Differential updating method, device and equipment for map data

Abstract

The invention discloses a differential updating method, device and equipment for map data, and the method comprises the steps: determining a historical data set of a first semantic object based on a counting attribute in an attribute segment of the first semantic object in historical data and a recent data list; clustering the historical data set of the first semantic object and a second semantic object in the new data to obtain a plurality of clusters formed by clustering of the first semantic object and / or the second semantic object; and aggregating clusters formed by clustering, and performing differential updating on the map data by using the aggregated semantic objects. According to the method, the attribute segment of the semantic object is fully utilized to improve the weight of the historical data so as to increase the density of the historical data of the current position, so that the contribution of the historical data is fully considered during clustering and aggregation; the method solves the problems that local sensing data in historical data is less, clustering fitting of semantic objects cannot be achieved, and high-precision map changes are large and mapping results are unstable due to sensing data differences.

Description

Technical Field

[0001] The present invention relates to the field of geographic information technology, and in particular to a differential updating method, device and equipment for map data. Background Art

[0002] High-precision maps can be used for autonomous driving, and the recognition and production of road signs are important components of generating high-precision maps. Road signs (semantic objects) are generally identified and extracted through road image information collected by cameras mounted on vehicles.

[0003] In the process of making high-precision maps, in order to solve the drawback of large resource investment in single-trip closed-loop collection (closed-loop single-trip collection) and optimize application scenarios, the inventor previously proposed a solution for making high-precision maps based on multi-source crowdsourcing data, that is, by using private cars as collection equipment, data collection is performed without interfering with the stable and fast driving of the vehicle. In the process of using private cars to collect data, since it is impossible to stipulate that a certain vehicle strictly follows a closed-loop trajectory, most vehicle trajectories are only one-way trajectories from the starting point to the destination, so multiple single-track data is used in the later production of high-precision maps, that is, multiple / multiple image collections are required for the same location, and then the road sign information in each collected image is extracted separately, and finally the same road sign collected in multiple trips is fused, and then a high-precision map is produced based on multi-source crowdsourcing data.

[0004] After the high-precision map is produced, it will also face the problem of map updating during use. There are two common ways. One is to overlay on the historical data, that is, the deletion scenario of historical data is not covered, which will inevitably cause the historical database to become infinitely large; the second is to update based on data differentials. Although the historical database is no longer infinitely large, due to the aggregation nature and fusion characteristics of semantic objects, direct differential updates cannot be applied to scenarios with multi-source crowdsourcing data; for example, when updating, the focus is on how to use the data that has not been successfully clustered, without considering the historical data and the weight of the historical data; each time an update is made, the perceived data is not retained, resulting in slow updates of historical data when the actual road data changes. Summary of the invention

[0005] In order to update high-precision maps produced based on multi-source crowdsourcing data, so as to fully consider the contribution of historical data to semantic objects, and thereby enrich technical routes and increase selection space, a differential update method, device and equipment for map data are provided in an embodiment of the present invention.

[0006] In a first aspect, an embodiment of the present invention provides a method for differential updating of map data, which may include:

[0007] Determining a historical data set of the first semantic object based on a count attribute in an attribute segment of the first semantic object in the historical data and a recent data list;

[0008] Clustering the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object;

[0009] The clusters formed by clustering are aggregated, and the map data is differentially updated with the aggregated semantic objects.

[0010] Optionally, determining the historical data set of the first semantic object based on the count attribute in the attribute segment of the first semantic object in the historical data and the recent data list may include:

[0011] Based on the recent historical data in the recent data list, the recent historical data is divided into several files according to the chronological order to obtain the recent historical data of several files;

[0012] Based on the counting attribute and a counting ratio threshold preset for each bin, the recent historical data in each bin is copied respectively to determine a historical data set of the first semantic object.

[0013] Optionally, the grading in chronological order to obtain several gradations of recent historical data may include:

[0014] Sorting the recent historical data from new to old in time, and dividing the recent historical data into several files in time from new to old, so as to obtain several files of recent historical data;

[0015] The preset counting ratio threshold of the bins composed of recent historical data with new time is greater than the preset counting ratio threshold of the bins composed of recent historical data with old time.

[0016] Optionally, aggregating the clusters formed by clustering and performing differential update on the map data using the aggregated semantic objects may include:

[0017] Determine whether the cluster formed by clustering contains the historical data;

[0018] If the cluster only includes the first semantic object, updating the attribute segment of the first semantic object;

[0019] If the cluster only includes the second semantic object, aggregating a new cluster formed by aggregating the second semantic objects, so as to update the map data with the aggregated new semantic objects;

[0020] If the cluster includes the first semantic object and the second semantic object, a mixed cluster formed by the aggregation of the first semantic object and the second semantic object is aggregated to perform a differential update between the aggregated new semantic object and the first semantic object.

[0021] Optionally, the aggregating the mixed cluster formed by aggregating the first semantic object and the second semantic object to differentially update the aggregated new semantic object and the first semantic object may include:

[0022] Determining the number of the first semantic objects included in the mixed cluster;

[0023] If the first semantic object is one, the hybrid cluster is used as a first hybrid cluster, and the first hybrid cluster inherits the ID of the first semantic object, so as to form a new semantic object by aggregating the first hybrid cluster, and differentially update the first semantic object in the map data based on the ID;

[0024] If there are at least two first semantic objects, the mixed cluster is used as a second mixed cluster, and the second mixed cluster inherits the IDs of at least two of the first semantic objects to form a new semantic object by aggregating the second mixed clusters, and differentially update at least two first semantic objects in the map data based on the IDs.

