A map construction method, system, electronic device and storage medium

Abstract

Embodiments of the present application provide a kind of map construction method, system, electronic equipment and storage medium, comprising: each original map data fusion splicing, obtain preliminary mapping result;Obtain each target section information and target connection point information in preliminary mapping result, two adjacent target section and target connection point between them constitute target connection;For each target connection, obtain each target original map data, carry out graph optimization to each target original map data and adjacent target section, obtain optimized target original map data and optimized adjacent target section, aggregate and obtain after updating adjacent target section and its connection relationship;Based on each adjacent target section after updating and its connection relationship, generate target map.Using the embodiments of the present application, by using the original local map data between adjacent sections to optimize each part in the section, the missing of the connection between the sections can be repaired, the connection between the sections is smooth, and the map accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of map construction, and in particular to a map construction method and system, an electronic device and a storage medium. BACKGROUND

[0002] At present, automatic driving technology has been widely applied. High-precision map is an important factor to realize vehicle automatic driving. The ordinary map construction method has a long update cycle, which leads to low accuracy of the map and cannot meet the demand of vehicle automatic driving for high-precision map. Therefore, the method of crowdsourcing mapping is usually used to construct the automatic driving high-precision map.

[0003] The object of crowdsourcing mapping is a road network (graph) composed of various roads. The road network (graph) is composed of various road segments (edges) and connecting points (nodes) between the road segments. The crowdsourcing mapping is to map each road segment (edge) in the road network respectively, and then connect them through the connecting points (nodes) to achieve the purpose of mapping the entire road network. Since the cost of the collection device of crowdsourcing mapping is small and the precision is low, the map constructed for each road segment has a large absolute error, and due to the possible existence of intersections, crossroads and other scenes between road segments, there may be misalignment, missing and other situations between different road segments, which cannot be directly connected for use. This also leads to low precision of the map. SUMMARY

[0004] The purpose of the embodiment of the present application is to provide a map construction method, system, electronic device and storage medium to improve the precision of map construction. The specific technical solutions are as follows:

[0005] In one aspect of the present application, a map construction method is provided, which comprises:

[0006] Fusing and splicing each original map data to obtain a preliminary mapping result;

[0007] Obtaining each target road segment information in the preliminary mapping result and target connecting point information between adjacent target road segments; wherein, any two adjacent target road segments and the target connecting point between the adjacent target road segments constitute a target connection;

[0008] For each target connection, obtaining each target original map data including the target connecting point;

[0009] For each target connection, based on the each target original map data and the adjacent target road segment information, performing map optimization on the each target original map data and the adjacent target road segment to obtain each optimized target original map data and an optimized adjacent target road segment;

[0010] For each target connection, the each optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections.

[0011] Based on the each updated adjacent target road section and the connection relationship between the each updated adjacent target road section, a target map is generated.

[0012] In an embodiment of the present application, before the each target original map data and the adjacent target road section are graph-optimized based on the each target original map data and the adjacent target road section information to obtain each optimized target original map data and an optimized adjacent target road section, the method further comprises:

[0013] The each target original map data is aggregated according to a preset number to obtain each aggregated map data;

[0014] The each target original map data and the adjacent target road section are graph-optimized based on the each target original map data and the adjacent target road section information to obtain each optimized target original map data and an optimized adjacent target road section, comprising:

[0015] The each aggregated map data and the adjacent target road section are graph-optimized based on the each aggregated map data and the adjacent target road section information to obtain each optimized aggregated map data and an optimized adjacent target road section;

[0016] The each optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections, comprising:

[0017] The each optimized aggregated map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections.

[0018] In an embodiment of the present application, the aggregated map data and the adjacent target road section correspond to a plurality of trajectory points respectively;

[0019] The each aggregated map data and the adjacent target road section are graph-optimized based on the each aggregated map data and the adjacent target road section information to obtain each optimized aggregated map data and an optimized adjacent target road section, comprising:

[0020] The each aggregated map data and the adjacent target road section are cut according to a preset length to obtain each aggregated map slice and each target road section slice;

[0021] adding the first track points corresponding to the each aggregated map data and the second track points corresponding to the each target road section into a graph;

[0022] adding edges between the first track points corresponding to the same aggregated map data and edges between the second track points corresponding to the same target road section in the graph;

[0023] adding matching edges between the first track points corresponding to different aggregated map data, between the second track points corresponding to different target road sections and between the first track points and the second track points in the graph based on the each aggregated map data and the target road section information;

[0024] performing graph optimization on the each aggregated map data and the adjacent target road section based on the graph to obtain each optimized aggregated map data and optimized adjacent target road section.

[0025] In an embodiment of the present application, before the each aggregated map data and the adjacent target road section are divided into each aggregated map data slice and each target road section slice according to the preset length, the method further comprises:

[0026] for the adjacent target road section, dividing the adjacent target road section at a preset distance from the target connection point to obtain an interface area road section; wherein the interface area road section is a part close to the target connection point;

[0027] the adding edges between the first track points corresponding to the same aggregated map data and edges between the second track points corresponding to the same target road section in the graph comprises:

[0028] adding edges between the first track points corresponding to the same aggregated map data and edges between the second track points corresponding to the same interface area road section in the graph;

[0029] the adding matching edges between the first track points corresponding to different aggregated map data, between the second track points corresponding to different target road sections and between the first track points and the second track points in the graph comprises:

[0030] adding matching edges between the first track points corresponding to different aggregated map data, between the second track points corresponding to different interface area road sections and between the first track points and the second track points in the graph;

[0031] the performing graph optimization on the each aggregated map data and the adjacent target road section based on the graph to obtain each optimized aggregated map data and optimized adjacent target road section comprises:

[0032] Based on the graph, the each optimized aggregated map data and the interface area road segment in the adjacent target road segment are graph optimized, to obtain each optimized aggregated map data and the optimized interface area road segment.

