corresponding to the same type of map element are determined, and the map element fragment group corresponding to the same type of map element is generated according to the multiple map element fragments corresponding to the same type of map element.

3. The map generation method according to claim 2, characterized in that, Each group of map element fragments is rasterized to obtain multiple grids corresponding to the multiple map element fragments. The multiple element points included in each grid are sorted to obtain the sorted grids, including: Based on the divide-and-conquer method, each map element fragment included in each map element fragment group is rasterized to obtain the raster corresponding to each map element fragment; Based on the density clustering algorithm, the multiple element points included in each grid are clustered into different classes; Based on the shortest path algorithm, the shortest path between each element point in the class is obtained, and the element points included in each grid are sorted according to the shortest path to obtain the sorted grid.

4. The map generation method according to claim 3, characterized in that, Each of the target map element fragments is segmented and fitted to obtain target map fragments corresponding to the same type of map element, including: Each of the target map element fragments is segmented to obtain multiple map element blocks corresponding to each target map element fragment, wherein each map element block has an overlapping area with another map element block; The multiple map element blocks are fitted to obtain the fitting result; Clustering is performed on the fitting results to obtain the confidence scores of the repeated regions of the multiple map element blocks corresponding to each target map element fragment; The target map fragment is obtained from the map element block with the highest confidence level in the repeated regions.

5. The map generation method according to claim 4, characterized in that, Identifying key element points among multiple element points in the target map fragment corresponding to the same type of map element, including: Based on the dynamic programming method, the direction vectors of the plurality of element points are determined, and the element points that deviate from the direction vectors by a preset distance among the plurality of element points are identified as the key element points.

6. The map generation method according to claim 5, characterized in that, Generate a target map based on the target map elements corresponding to each map element of the same type, including: The target map is obtained by stitching together the target map elements corresponding to each map element of the same type.

7. The map generation method according to any one of claims 1-6, characterized in that, Based on the key element points, generate target map elements corresponding to the same type of map elements, including: The target map element is generated based on the key element points and target association information, where the target association information is lane line related information.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes the computer execution instructions stored in the memory to perform the map generation method as described in any one of claims 1-7.

9. A vehicle, characterized in that, The vehicle is used to perform the map generation method as described in any one of claims 1-7.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, performs the map generation method as described in any one of claims 1-7.

Images