[0025] Optionally, after performing differential updating on the first semantic object in the map data based on the ID, the method may further include:

[0026] Selecting an object from the second semantic object that has the highest similarity to the new semantic object formed by the aggregation of the first mixed cluster, and adding it to the recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the new semantic object includes: updating the count attribute, updating the type confidence, updating the existence confidence, and updating the recent data list;

[0027] After differentially updating at least two first semantic objects in the map data based on the ID, the method further includes: selecting an object from the second semantic objects that has the highest similarity to a new semantic object formed by aggregation of the second mixed cluster, and adding the object to a recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the semantic object includes: merging the attribute segments of at least two first semantic objects, and updating the merged attribute segment based on the newly added second semantic object.

[0028] Optionally, updating the attribute segment of the first semantic object includes:

[0029] Based on the number of newly added segments in the group where the first semantic object is located, the existence confidence of the first semantic object is reduced to update the attribute segment of the first semantic object.

[0030] Optionally, the method may further include: comparing a preset existence confidence threshold with the existence confidence in the updated semantic object attribute segment, and deleting semantic objects corresponding to existence confidences lower than the existence confidence threshold.

[0031] Optionally, the method may further include: combining the second semantic objects that have not been successfully clustered into a new segment data for storage.

[0032] Optionally, before clustering the first semantic object in the historical data and the second semantic object in the new data, the following steps may be further included:

[0033] Acquire new data, wherein the new data includes at least one newly added single trip data;

[0034] Based on the association relationship between the lane and the semantic object in the newly added single trip data, calibrating the position relationship of the semantic object in the newly added single trip data;

[0035] Segmenting the calibrated newly added single trip data based on the position of the preset mark in the newly added single trip data to obtain a plurality of segment data after the newly added single trip data is segmented; wherein each segment data includes boundary information, vehicle lanes and semantic objects;

[0036] Determine the correspondence between each segment data in the newly added single-pass data and the group data composed of the segment data at the same position after the multi-pass data in the historical data are segmented, so as to cluster the second semantic object in each segment data in the newly added single-pass data with the first semantic object in the group data in the historical data corresponding to the segment data.

[0037] In a second aspect, an embodiment of the present invention provides a map data differential update device, which may include:

[0038] a determination module, configured to determine a historical data set of the first semantic object based on a count attribute in an attribute segment of the first semantic object in the historical data and a recent data list;

[0039] A clustering module, configured to cluster the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object;

[0040] The updating module is used to aggregate the clusters formed by clustering and perform differential update on the map data with the aggregated semantic objects.

[0041] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the differential update method for map data as described in the first aspect.

[0042] In a fourth aspect, an embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the differential update method for map data as described in the first aspect is implemented.

[0043] The beneficial effects of the above technical solution provided by the embodiment of the present invention include at least:

[0044] The embodiments of the present invention provide a differential update method, device and equipment for map data. The differential update method has the following beneficial effects: first, the method makes full use of the attribute segments of semantic objects, and improves the weight of historical data through the attribute segments, so as to increase the density of historical data at the current location, and then fully considers the contribution of historical data during clustering and aggregation; second, it solves the problem that there is little local perception data in historical data, and it is impossible to cluster and fit into semantic objects and produce high-precision maps; third, it solves the problem that there is little local perception data in new data, and it is impossible to update and produce high-precision maps separately; fourth, it solves the problem that the high-precision map changes are large and the mapping results are unstable due to the differences in perception data in multiple passes of data.

[0045] Furthermore, this updating method solves the technical problems of uncertain data collection scope, uncertain data collection quantity, and unstable data collection quality of multi-source crowdsourcing data, and uses historical data to regulate and constrain the final generated high-precision map. Compared with a single-production high-precision map, its data range is more complete, has a higher fault tolerance, and can reflect changes in single data, as well as reflect real road changes from cumulative changes.

[0046] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0047] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0049] Figure 1A flowchart of a differential updating method for map data provided in an embodiment of the present invention;

[0050] Figure 2 A flowchart of determining a historical data set of a first semantic object provided in an embodiment of the present invention;

[0051] Figure 3 A specific flow chart of differential updating of map data provided in an embodiment of the present invention;

[0052] Figure 4 A flowchart of preprocessing single-trip data provided in an embodiment of the present invention;

[0053] Figure 5 One of the flowcharts of the detailed map data differential updating method provided in the embodiment of the present invention;

[0054] Figure 6 This is a second flowchart of a detailed map data differential updating method provided in an embodiment of the present invention;

[0055] Figure 7 A schematic diagram of segment data provided in an embodiment of the present invention;

[0056] Figure 8 A schematic diagram of line production provided in an embodiment of the present invention;

[0057] Fig. 9 A schematic diagram of multiple data packets provided in an embodiment of the present invention;

[0058] Fig.10 An example of an update process provided in an embodiment of the present invention;

[0059] Fig.11 It is a schematic diagram of the structure of a differential updating device for map data provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0061] The inventors found the following problems when updating the high-precision map after actually making a high-precision map based on multi-source crowdsourcing data: (1) There are few local perception data (semantic objects, etc.) in the historical data, and clustering and mapping output cannot be achieved. When making a high-precision map, it is necessary to accumulate new data, and then cluster the perception data at the location and map it; (2) There are enough local perception data in the historical data to cluster it into a map, but there are few new data, and the new data alone cannot be clustered into a map; (3) There are perception differences between the historical data and the new data, such as differences in data type, data shape, and data location, which leads to large changes in clustering results and unstable clustering results. In view of the above problems, the present invention is proposed to provide a differential update method, device and equipment for map data that overcomes the above problems or at least partially solves the above problems. The method is applicable to high-precision maps made based on multi-source crowdsourcing data. It should be noted that the above clustering mapping mainly refers to clustering into specific semantic objects (roadside signs), such as lane lines, straight arrows, turn arrows, sidewalks, etc.