[0033] In an embodiment of the present application, after the adjacent target road segment is split, a non-interface area road segment is also obtained;

[0034] The aggregation of the each optimized aggregated map data and the optimized adjacent target road segment obtains an updated adjacent target road segment and a connection relationship between the updated adjacent target road segments, which includes:

[0035] The aggregation of the each optimized aggregated map data and the optimized interface area road segment obtains a target aggregated road segment;

[0036] Based on the target connection point, the target aggregated road segment is split to obtain two target interface area road segments;

[0037] For each target road segment, based on the position of the target interface area road segment and the position of the non-interface area road segment, the target interface area road segment and the non-interface area road segment are connected to obtain an updated adjacent target road segment and a connection relationship between the updated adjacent target road segments.

[0038] In an embodiment of the present application, the each target road segment information includes a target road segment identifier;

[0039] The splitting of the each aggregated map data and the adjacent target road segment according to the preset length to obtain each aggregated map slice and each target road segment slice further includes:

[0040] A preset field is used to save the target road segment identifier corresponding to each target road segment slice;

[0041] The connection of the target interface area road segment and the non-interface area road segment for each target road segment based on the position of the target interface area road segment and the position of the non-interface area road segment to obtain an updated adjacent target road segment and a connection relationship between the updated adjacent target road segments includes:

[0042] The connection of the target interface area road segment and the non-interface area road segment for each target road segment based on the position of the target interface area road segment and the position of the non-interface area road segment to obtain an updated adjacent target road segment and a connection relationship between the updated adjacent target road segments includes:

[0043] In another aspect of the present application, a map construction system is provided, which includes a first geometric module, a second geometric module and a logic module.

[0044] The first geometry module is configured to fuse and splice each original map data to obtain a preliminary mapping result, acquire target road section information in the preliminary mapping result and target connection point information between adjacent target road sections, wherein any two adjacent target road sections and the target connection points between the adjacent target road sections form a target connection, and acquire each target original map data including the target connection point for each target connection;

[0045] The second geometry module is configured to perform graph optimization on each target original map data and the adjacent target road section based on the each target original map data and the adjacent target road section information for each target connection to obtain each optimized target original map data and an optimized adjacent target road section.

[0046] The logic module is configured to aggregate each optimized target original map data and the optimized adjacent target road section to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections for each target connection, and generate a target map based on each updated adjacent target road section and the connection relationship between the updated adjacent target road sections.

[0047] In another aspect of the embodiment of the present application, an electronic device is provided, which comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete communication with each other through the communication bus.

[0048] The memory is configured to store a computer program.

[0049] The processor is configured to execute the program stored on the memory to implement the steps of the map construction method.

[0050] In another aspect of the embodiment of the present application, a computer readable storage medium is provided, characterized in that the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the map construction method.

[0051] The embodiment of the present application further provides a computer program product comprising instructions which, when executed on a computer, cause the computer to perform the map construction method.

[0052] The embodiment of the present application has the following beneficial effects:

[0053] The map construction method provided by the embodiment of the present application fuses and splices each original map data to obtain a preliminary mapping result; target road section information in the preliminary mapping result and target connection point information between adjacent target road sections are acquired; wherein, any two adjacent target road sections and the target connection points between the adjacent target road sections form a target connection; for each target connection, each target original map data including the target connection point is acquired; based on the each target original map data and the adjacent target road section information, the each target original map data and the adjacent target road section are subjected to graph optimization to obtain each optimized target original map data and an optimized adjacent target road section; the each optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections; based on each updated adjacent target road section and the connection relationship between the updated adjacent target road sections, a target map is generated. By applying the embodiment of the present application, for each adjacent target road section in the preliminary mapping result, target map original map data including the connection points between the adjacent target road sections, that is, original local map data between the target road sections, is acquired. Since the target road section is a preliminary result of mapping according to the original map data, its relative accuracy and completeness rate are better, and by using the original local map data between the adjacent road sections to optimize each part in each road section in the preliminary mapping result, the missing of the connection between the road sections can be repaired, the connection between the road sections is smooth, and the map accuracy is further improved.

[0054] Of course, implementing any product or method of the present application does not necessarily require achieving all the advantages mentioned above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and other embodiments can also be obtained by those skilled in the art based on these drawings.

[0056] Figure 1a A flowchart of the map construction method provided by the embodiment of the present application;

[0057] Figure 1b A schematic diagram of a "connection" in the embodiment of the present application;

[0058] Figure 2 A second flowchart of the map construction method provided by the embodiment of the present application;

[0059] Figure 3A schematic diagram of aggregating a preset number of target map data in an embodiment of the present application;

[0060] Figure 4 A flowchart of map optimization in an embodiment of the present application;

[0061] Figure 5 Another flowchart of map optimization in an embodiment of the present application;

[0062] Figure 6 A schematic diagram of a map in an embodiment of the present application;

[0063] Figure 7 A flowchart of obtaining an updated target road segment and its connection relationship in an embodiment of the present application;

[0064] Figure 8 A schematic diagram of obtaining an updated target road segment and its connection relationship in an embodiment of the present application;

[0065] Figure 9 A schematic diagram of connecting each updated target road segment in an embodiment of the present application;

[0066] Figure 10 A structural schematic diagram of a map construction system provided in an embodiment of the present application;

[0067] Figure 11 An execution process schematic diagram of a map construction system provided in an embodiment of the present application;

[0068] Figure 12 A structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0069] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art based on the present application belong to the scope of protection of the present application.

[0070] Crowdsourcing mapping refers to uploading image data collected by multiple vehicles to the cloud, and aggregating and optimizing the data uploaded by each vehicle in the cloud to construct a map. As described above, the current crowdsourcing mapping has low precision and cannot meet the needs of autonomous driving.

[0071] Therefore, in order to improve the precision of map construction, the present application provides a map construction method, system, electronic device and storage medium.

[0072] The map construction method provided by the embodiments of the present application can be applied to electronic devices such as servers and computers, and the present application does not make specific limitations in this regard.

[0073] First, the map construction method provided by the present application is exemplarily described as follows:

[0074] Referring to Figure 1a , Figure 1a is a flowchart of the map construction method provided by the present application, which can include the following steps:

[0075] In step S110, each original map data is fused and spliced to obtain a preliminary mapping result.

[0076] In step S120, each target road section information in the preliminary mapping result and target connection point information between adjacent target road sections are obtained.

[0077] Among them, any two adjacent target road sections and the target connection points between the adjacent target road sections constitute a target connection.

[0078] In step S130, for each target connection, each target original map data including the target connection point is obtained.

[0079] In step S140, for each target connection, based on each target original map data and the adjacent target road section information, the each target original map data and the adjacent target road section are graphically optimized to obtain each optimized target original map data and an optimized adjacent target road section.

[0080] In step S150, for each target connection, the each optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections.

[0081] In step S160, based on each updated adjacent target road section and the connection relationship between the updated adjacent target road sections, a target map is generated.

[0082] The map construction method provided in the embodiments of the present application fuses and splices each original map data to obtain a preliminary mapping result; target road section information in the preliminary mapping result and target connection point information between adjacent target road sections are acquired; wherein, any two adjacent target road sections and the target connection points between the adjacent target road sections form a target connection; for each target connection, each target original map data including the target connection point is acquired; based on the target original map data and the adjacent target road section information, the target original map data and the adjacent target road section are graph-optimized to obtain each optimized target original map data and an optimized adjacent target road section; the optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections; based on each updated adjacent target road section and the connection relationship between the updated adjacent target road sections, a target map is generated. According to the embodiments of the present application, for each adjacent target road section in the preliminary mapping result, target map original map data including the connection points between the adjacent target road sections, that is, original local map data between the target road sections, is acquired. Since the target road section is a preliminary result of mapping according to the original map data, the relative accuracy and the completeness rate are better, and by optimizing each part in each road section in the preliminary mapping result by using the original local map data of the target between the adjacent road sections, the missing of the connection between the road sections can be repaired, the connection between the road sections is smooth, and the map accuracy is further improved.

[0083] The steps S110-S160 are exemplarily described as follows:

[0084] In the embodiments of the present application, the data collection vehicle carrying the image collection device can be used to collect the original map data. The original map data can include the identification (such as ID) of each original map data, the collection time, the pose information, the coordinate information (the coordinate can be a geographic coordinate), and the semantic information, etc. The semantic information can be the road element information extracted by the chip carried by the vehicle for full-map semantic segmentation of the image data. The road element information can include lane lines, traffic signs, stop lines, zebra crossings, street lamps, etc. The original map data is point cloud data, that is, composed of a large number of points reflecting the appearance of elements. The coordinate information of the points of the road elements on the ground can be calculated by the vehicle based on the current vehicle coordinate system, and the coordinate information of the road elements above the ground (such as traffic signs) can be determined by fusion using the multi-view geometry method (such as the BA (Bundle Adjustment) method). The pose information of the original map data can also be calculated by the vehicle based on the vehicle coordinate system.

[0085] The information of each original map data uploaded on each data collection vehicle can further include trajectory information of the vehicle when collecting the original map data. The trajectory information can be a series of trajectory points constituting a driving path of the vehicle. In the embodiment of the present application, a data collection interval of the data collection vehicle can be set, for example, an image can be collected every 0.1s or 0.2s and uploaded to the cloud. The trajectory information corresponding to the image is the driving path of the vehicle when collecting the image.

[0086] In the embodiment of the present application, the original map data can be crowd-sourced collection data or high-precision map data. Compared with the crowd-sourced collection data, the high-precision map data contains richer semantic information and road state information.

[0087] After obtaining the original map data, the original map data can be fused and spliced, and the result of the fusion and splicing can be used as a preliminary mapping result. Generally, there is an overlapping part between two adjacent images collected by the collection device in the original map data collected by the collection device. Therefore, as a specific implementation, each original map data can be fused and spliced based on the information of each original map data to obtain the preliminary mapping result. The preliminary mapping result contains a plurality of road segments (edges), connection points (nodes) between the road segments, and semantic information of each road segment.

[0088] Of course, in the embodiment of the present application, after obtaining the result of the fusion and splicing of the original map data, the result can be preliminarily optimized, and the result of the preliminary optimization can be used as the preliminary mapping result. Generally, in the process of preliminary optimization, each road segment is optimized separately, and the connection relationship between the road segments is not considered. Therefore, the preliminary mapping result still has a situation that the road segments cannot be well connected. Therefore, in the embodiment of the present application, each road segment and the connection point between the road segments can be optimized by the local map at the connection point.

[0089] In the embodiment of the present application, when each road segment and the connection point between the road segments are optimized, the processing is performed in units of “connection”. A connection is composed of two adjacent road segments and the connection point between them. The connection point (node) represents the boundary between the road segments. As shown in FIG. 1, the adjacent target road segment 1, the target road segment 2, and the target connection point therebetween constitute a target connection. Figure 1b Therefore, in step S120, the information of each target road segment in the preliminary mapping result and the information of the target connection point between the adjacent target road segments are obtained, that is, a plurality of target connections are obtained.