[0062] In an embodiment of the present invention, a method for differential updating of map data is provided. Figure 1 As shown, the method may include the following steps:

[0063] Step S11: Determine a historical data set of the first semantic object based on the count attribute in the attribute segment of the first semantic object in the historical data and the recent data list.

[0064] Step S12: clustering the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object.

[0065] Step S13: Aggregate the clusters formed by clustering, and perform differential update on the map data with the aggregated semantic objects.

[0066] It should be noted that a single semantic object in historical data is fitted by objects in multiple passes of data, and it can represent multiple geometric bodies. In order to facilitate the distinction and explanation of semantic objects in historical data and new data, in the embodiment of the present invention, the semantic object in historical data is called the first semantic object, and the semantic object in new data is called the second semantic object. The above-mentioned first semantic object and second semantic object have different attributes (historical data attributes and new data attributes). Those skilled in the art can understand that there are several semantic objects in historical data and several semantic objects in new data, and they are not limited to one semantic object data.

[0067] The embodiment of the present invention can cluster the semantic objects in a segment of data. Each segment of data corresponding to each pass can have multiple semantic objects. If the segment of data is historical data, the semantic objects in the segment of data are all first semantic objects; if the segment of data is new data, the semantic objects in the segment of data are all second semantic objects. Due to the difference in signal quality during acquisition, etc., the semantic objects in a single pass of data are missing. Therefore, in the clustering process, the cluster initially clustered at the same position may only contain the first semantic object, or only contain the second semantic object, or of course, may contain the first semantic object and the second semantic object.

[0068] In the above step S11 in the embodiment of the present invention, when clustering, the first semantic object in the historical data uses the count attribute to express the weight of the semantic object. The purpose of this is to increase the density of the current position data (semantic object) by count when clustering, so as to fully consider the role of historical data in clustering. In the embodiment of the present invention, it should also be noted that due to the huge volume of historical data, in order to save storage space and improve operating computing power, the original semantic objects in all historical data in this embodiment are not all stored, that is, only part of the historical data is stored in the recent data list (recent_list). For example, the recent data list storage limit for each semantic object in this embodiment is 20 data. It should be further noted that the original data in the above recent data list is the semantic object (such as length, area, position, etc.) that is most similar to the aggregation result of each clustering, so that it can best characterize the characteristics of historical data of different clustering processes.

[0069] The differential update method for the above-mentioned map data provided in an embodiment of the present invention utilizes the attribute segments of the semantic objects in the historical data to form a historical data set of the first semantic object, and improves the weight of the historical data through the attribute segments to increase the density of the historical data of the current location, thereby fully considering the contribution of the historical data during clustering and aggregation; the update method solves the technical problems of uncertain data collection scope, uncertain data collection quantity, and unstable data collection quality of multi-source crowdsourcing data, and utilizes historical data to regulate and constrain the high-precision map finally generated; compared with a high-precision map produced once, its data range is more complete and has a high fault tolerance, and it can reflect the changes in single data, as well as the real road surface changes from the cumulative changes.

[0070] In an optional embodiment, in the above step S11, the process of determining the historical data set of the first semantic object can be to divide the data in the recent data list according to count, with the data having a long history having a low proportion and the data having a new history having a high proportion. When the data having a high proportion is formed into a set, the number of times it is used will increase. Figure 2 As shown, the following steps may be specifically included:

[0071] Step S21: Based on the recent historical data in the recent data list, the recent historical data is divided into several files in chronological order to obtain a plurality of files of recent historical data.

[0072] During the specific implementation of this step, the recent historical data is sorted from new to old in time, and the recent historical data is divided into several grades in time from new to old to obtain several grades of recent historical data; wherein the preset counting ratio threshold of the grades composed of recent historical data with new time is greater than the preset counting ratio threshold of the grades composed of recent historical data with old time.

[0073] In a specific example, the data in the recent data list are sorted from new to old according to time, for example, divided into three levels. The division can be done by dividing the number of data in the recent data list by three. If the number is not divisible by three, the number of data in the second and third levels is the integer rounded down after dividing by three, and the rest is the number of data in the first level. The calculation process is as follows: If there are n data in the recent data list, the number of data in the first level is n- The number of data in the second and third tiers is If n is 2 in the recent data list, then the number of data in the first file is 2, and the number of data in the second and third files is 0; if n is 10 in the recent data list, then the number of data in the first file is 4, and the number of data in the second and third files is 3.

[0074] Step S22: based on the counting attribute and the counting ratio threshold preset for each bin, the recent historical data in each bin is copied respectively to determine the historical data set of the first semantic object.

[0075] In this step, each data in each bin has a count ratio threshold. In the embodiment of the present invention, in order to increase the weight of the latest data, when setting the above thresholds, the count ratio threshold in the first bin, the count ratio threshold in the second bin, and the count ratio threshold in the third bin are sorted from large to small, that is, the first bin accounts for a larger proportion in count than the second bin, the second bin accounts for a larger proportion in count than the third bin, and the cumulative sum of the count ratio thresholds of each bin is 1.