[0090] After the above plurality of target connections are acquired, for each target connection, target original map data including the connection point in the connection can be acquired from the original map data based on the target connection point information contained in the connection. As a specific implementation, since the coordinate information of the point in the original map data is contained in each of the above original map data information, each target original map data node_RMS containing the target connection point coordinates can be acquired based on the coordinate information of the target connection point. The above each target original map data is the original local map at the target connection point. In the embodiment of the application, in order to reduce the data processing amount, for each target original map data, only the points indicating the element information and the corresponding semantic information, coordinate information, and pose information can be retained for subsequent processing.

[0091] Next, the above target road sections can be graph optimized based on the above target original map data, so that the above target road sections can be better connected together.

[0092] Graph optimization is to express a conventional optimization problem in the form of a graph. A graph is a structure composed of vertices and edges, where the edge can be referred to as a constraint relationship between the vertices. In the embodiment of the application, graph optimization is to optimize the target road sections and the target original map data based on the graph.

[0093] Generally, the element information and the pose of a road section can be quite different at different positions. Therefore, in order to more accurately repair the road section based on the original local map, in the embodiment of the application, after the above target original map data is acquired, the original map data can be segmented. For example, the target original map data can be segmented into small segments of a length of 50m or 30m, and the target road section can be optimized based on the small segments.

[0094] As described above, the original map data is the original data collected by the vehicle end, and the data accuracy is often not high. In addition, due to the occlusion in the collection process, the completeness of the original map data cannot be guaranteed. The data of the above target road section is the preliminary result of road network mapping based on the original map data, and compared with the original map data, the data accuracy of the target road section is higher, and the completeness is better. Therefore, if the above target original map data is optimized based on one target original map data each time, since the data of the target road section is aggregated from a plurality of original map data, the data quality of the target original map data is unbalanced, and the effect of graph optimization cannot be completely guaranteed. Therefore, in the embodiment of the application, the above target original map data can be aggregated, and the target road section can be graph optimized based on the aggregated data and the target road section data. In this way, the data quality is more balanced, and the graph optimization can achieve better results.

[0095] As one specific implementation method, based on Figure 1a ,like Figure 2 As shown, before performing map optimization on the original map data of each target and the adjacent target road segments based on the original map data of each target and the adjacent target road segments information, the following steps may also be included:

[0096] Step S240: Aggregate the original map data of each target according to a preset quantity to obtain aggregated map data.

[0097] Accordingly, based on Figure 1a ,like Figure 2 As shown, the above steps S140 and S150 can be further broken down into the following steps:

[0098] Step S141: Based on the aggregated map data and the adjacent target road segment information, perform graph optimization on the aggregated map data and the adjacent target road segments to obtain optimized aggregated map data and optimized adjacent target road segments.

[0099] Step S151: Aggregate the optimized aggregated map data and the optimized adjacent target road segments to obtain the updated adjacent target road segments and the connection relationship between the updated adjacent target road segments.

[0100] In this embodiment of the invention, the preset quantity can be set manually, such as 10, 20, etc. A preset quantity of target raw map data forms a cluster, and aggregation of this cluster yields aggregated map data. Typically, there are many data acquisition vehicles, and a sufficient number of target raw map data, including target connection points, to satisfy the conditions for cluster aggregation. If there are N target raw map data, and each cluster contains k target raw data, then after cluster aggregation, n aggregated map data can be obtained, where n = N / k.

[0101] For the target original map data in each cluster, algorithms such as K-means clustering, DBSCAN, and Gaussian mixture model expectation-maximum clustering can be used to aggregate a preset number of target original map data. In this embodiment of the invention, the DBSCAN algorithm can be used to cluster a preset number of target original map data. As mentioned above, the original map data is point cloud data, and the DBSCAN algorithm only outputs the aggregation result when the point density reaches a preset threshold. Therefore, using the DBSCAN algorithm to cluster a preset number of target original map data can eliminate noise in the target original map data to a certain extent.

[0102] like Figure 3 As shown,Figure 3 A specific example of dividing target original map data into batches for separate aggregation is shown. In this example, N target original map data (node_RMS data) is randomly divided into several batches, each batch having k target original map data. Each data is composed of multiple segments of 50m in length. By clustering aggregation, multiple target original data in each batch can be integrated into one aggregated map data, achieving the reduction of subsequent data processing while improving the accuracy of the original data.

[0103] As described above, each original map data corresponds to a series of trajectory points. When aggregating target original map data, the trajectory points corresponding to the target original map data can be aggregated accordingly to obtain trajectory points corresponding to the aggregated map data. When performing preliminary mapping based on original map data, the trajectory points corresponding to the original map data can also be fused and spliced accordingly to obtain trajectory points corresponding to each segment.

[0104] As an embodiment of the present application, as shown in Figure 4 Based on the above aggregated map data and target segment information, the aggregated map data and the target segment can be graph optimized by the following steps:

[0105] Step S410: The each aggregated map data and the adjacent target segment are cut according to a preset length to obtain each aggregated map slice and each target segment slice.

[0106] The above preset length can be manually set, such as 50m, 30m, etc. For example, the above target segment can be cut every 50m to obtain each target segment slice (block), and each target segment slice has a length of 50m. At the same time, a series of trajectory points corresponding to the target segment can also be cut to obtain trajectory points corresponding to each target segment slice.

[0107] As described above, the above target original map data is point cloud data. When aggregating target original map data, point cloud data in the target original map data is extracted, and each point cloud data is aggregated to obtain aggregated map data which is also point cloud data. As described above, the target segment is composed of a large number of points, and the aggregated map data is point cloud data and is also composed of a large number of points, so the aggregated map data can also be regarded as a segment, and therefore the same method can be used to obtain each aggregated map slice and trajectory points corresponding to each aggregated map slice.

[0108] Step S420: The first trajectory points corresponding to each aggregated map slice and the second trajectory points corresponding to each target segment slice are added to the graph.

[0109] In the embodiment of the present application, one of the trajectory points corresponding to the aggregated map slice and the target road segment slice can be used to represent the corresponding aggregated map slice and the target road segment slice. Generally, the trajectory point used to represent the aggregated map slice and the target road segment slice is referred to as a key frame. Subsequently, the semantics and pose of the trajectory point can be optimized through graph optimization, and the motion probability can be propagated through feature matching to optimize the semantics and pose of each slice, thereby optimizing the aggregated map data and the target road segment.

[0110] The first trajectory point can be the first trajectory point corresponding to each aggregated map slice, and can also be the third trajectory point, the sixth trajectory point, etc. The first trajectory point can be set according to actual needs. Similarly, the second trajectory point can be the first trajectory point, the fourth trajectory point, the fifth trajectory point, etc. corresponding to the target road segment slice. The present application does not make specific limitations on this.

[0111] As described above, a graph is a structure composed of vertices and edges. The first trajectory point and the second trajectory point are vertices in the graph.

[0112] In step S430, edges are added between the first trajectory points corresponding to the same aggregated map data and between the second trajectory points corresponding to the same target road segment in the graph.

[0113] The edges are constraint relationships. In step S430, constraint relationships are added between the first trajectory points corresponding to the same aggregated map data and between the second trajectory points corresponding to the same target road segment in the graph.

[0114] In step S440, based on the aggregated map data and the target road segment information, matching edges are added between the first trajectory points corresponding to different aggregated map data, between the second trajectory points corresponding to different target road segments, and between the first trajectory points and the second trajectory points in the graph.

[0115] Generally, when matching point cloud data, a semantic ICP (Iterative Closest Point) algorithm, an NDT (Normal Distribution Transformation) algorithm, etc. can be used. In the embodiment of the present application, a semantic ICP algorithm can be used to match the corresponding trajectory points based on the aggregated map data and the target road segment information, to obtain the matching edges (i.e., constraint relationships) and add them between the corresponding trajectory points in the graph.

[0116] For example, two trajectory points that correspond to any two different aggregated map data points or target road segments can be matched based on the corresponding aggregated map data or target road segment information. If a matching relationship exists between the two trajectory points, a matching edge can be added between them.

[0117] Step S450: Based on the graph, perform graph optimization on each aggregated map data and the adjacent target road segments to obtain each optimized aggregated map data and the optimized adjacent target road segments.

[0118] In this embodiment of the invention, the GTSAMA graph optimization algorithm can be used to optimize the aggregated map data and the adjacent target road segments based on the graph.

[0119] As mentioned above, the accuracy of raw map data directly acquired by the vehicle-mounted image acquisition device is low, and the accuracy of the pose and coordinate information of each trajectory point corresponding to the raw map data is also correspondingly low. Aggregated map data is obtained by aggregating the target raw map data, and the target road segment is obtained by simply fusion and stitching the raw map data. Compared with the raw map data, the accuracy of the above-mentioned aggregated map data and target road segment is slightly improved, but still low. Correspondingly, the accuracy of the trajectory points corresponding to the aggregated map data and the trajectory points corresponding to the target road segment is also low. Therefore, in this embodiment of the invention, the accuracy can be improved by optimizing the pose of each trajectory point in the above figure, and then the aggregated map data and target road segment can be optimized based on the correspondence between the aggregated map data and the target road segment and the trajectory points in the figure.

[0120] In this embodiment of the invention, the object of graph optimization is the pose of each trajectory point in the graph. After graph optimization, the optimal pose of each trajectory point can be obtained. Then, based on the optimal pose of each trajectory point and the constraint relationship between trajectory points, the pose information of each aggregated map data and the target road segment can be optimized to obtain each optimized aggregated map data and the optimized adjacent target road segments.

[0121] As mentioned above, initial optimization typically focuses on isolated road segments without considering connections between them. Therefore, preliminary mapping often results in poor connections between adjacent road segments, while the probability of problems in the middle sections is relatively low. Thus, in this embodiment of the invention, graph optimization can be performed only on the portions near the target connection points. Afterward, the optimized portions near the target connection points are simply connected to the remaining portions of the target road segments.

[0122] As one specific implementation method, based on Figure 4 ,like Figure 5 As shown, the following steps may also be included before slicing:

[0123] Step S510, for the adjacent target road section, the adjacent target road section is split at a preset distance from the target connection point to obtain an interface area road section.

[0124] In the embodiment of the present application, the preset distance (buffer length) can be manually set, such as 50 m, 60 m, etc. After the adjacent target road section is split at the preset distance from the target connection point, an interface area (junction) road section and a non-interface area road section can be obtained. The interface area road section is the part close to the target connection point, and the non-interface area road section is the part of the target road section other than the interface area road section.

[0125] As shown above Figure 1b The area within the preset distance near the target connection point is referred to as an interface area region, and the area outside the preset distance is referred to as a non-interface area region.

[0126] As Figure 5 shown, the steps S430-S450 can be refined as follows:

[0127] Step S431, edges are added between the first trajectory points corresponding to the same aggregated map data and between the second trajectory points corresponding to the same interface area road section in the graph.

[0128] Step S441, based on the aggregated map data and the target road section information, matching edges are added between the first trajectory points corresponding to different aggregated map data, between the second trajectory points corresponding to different interface area road sections, and between the first trajectory points and the second trajectory points in the graph.

[0129] Step S451, based on the graph, the aggregated map data and the interface area road sections in the adjacent target road section are graph-optimized to obtain the optimized aggregated map data and the optimized interface area road sections.