[0076] In one example, the count ratio thresholds of the above three bins can be 50%, 30% and 20% respectively. The calculation process is as follows: if count is m, the number of the first bin is: The number of the second tier is: The number of the third tier is: If the counting attribute count in the attribute segment is 7, the count in the first gear is 4, the count in the second gear is 2, and the count in the third gear is 1; if the counting attribute count in the attribute segment is 15, the count in the first gear is 8, the count in the second gear is 4, and the count in the third gear is 3.

[0077] In the embodiment of the present invention, after the data in each file is determined in step S21 and the count ratio threshold of each file is determined in step S22, the historical data set of the first semantic object can be copied according to the count attribute. The following is a specific example of this embodiment:

[0078] There are 5 data in the recent data list of the first semantic object in a historical data. After sorting from new to old in time, they are [D1, D2, D3, D4, D5], which are divided into three levels: [D1, D2, D3], [D4], [D5]. The count of the first semantic object is 15, and the final number of each group is 8, 4, and 3.

[0079] The first file of data is copied cyclically to obtain 8 data, that is, the first group is: [D1,D2,D3,D1,D2,D3,D1,D2]; the data in the second file is copied cyclically to obtain 4 data, that is, the second group is: [D4,D4,D4,D4]; the data in the third file is copied cyclically to obtain 3 data, that is, the third group is: [D5,D5,D5], and finally the historical data set of the first semantic object is: [D1,D2,D3,D1,D2,D3,D1,D2,D4,D4,D4,D4,D5,D5,D5].

[0080] In the embodiment of the present invention, the process of determining the historical data set of the first semantic object fully considers the historical data time factor and the historical clustering result factor. It is not necessary to store all the historical data. It is only necessary to retain the historical data in the historical clustering that is most similar to the aggregation result. This saves storage space and can represent the changes in each clustering in the historical data. Furthermore, in the process of constructing the historical data set of the above-mentioned first semantic object, the proportion of new historical data is increased, and the proportion of old historical data is reduced. When the constructed historical data set is clustered with the second semantic object of the new data, the density of the historical data is improved, and the clustering results and subsequent aggregation graphs are more in line with the elements on the actual road.

[0081] In order to improve computing power and save storage space in the embodiment of the present invention, the count value range is 1-100, and it will not increase after accumulating to 100. Only 20 data in the recent data list are saved. According to the first-in-first-out principle, after reaching 20, more data is added to it, and the first data added to the list is deleted. In a specific example, the attribute segment of the first semantic object in the historical data can be shown in Table 1 as follows:

[0082] Table 1 Example table of semantic object attribute segments

[0083]

[0084] In another optional embodiment, the above step S13 of performing differential update on the map data may specifically be parameter Figure 3 As shown, the following steps may be included:

[0085] Step S31, determine whether the cluster formed by clustering contains historical data; if the cluster only contains the first semantic object, execute step S32; if the cluster only contains the second semantic object, execute step S33; if the cluster contains the first semantic object and the second semantic object, execute step S34.

[0086] Step S32: Update the attribute segment of the first semantic object.

[0087] Step S33: Aggregate the newly added clusters formed by aggregating the second semantic objects, so as to update the map data with the aggregated newly added semantic objects.

[0088] Step S34: Aggregate the mixed cluster formed by aggregating the first semantic object and the second semantic object, so as to perform a differential update between the aggregated new semantic object and the first semantic object.

[0089] When performing differential update in step S34, first determine the number of first semantic objects contained in the hybrid cluster; if the number of first semantic objects is one, the hybrid cluster is taken as the first hybrid cluster, and the first hybrid cluster inherits the ID of the first semantic object, so as to aggregate the first hybrid cluster into a new semantic object, and perform differential update on the first semantic objects in the map data based on the ID; if the number of first semantic objects is at least two, the hybrid cluster is taken as the second hybrid cluster, and the second hybrid cluster inherits the ID of at least two first semantic objects, so as to aggregate the second hybrid cluster into a new semantic object, and perform differential update on at least two first semantic objects in the map data based on the ID.

[0090] Step S35: Update the attribute segment of the semantic object.

[0091] Select an object in the second semantic object that has the highest similarity to the new semantic object formed by the first mixed cluster, and add it to the recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the new semantic object includes: updating the count attribute, updating the type confidence, updating the existence confidence and updating the recent data list;

[0092] After differentially updating at least two first semantic objects in the map data based on the ID, the method further includes: selecting an object from the second semantic objects that has the highest similarity to a new semantic object formed by aggregation of the second mixed cluster, and adding the object to a recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the semantic object includes: merging the attribute segments of at least two first semantic objects, and updating the merged attribute segment based on the newly added second semantic object.

[0093] This step increases the existence confidence (N\A confidence) of the semantic object that is successfully fitted to the mixed cluster. The specific method can generate different update and increase schemes depending on whether the multi-pass data is segmented and grouped or not. For example: if the multi-pass data is not segmented and grouped, the preset existence confidence value is directly increased for the semantic object that is successfully fitted to the mixed cluster and other semantic objects within the preset range to increase the existence confidence of the semantic object; if the multi-pass data is grouped and classified, the specific execution method can be as follows: count the number of segments of the new data on the bunch, for example, there are n new segments on the bunch, and the fitting result also inherits the ID of the historical data, then use the N\A confidence of the historical data to perform n times N=N+(100-N)*10% on the fitting result (semantic object), thereby increasing the existence confidence of the semantic object fitted by the mixed cluster.

[0094] After the aggregation is completed, an object (length, area, position) close to the aggregation result is selected from the non-historical data (i.e., new data) in the cluster and put into the recent data list of the aggregation result. In this way, samples of the perception data of each task are retained. When used next time, the recent data list formed by using the retained sample is closer to the actual perception data than directly using historical data. The specific judgment method is: the difference between the length or area and the aggregation result accounts for 50%, and the position distance accounts for 50%. The one with the smallest comprehensive value is the closest, and the count field of the aggregation result is updated.

[0095] In the embodiment of the present invention, updating the attribute segment of the first semantic object may include: based on the number of newly added segments in the group where the first semantic object is located, reducing the existence confidence of the first semantic object to update the attribute segment of the first semantic object.

[0096] When this step is implemented, the specific reduction method can generate different update and increase schemes depending on whether the multi-pass data is segmented and grouped or not. For example, if the multi-pass data is not segmented and grouped, the semantic object that is successfully fitted by the mixed cluster and other semantic objects within the preset range are directly increased by the preset existence confidence value to increase the existence confidence of the semantic object; if the multi-pass data is grouped and classified, the specific implementation method can be as follows: the existence confidence of the semantic object fitted by the historical cluster is reduced by counting the number of new data segments on the bunch. For example, if there are n new segments on the bunch, the current existence confidence (N\A confidence) N of the historical cluster needs to be reduced n times N=N-(100-N)*10%. The purpose of this reduction is to combine the existence confidence with the number of new data, ensuring the consistency of the results of multi-pass one-time calculation and multi-pass multiple calculation.

[0097] In another alternative embodiment, referring to Figure 1 As shown, the method may further include the following steps: Step S14, comparing the existence confidence in the updated semantic object attribute segment with the preset existence confidence threshold, and deleting the semantic objects corresponding to the existence confidences lower than the existence confidence threshold.

[0098] This step compares the fitting results with historical data, updates historical data, checks the existence confidence of historical data (N\A confidence), and deletes historical data with too low confidence values ​​(by setting confidence thresholds); for example, after clustering, there are only historical data in the cluster, and the N\A confidence of the historical data will be reduced. After fitting, before differential update, check all historical data and delete historical data with N\A confidence less than a preset threshold (for example, the threshold can be set to 0). For mixed clusters, when updating, the existence confidence of the fitted semantic objects is increased, and the type confidence is updated. For newly added clusters, they are updated directly, and the semantic objects fitted by the newly added clusters are added to the map data.

[0099] In another alternative embodiment, referring to Figure 1 As shown, the method may further include the following steps: Step S15, combining the second semantic objects that are not successfully clustered into a new segment data for storage.

[0100] The second semantic object that has not been successfully clustered does not mean that the semantic object does not exist in the actual road. It may be due to the accuracy of the collected signal or the newly added road surface signs (semantic objects), which causes some data to fail to be clustered successfully. However, it is impossible to determine whether these data actually exist. Therefore, the semantic objects that have not been successfully clustered will be combined into a new segment data for storage. It should be noted here that the semantic objects that have not been successfully clustered will not be output again, so this type of semantic object will not be displayed in the map data. If clustering can be successful when there is new data updated later, the semantic object formed by the cluster after successful clustering will be output at this time. This embodiment retains the data that has not been successfully clustered for backup, which improves the accuracy of the map data updated later.

[0101] In another optional embodiment, before clustering, the following steps need to be performed, that is, pre-processing the single-pass data included in the new data, referring to Figure 4 As shown, the following steps may be included:

[0102] Step S41: Acquire new data, where the new data includes at least one newly added single trip data.

[0103] Step S42: based on the association relationship between the lanes and the semantic objects in the newly added single trip data, calibrate the position relationship of the semantic objects in the newly added single trip data.

[0104] Step S43, segmenting the calibrated newly added single trip data based on the position of the preset mark in the newly added single trip data to obtain a plurality of segment data after segmenting the newly added single trip data; wherein each segment data includes boundary information, vehicle lanes and semantic objects.

[0105] Step S44, determine the correspondence between each segment data in the newly added single-pass data and the group data composed of the segment data at the same position after the multi-pass data in the historical data are segmented, so as to cluster the second semantic object in each segment data in the newly added single-pass data with the first semantic object in the group data in the historical data corresponding to the segment data.

[0106] Since there are a large number of semantic objects in a single trip of data, and the starting and ending points of the routes of multi-source data are different, it would be very difficult to directly update the entire single trip of data. Therefore, in this embodiment, the single trip data needs to be divided into segments, and the segment data at the same position are used to form a group of data, and the semantic objects in a group of data are updated.

[0107] In a detailed embodiment, the present invention provides a method for differential updating of detailed map data, referring to Figure 5 and Figure 6 As shown, the method may include the following steps:

[0108] Step S501: Acquire new data, where the new data includes at least one newly added single trip data.

[0109] Reference Figure 7 As shown, a single trip of data is composed of several connected segments, each of which contains an identification semantic object and a lane semantic object. Multiple trips of segment data at the same position can be combined into a group of data. When the embodiment of the present invention updates the above map data, a group of data is used as an update processing unit, that is, the semantic objects in the segment data included in the group data are first clustered and fitted, and finally multiple trips of collected data at the same position are fitted into a semantic object for output.

[0110] Step S502: based on the association relationship between the lanes and the semantic objects in the newly added single trip data, calibrate the position relationship of the semantic objects in the newly added single trip data.