[0130] As Figure 6 shown, Figure 6 is a schematic diagram of the graph in the embodiment of the present application.

[0131] The graph is composed of different key frames (vertices) and constraint relationships between the key frames (edges between the vertices). As described above, after the data of the target road section is split, a key frame is added to the graph for each section, and a constraint relationship is added between adjacent key frames. After the aggregated map data is split, each section is also represented as a key frame, and a constraint relationship exists between adjacent key frames. If there are n aggregated map data, there are n+2 data in the connection area, and constraint relationships can be added between these data according to matching relationships.

[0132] When all the key frames and constraint relations are added to the graph, an optimization algorithm can be run to obtain the optimal pose of the key frames. Although the target original map data may not be very accurate as a local map, the quality of the global map can be improved as a whole after the target road section is added to the graph at the same time. This improves the absolute accuracy of the data of each section of the target road section, making up for the lack of accuracy of the target road section itself. Moreover, the target original map data may have more complete semantic information at intersections, merging areas, and other sections, which can also improve the completeness of the global map.

[0133] In the embodiments of the present application, after the poses of the target road section and the aggregated map data are optimized, the following steps can be performed, as shown in Figure 7

[0134] In step S710, the optimized aggregated map data and the optimized interface area road section are aggregated to obtain a target aggregated road section.

[0135] In the embodiments of the present application, the optimized aggregated map data and the optimized interface area road section can be aggregated based on the DBSCAN algorithm to obtain the target aggregated road section. The aggregation process is based on the semantic information of the optimized aggregated map data and the optimized interface area road section, including lane lines, poles, traffic signs, etc.

[0136] The target aggregated road section is the optimal road section through the target connection point. Part of it belongs to one of the target road sections in the target connection, and the other part belongs to another target road section.

[0137] In step S720, the target aggregated road section is divided based on the target connection point to obtain two target interface area road sections.

[0138] The target aggregated road section is also point cloud data, which includes points on semantic elements (such as lane lines) and the coordinates, semantics, and pose information of the points. Therefore, the target aggregated road section can be divided based on the coordinates of the target connection point to obtain two target interface road sections. The two target interface road sections belong to two adjacent target road sections in the target connection.

[0139] As shown in Figure 8 Figure 8 The process of aggregating multiple data of the interface area into a set of good quality data after optimization and dividing into two target road sections in the connection is shown. That is, the optimized aggregated map data and the two optimized interface area road sections are aggregated to obtain a target aggregated road section. And the target aggregated road section is divided based on the coordinates of the target connection point to obtain two target interface area road sections.

[0140] ​​Step S730, for each target road section, based on the position of the target interface area road section and the position of the non-interface area road section, connecting the target interface area road section and the non-interface area road section to obtain an updated adjacent target road section and the connection relationship between the updated adjacent target road sections.

[0141] The connection relationship between the updated adjacent target road sections can refer to the position connection relationship and the semantic connection relationship (such as being a crossroad, a lane line, etc.) between adjacent target road sections.

[0142] In the embodiment of the application, when the each aggregated map data and the adjacent target road section are cut according to the preset length to obtain each aggregated map slice and each target road section slice, the preset field can be used to save the target road section identifier corresponding to each target road section slice.

[0143] In a specific example, the id of each aggregated map data and the adjacent target road section before cutting can be reserved through the prv_UUID field.

[0144] Correspondingly, then, for each target road section, based on the position of the target interface area road section and the corresponding target road section identifier, and the position of the non-interface area road section and the corresponding target road section identifier, the target interface area road section and the non-interface area road section are connected to obtain an updated adjacent target road section and the connection relationship between the updated adjacent target road sections.

[0145] As shown in FIG. 7, Figure 9 Figure 9 how to update the connection relationship of the connection of the target road section is shown. As described above, in the embodiment of the application, for the target aggregated road section, the target connection point is cut based on the coordinates to obtain two target interface area road sections, and the target interface area road section is connected with the corresponding non-interface area road section to obtain an updated target road section. That is, for each target road section, the interface area between each target road section is cut. However, the semantic information of the cut interface area originally belongs to one instance (lane line, road identifier), and therefore, the two target interface road sections after cutting have a connection relationship. From the road level, the two roads obtained after cutting have a connection relationship; from the lane level, each semantic after cutting has a connection relationship, such as the lane solid line and the lane dashed line shown in the figure.

[0146] ​It can be seen that in the embodiment of the present application, the positions of the road segments on both sides of the connection point are optimized geometrically, the lane element information of the connection part is supplemented and repaired, the accuracy, completeness and consistency of the crowd-sourced map are improved, the connection relationship between adjacent road segments can be obtained logically, the completeness of the logical layer is improved, and the connection relationship is an indispensable part of the global logical relationship of the crowd-sourced map. The embodiment of the present application optimizes and repairs the global map through the local map, which helps to improve the accuracy, completeness and quality of the entire crowd-sourced map, and provides favorable conditions for the application of the downstream crowd-sourced map.

[0147] In another aspect of the present application, a map construction system is also provided, as shown in the figure, the system can include: a first geometric module 1010, a second geometric module 1020 and a logic module 1030. Figure 10

[0148] The first geometric module 1010 is configured to fuse and splice each original map data to obtain a preliminary mapping result, obtain target road segment information in the preliminary mapping result and target connection point information between adjacent target road segments, wherein any two adjacent target road segments and the target connection point between the adjacent target road segments constitute a target connection, and for each target connection, each target original map data including the target connection point is obtained

[0149] The second geometric module 1020 is configured to, for each target connection, based on each target original map data and the adjacent target road segment information, perform map optimization on each target original map data and the adjacent target road segment to obtain each optimized target original map data and an optimized adjacent target road segment.