[0111] In this step, only the newly added single-trip data is processed, and the single-trip data processing process in the historical data is the same. Only when the processing mechanism is the same can it be divided into the same group of data. The single-trip data processing process is as follows: Based on the line information, that is, the vehicle lane semantic object information, a position relationship is established in the horizontal and vertical directions. In the specific implementation, the vertical vehicle lane semantic object is made into a line (that is, the vertical relationship), and other semantic objects in the single-trip data (identification semantic objects, vehicle lane semantic objects, such as arrows, zebra crossings, etc.) are mounted on the line. Refer to Figure 8 As shown, the vertical thin line is the produced line. In the same frame of data, the absolute relationship between the above-mentioned car line and the semantic object is known, so the position relationship of the semantic object in a single trip of data can be calibrated.

[0112] Step S503, segmenting the calibrated newly added single trip data based on the position of the preset mark in the newly added single trip data to obtain a plurality of segment data after the newly added single trip data is segmented; wherein each segment data includes boundary information, vehicle lanes and semantic objects.

[0113] After the calibration is completed, due to the large amount of data and the different starting and ending points between the single-trip routes, it is not possible to use the entire route for update. At this time, the single-trip data needs to be segmented. In the specific implementation, the single-trip data is divided into several segments according to the locations of preset signs such as intersections, zebra crossings, stop lines, etc.

[0114] Step S504, determine the correspondence between each segment data in the newly added single-pass data and the group data composed of the segment data at the same position after the multi-pass data in the historical data are segmented, so as to cluster the second semantic object in each segment data in the newly added single-pass data with the first semantic object in the group data in the historical data corresponding to the segment data.

[0115] This step is to group the multi-trip data after segmentation, that is, to establish a group relationship for the multi-trip data, to divide multiple single-trip data representing the same section of road into one group, to establish a topological relationship between adjacent groups, and to write an ID to each group based on the longitudinal extension characteristics of the line. Fig. 9 As shown in the figure, the segments at the same position will form a group data. The endpoints of the group data are crosses. The cross represents the location point of the end of the bunch. Through the relationship of the cross, we can determine which bunch it belongs to, and also determine the relationship between adjacent bunches. Fig. 9 As shown, after grouping, the same semantic objects in the group need to be matched and aligned before updating. The purpose of matching is to adjust the geometric position of the semantic objects, and the purpose of alignment is to increase the weight of historical data when clustering after moving the semantic objects.

[0116] Step S505: Based on the recent historical data in the recent data list, the recent historical data is divided into several files in chronological order to obtain a plurality of files of recent historical data.

[0117] This step sorts the recent historical data from new to old in time, and divides the recent historical data into several grades in time from new to old, so as to obtain several grades of recent historical data; wherein the preset counting ratio threshold of the grades composed of recent historical data with new time is greater than the preset counting ratio threshold of the grades composed of recent historical data with old time.

[0118] Step S506: based on the count attribute and the count ratio threshold preset for each bin, the recent historical data in each bin is copied respectively to determine the historical data set of the first semantic object.

[0119] The specific implementation process of the above steps S505 and S506 can refer to the above steps S21 and S22, and the embodiment of the present invention will not be repeated here.

[0120] Step S507: cluster the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object. Fig.10 As shown, Fig.10 The process of clustering and aggregating (fitting) historical data and new data has evolved.

[0121] Step S508: Determine whether the cluster formed by clustering contains historical data.

[0122] Step S509: If the cluster only includes the first semantic object, update the attribute segment of the first semantic object.

[0123] In a specific implementation, based on the number of newly added time segments in the group where the first semantic object is located, the existence confidence of the first semantic object is reduced to update the attribute segment of the first semantic object.

[0124] Step S510: If the cluster only includes the second semantic object, then the newly added cluster formed by the aggregation of the second semantic objects is aggregated, so as to update the map data with the aggregated newly added semantic objects.

[0125] Step S511: If the cluster includes the first semantic object and the second semantic object, a mixed cluster formed by aggregating the first semantic object and the second semantic object is aggregated to perform a differential update between the aggregated new semantic object and the first semantic object.

[0126] In this step, the number of first semantic objects contained in the mixed cluster is first determined; if the first semantic object is one, the mixed cluster is taken as the first mixed cluster, and the first mixed cluster inherits the ID of the first semantic object, so as to aggregate the first mixed cluster into a new semantic object, and differentially update the first semantic objects in the map data based on the ID; if the first semantic objects are at least two, the mixed cluster is taken as the second mixed cluster, and the second mixed cluster inherits the IDs of at least two first semantic objects, so as to aggregate the second mixed cluster into a new semantic object, and differentially update at least two first semantic objects in the map data based on the ID.

[0127] The above steps S507 to S511 may refer to the above steps S31 to S35, and the embodiments of the present invention will not be described in detail herein.

[0128] Step S512: compare the existence confidence in the updated semantic object attribute segment with the preset existence confidence threshold, and delete the semantic objects corresponding to the existence confidence lower than the existence confidence threshold.

[0129] The above steps can be implemented with reference to the above step S14, which will not be described in detail here.

[0130] Step S513: combine the unclustered second semantic objects into a new segment data for storage.

[0131] The above steps can be implemented with reference to the above step S15, which will not be described in detail here.

[0132] The above-mentioned detailed differential update method of map data provided in the embodiment of the present invention has the following beneficial effects: first, the method makes full use of the attribute segments of semantic objects, and improves the weight of historical data through the attribute segments, so as to increase the density of historical data of the current location, and then fully considers the contribution of historical data during clustering and aggregation; second, it solves the problem that there is little local perception data in historical data, and it is impossible to cluster and fit it into semantic objects and produce high-precision maps; third, it solves the problem that there is little local perception data in new data, and it is impossible to update it separately and produce high-precision maps; fourth, it solves the problem that the high-precision map changes are large and the mapping results are unstable due to the differences in perception data in multiple trips of data.

[0133] Furthermore, this updating method solves the technical problems of uncertain data collection scope, uncertain data collection quantity, and unstable data collection quality of multi-source crowdsourcing data, and uses historical data to regulate and constrain the final generated high-precision map. Compared with a single-production high-precision map, its data range is more complete, has a higher fault tolerance, and can reflect changes in single data, as well as reflect real road changes from cumulative changes.

[0134] Based on the same inventive concept, a differential updating device for map data is provided in an embodiment of the present invention. Fig.11 As shown, it may include:

[0135] The determination module 111 is used to determine a historical data set of the first semantic object based on the count attribute in the attribute segment of the first semantic object in the historical data and the recent data list;

[0136] The clustering module 112 is used to cluster the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object;

[0137] The aggregation module 113 is used to aggregate the clusters formed by clustering;

[0138] The updating module 114 is used to perform differential update on the map data using the aggregated semantic objects.

[0139] In an optional embodiment, the determination module 111 is used to: divide the recent historical data in the recent data list into files according to the chronological order to obtain several files of recent historical data;

[0140] Based on the counting attribute and a counting ratio threshold preset for each bin, the recent historical data in each bin is copied respectively to determine a historical data set of the first semantic object.

[0141] In a specific embodiment, the determination module 111 is also used to: sort the recent historical data from new to old in time, and divide the recent historical data into several files in time from new to old to obtain several files of recent historical data; wherein the preset counting ratio threshold of the files composed of recent historical data with new time is greater than the preset counting ratio threshold of the files composed of recent historical data with old time.

[0142] In an optional embodiment, the updating module 114 is specifically configured to:

[0143] Determine whether the cluster formed by clustering contains the historical data;

[0144] If the cluster only includes the first semantic object, updating the attribute segment of the first semantic object;

[0145] If the cluster only includes the second semantic object, aggregating a new cluster formed by aggregating the second semantic objects, so as to update the map data with the aggregated new semantic objects;

[0146] If the cluster includes the first semantic object and the second semantic object, a mixed cluster formed by the aggregation of the first semantic object and the second semantic object is aggregated to perform a differential update between the aggregated new semantic object and the first semantic object.

[0147] In a specific embodiment, the update module 114 may also be used to:

[0148] Determining the number of the first semantic objects included in the mixed cluster;

[0149] If the first semantic object is one, the hybrid cluster is used as a first hybrid cluster, and the first hybrid cluster inherits the ID of the first semantic object, so as to form a new semantic object by aggregating the first hybrid cluster, and differentially update the first semantic object in the map data based on the ID;

[0150] If there are at least two first semantic objects, the mixed cluster is used as a second mixed cluster, and the second mixed cluster inherits the IDs of at least two of the first semantic objects to form a new semantic object by aggregating the second mixed clusters, and differentially update at least two first semantic objects in the map data based on the IDs.

[0151] In a specific embodiment, the update module 114 may also be used to:

[0152] After the first semantic object in the map data is differentially updated based on the ID, the method further includes: selecting an object with the highest similarity to the new semantic object formed by the aggregation of the first mixed cluster from the second semantic object, and adding the object to the recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the new semantic object includes: updating the count attribute, updating the type confidence, updating the existence confidence and updating the recent data list;

[0153] After differentially updating at least two first semantic objects in the map data based on the ID, the method further includes: selecting an object from the second semantic objects that has the highest similarity to a new semantic object formed by aggregation of the second mixed cluster, and adding the object to a recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the semantic object includes: merging the attribute segments of at least two first semantic objects, and updating the merged attribute segment based on the newly added second semantic object.

[0154] In a specific embodiment, the update module 114 may also be used to:

[0155] Based on the number of newly added segments in the group where the first semantic object is located, the existence confidence of the first semantic object is reduced to update the attribute segment of the first semantic object.

[0156] In another optional embodiment, the updating module 114 may be specifically configured to: compare a preset existence confidence threshold with the existence confidence in the updated semantic object attribute segment, and delete semantic objects corresponding to existence confidences lower than the existence confidence threshold.

[0157] In another alternative embodiment, referring to Fig.11 As shown, the device may further include: a storage module, the storage module 115 is used to combine the unsuccessfully clustered second semantic objects into a new segment data for storage. Of course, the updated map data is also stored.

[0158] In another alternative embodiment, referring to Fig.11 As shown, the device may further include: an acquisition module 116, a calibration module 117, and a segmentation module 118, and its working principle is as follows:

[0159] The acquisition module 116 is used to acquire new data, wherein the new data includes at least one newly added single trip data;

[0160] The calibration module 117 is used to calibrate the position relationship of the semantic objects in the newly added single trip data based on the association relationship between the lanes and the semantic objects in the newly added single trip data;

[0161] The segmentation module 118 is used to segment the calibrated newly added single trip data based on the position of the preset mark in the newly added single trip data to obtain a plurality of segment data after the newly added single trip data is segmented; wherein each segment data includes boundary information, vehicle lanes and semantic objects;

[0162] Correspondingly, the determination module 111 determines the correspondence between each segment data in the newly added single-pass data and the group data composed of the segment data at the same position after the multi-pass data in the historical data are segmented, and the clustering module 112 clusters the second semantic object in each segment data in the newly added single-pass data with the first semantic object in the group data in the historical data corresponding to the segment data.