[0150] The logic module 1030 is configured to, for each target connection, aggregate each optimized target original map data and the optimized adjacent target road segment to obtain an updated adjacent target road segment and a connection relationship between the updated adjacent target road segments, and generate a target map based on each updated adjacent target road segment and the connection relationship between the updated adjacent target road segments.

[0151] ​The map construction system provided in the embodiments of the present application fuses and splices each original map data to obtain a preliminary mapping result; target road section information in the preliminary mapping result and target connection point information between adjacent target road sections are acquired; wherein, any two adjacent target road sections and the target connection points between the adjacent target road sections form a target connection; for each target connection, each target original map data including the target connection point is acquired; based on the each target original map data and the adjacent target road section information, the each target original map data and the adjacent target road section are graph-optimized to obtain each optimized target original map data and an optimized adjacent target road section; the each optimized target original map data and the optimized adjacent target road section are aggregated to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections; based on the each updated adjacent target road section and the connection relationship between the each updated adjacent target road section, a target map is generated. According to the embodiments of the present application, for each adjacent target road section in the preliminary mapping result, target map original map data containing the connection points between the adjacent target road sections, that is, original local map data between the target road sections, is acquired. Since the target road section is a preliminary result of mapping according to the original map data, the relative accuracy and the completeness rate are better, and by using the original local map data between the adjacent road sections to optimize each part in each road section in the preliminary mapping result, the missing of the connection between the road sections can be repaired, the connection between the road sections is smooth, and the map accuracy is further improved.

[0152] Figure 11 An embodiment of the present application shows a specific example of the execution process of the above system, which can specifically include:

[0153] The first geometric layer (first geometric module in the embodiments of the present application) acquires N target original data (N times of node_RMS data), two adjacent target road sections (edge1, edge2) for a target connection. The target original data is clustered and aggregated to obtain n aggregated aggregated map data (node_junction); the two adjacent target road sections are divided into an interface area road section (edge_junction) and a non-interface area road section (edge_non_junction). The above results are input to the second geometric layer.

[0154] The second geometry layer (second geometry module in the embodiment of the application) performs graph optimization based on the data to obtain optimized aggregated map data and optimized target road segments. The optimized aggregated map data and the optimized target road segments of the interface area are aggregated to obtain a target aggregated road segment. The target aggregated road segment of the interface area is segmented to obtain a target interface area road segment, and then the target interface area road segment is connected and smoothed with the non-interface area road segment.

[0155] The logic layer (logic module in the embodiment of the application) updates the connection relationship (updates the logic relationship) between the target road segments according to the data generated by the second geometry module. Finally, the target map is generated.

[0156] The embodiment of the application also provides an electronic device, such as Figure 12 As shown in the figure, the electronic device comprises a processor 1201, a communication interface 1202, a memory 1203 and a communication bus 1204, wherein the processor 1201, the communication interface 1202 and the memory 1203 complete mutual communication through the communication bus 1204,

[0157] The memory 1203 is used for storing a computer program.

[0158] The processor 1201 is used for executing the program stored in the memory 1203 to realize the following steps:

[0159] The original map data is fused and spliced to obtain a preliminary mapping result;

[0160] Target road segment information in the preliminary mapping result and target connection point information between adjacent target road segments are obtained; wherein any two adjacent target road segments and the target connection points between the adjacent target road segments constitute a target connection.

[0161] For each target connection, each target original map data including the target connection point is obtained.

[0162] For each target connection, the target original map data and the adjacent target road segment are graph-optimized based on the target original map data and the adjacent target road segment information to obtain optimized target original map data and optimized adjacent target road segments.

[0163] For each target connection, the optimized target original map data and the optimized adjacent target road segments are aggregated to obtain updated adjacent target road segments and connection relationships between the updated adjacent target road segments.

[0164] A target map is generated based on the updated adjacent target road segments and the connection relationships between the updated adjacent target road segments.

[0165] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0166] The communication interface is used for communication between the electronic device and other devices.

[0167] The memory can include a Random Access Memory (RAM) and can also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located away from the aforementioned processor.

[0168] The processor mentioned above can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0169] In another embodiment provided by the present application, a computer readable storage medium is also provided, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the steps of any of the above map construction methods.

[0170] In another embodiment provided by the present application, a computer program product containing instructions is also provided, and when the computer program product is run on a computer, the computer is caused to execute any of the map construction methods in the above embodiments.

[0171] In the embodiments described above, all or some of the steps can be implemented by software, hardware or firmware, or any combination thereof. When implemented by software, all or some of the steps can be implemented in the form of one or more computer programs. The computer program can be stored in any computer readable medium, and loaded into the computer for execution. The computer readable medium includes computer storage media and communication media. The computer storage media includes any tangible or physical medium for storing or transmitting the program. The computer storage media can be a volatile (such as RAM) or non-volatile (such as ROM, disk, or CD) storage medium. The communication media typically include computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism, and includes any information delivery media. The computer readable medium can be a computer program product.

[0172] It should be noted that, in the present document, the terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Also, the terms "comprising", "containing", or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article or apparatus that comprises a list of elements does not only include those elements, but also other elements not expressly listed, or other elements inherent in such process, method, article or apparatus. Without more limitations, an element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0173] Each of the embodiments in the present document is described in a related manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the system, electronic device, storage medium and program product embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0174] The above merely describes the preferred embodiments of the present application, but is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A map construction method characterized by comprising: The method comprises: fuse and splice each original map data to obtain a preliminary mapping result; obtain target road section information in the preliminary mapping result and target connection point information between adjacent target road sections; wherein, any two adjacent target road sections and the target connection points between the adjacent target road sections form a target connection; for each target connection, obtain each target original map data including the target connection point; for each target connection, based on the target original map data and the adjacent target road section information, graph optimization is performed on the target original map data and the adjacent target road section to obtain optimized target original map data and optimized adjacent target road section; for each target connection, the optimized target original map data and the optimized adjacent target road section are aggregated to obtain updated adjacent target road sections and connection relationships between the updated adjacent target road sections; based on each updated adjacent target road section and the connection relationship between each updated adjacent target road section, a target map is generated.

2. The method of claim 1, wherein, Before the graph optimization of the target original map data and the adjacent target road section information, the method further comprises: aggregating the target original map data according to a preset number to obtain aggregated map data; the graph optimization of the target original map data and the adjacent target road section information comprises: based on the aggregated map data and the adjacent target road section information, graph optimization is performed on the aggregated map data and the adjacent target road section to obtain optimized aggregated map data and optimized adjacent target road section; the aggregation of the optimized target original map data and the optimized adjacent target road section comprises: the aggregation of the optimized aggregated map data and the optimized adjacent target road section to obtain updated adjacent target road sections and connection relationships between the updated adjacent target road sections.

3. The method of claim 2, wherein, The aggregated map data and the adjacent target road section correspond to a plurality of trajectory points respectively; the graph optimization of the aggregated map data and the adjacent target road section information comprises: the aggregated map data and the adjacent target road section are cut according to a preset length to obtain aggregated map slices and target road section slices; the first trajectory points corresponding to the aggregated map slices and the second trajectory points corresponding to the target road section slices are added to a graph; adding edges between first trajectory points corresponding to the same aggregated map data and between second trajectory points corresponding to the same target road segment in the graph; adding matching edges between first trajectory points corresponding to different aggregated map data, between second trajectory points corresponding to different target road segments, and between first trajectory points and second trajectory points in the graph based on the graph, the aggregated map data, and the target road segment information; performing graph optimization on the aggregated map data and the adjacent target road segments based on the graph to obtain optimized aggregated map data and optimized adjacent target road segments.

4. The method of claim 3, wherein, Before the step of splitting the aggregated map data and the adjacent target road segments according to the preset length to obtain aggregated map slices and target road segment slices, the method further includes: splitting the adjacent target road segment at a preset distance from the target connection point to obtain an interface area road segment, wherein the interface area road segment is a portion close to the target connection point; the step of adding edges between first trajectory points corresponding to the same aggregated map data and between second trajectory points corresponding to the same target road segment in the graph includes: adding edges between first trajectory points corresponding to the same aggregated map data and between second trajectory points corresponding to the same interface area road segment in the graph; the step of adding matching edges between first trajectory points corresponding to different aggregated map data, between second trajectory points corresponding to different target road segments, and between first trajectory points and second trajectory points in the graph includes: adding matching edges between first trajectory points corresponding to different aggregated map data, between second trajectory points corresponding to different interface area road segments, and between first trajectory points and second trajectory points in the graph; the step of performing graph optimization on the aggregated map data and the adjacent target road segments based on the graph to obtain optimized aggregated map data and optimized adjacent target road segments includes: performing graph optimization on the aggregated map data and the interface area road segments in the adjacent target road segments based on the graph to obtain optimized aggregated map data and optimized interface area road segments.

5. The method of claim 4, wherein, splitting the adjacent target road segment further obtains a non-interface area road segment; the step of aggregating the optimized aggregated map data and the optimized adjacent target road segments to obtain updated adjacent target road segments and connection relationships between the updated adjacent target road segments includes: aggregating the optimized aggregated map data and the optimized interface area road segments to obtain target aggregated road segments; splitting the target aggregated road segments based on the target connection points to obtain two target interface area road segments; for each target road segment, connecting the target interface area road segment and the non-interface area road segment based on positions of the target interface area road segment and positions of the non-interface area road segment to obtain updated adjacent target road segments and connection relationships between the updated adjacent target road segments.

6. The method of claim 5, wherein, the target road segment information includes a target road segment identifier; The cutting the each aggregated map data and the adjacent target road section according to the preset length obtains each aggregated map slice and each target road section slice, and further includes: A preset field is used to save the target road section identifier corresponding to each target road section slice; The target interface area road section and the non-interface area road section are connected based on the position of the target interface area road section and the position of the non-interface area road section, to obtain the updated adjacent target road section and the connection relationship between the updated adjacent target road sections. The target interface area road section and the non-interface area road section are connected based on the position of the target interface area road section and the position of the non-interface area road section, to obtain the updated adjacent target road section and the connection relationship between the updated adjacent target road sections.

7. A map construction system characterized by comprising: It includes: A first geometric module, a second geometric module and a logic module; The first geometric module is used to fuse and splice each original map data to obtain a preliminary mapping result, obtain target road section information in the preliminary mapping result and target connection point information between adjacent target road sections, any two adjacent target road sections and the target connection point between the adjacent target road sections constitute a target connection, and each target original map data including the target connection point is obtained for each target connection; The second geometric module is used to perform graph optimization on each target original map data and the adjacent target road section based on the each target original map data and the adjacent target road section information for each target connection, to obtain each optimized target original map data and an optimized adjacent target road section; The logic module is used to aggregate each optimized target original map data and the optimized adjacent target road section for each target connection, to obtain an updated adjacent target road section and a connection relationship between the updated adjacent target road sections, and generate a target map based on each updated adjacent target road section and the connection relationship between the updated adjacent target road sections.

8. An electronic device, comprising: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus; The memory is used to store a computer program; The processor is used to execute the program stored on the memory, to realize the method steps in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer program stored in the computer readable storage medium is executed by the processor to realize the method steps in any one of claims 1-6.

Images