[0163] Based on the same inventive concept, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned map data differential update method is implemented.

[0164] Based on the same inventive concept, a computer device is provided in an embodiment of the present invention, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the program, the differential update method of map data as described above is implemented.

[0165] The principles of the problems solved by the above-mentioned devices, media, and related equipment in the embodiments of the present invention are similar to those of the above-mentioned methods, so their implementation can refer to the implementation of the above-mentioned methods, and the repeated parts will not be repeated.

[0166] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0167] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0168] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0170] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A differential updating method for map data, characterized in that: include: Determining a historical data set of the first semantic object based on a count attribute in an attribute segment of the first semantic object in the historical data and a recent data list; Clustering the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object; The clusters formed by clustering are aggregated, and the map data is differentially updated with the aggregated semantic objects.

2. The method according to claim 1, characterized in that The determining of the historical data set of the first semantic object based on the count attribute in the attribute segment of the first semantic object in the historical data and the recent data list includes: Based on the recent historical data in the recent data list, the recent historical data is divided into several files according to the chronological order to obtain the recent historical data of several files; Based on the counting attribute and a counting ratio threshold preset for each bin, the recent historical data in each bin is copied respectively to determine a historical data set of the first semantic object.

3. The method according to claim 2, characterized in that The data is divided into several files in chronological order to obtain recent historical data, including: Sorting the recent historical data from new to old in time, and dividing the recent historical data into several files in time from new to old, so as to obtain several files of recent historical data; The preset counting ratio threshold of the bins composed of recent historical data with new time is greater than the preset counting ratio threshold of the bins composed of recent historical data with old time.

4. The method according to claim 1, characterized in that The clusters formed by clustering are aggregated, and the map data is differentially updated with the aggregated semantic objects, including: Determine whether the cluster formed by clustering contains the historical data; If the cluster only includes the first semantic object, updating the attribute segment of the first semantic object; If the cluster only includes the second semantic object, aggregating a new cluster formed by aggregating the second semantic objects, so as to update the map data with the aggregated new semantic objects; If the cluster includes the first semantic object and the second semantic object, a mixed cluster formed by the aggregation of the first semantic object and the second semantic object is aggregated to perform a differential update between the aggregated new semantic object and the first semantic object.

5. The method according to claim 4, characterized in that The step of aggregating the mixed cluster formed by aggregating the first semantic object and the second semantic object to differentially update the aggregated new semantic object and the first semantic object includes: Determining the number of the first semantic objects included in the mixed cluster; If the first semantic object is one, the hybrid cluster is used as a first hybrid cluster, and the first hybrid cluster inherits the ID of the first semantic object, so as to form a new semantic object by aggregating the first hybrid cluster, and differentially update the first semantic object in the map data based on the ID; If there are at least two first semantic objects, the mixed cluster is used as a second mixed cluster, and the second mixed cluster inherits the IDs of at least two of the first semantic objects to form a new semantic object by aggregating the second mixed clusters, and differentially update at least two first semantic objects in the map data based on the IDs.

6. The method according to claim 5, characterized in that After the first semantic object in the map data is differentially updated based on the ID, the method further includes: Selecting an object from the second semantic object that has the highest similarity to the new semantic object formed by the aggregation of the first mixed cluster, and adding it to the recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the new semantic object includes: updating the count attribute, updating the type confidence, updating the existence confidence, and updating the recent data list; After differentially updating at least two first semantic objects in the map data based on the ID, the method further includes: selecting an object from the second semantic objects that has the highest similarity to a new semantic object formed by aggregation of the second mixed cluster, and adding the object to a recent data list of the new semantic object to update the attribute segment of the semantic object; wherein updating the attribute segment of the semantic object includes: merging the attribute segments of at least two first semantic objects, and updating the merged attribute segment based on the newly added second semantic object.

7. The method according to claim 4, characterized in that The updating of the attribute segment of the first semantic object comprises: Based on the number of newly added segments in the group where the first semantic object is located, the existence confidence of the first semantic object is reduced to update the attribute segment of the first semantic object.

8. The method according to any one of claims 1 to 7, characterized in that: Also includes: According to a preset existence confidence threshold, the existence confidence in the updated semantic object attribute segment is compared, and the semantic objects corresponding to the existence confidences lower than the existence confidence threshold are deleted.

9. The method according to any one of claims 1 to 7, characterized in that: Also includes: The second semantic objects that are not successfully clustered are combined into a new segment data for storage.

10. The method according to any one of claims 1 to 7, characterized in that: Before clustering the first semantic object in the historical data and the second semantic object in the new data, the method further includes: Acquire new data, wherein the new data includes at least one newly added single trip data; Based on the association relationship between the lane and the semantic object in the newly added single trip data, calibrating the position relationship of the semantic object in the newly added single trip data; Segmenting the calibrated newly added single trip data based on the position of the preset mark in the newly added single trip data to obtain a plurality of segment data after the newly added single trip data is segmented; wherein each segment data includes boundary information, vehicle lanes and semantic objects; Determine the correspondence between each segment data in the newly added single-pass data and the group data composed of the segment data at the same position after the multi-pass data in the historical data are segmented, so as to cluster the second semantic object in each segment data in the newly added single-pass data with the first semantic object in the group data in the historical data corresponding to the segment data.

11. A map data differential updating device, characterized in that: include: a determination module, configured to determine a historical data set of the first semantic object based on a count attribute in an attribute segment of the first semantic object in the historical data and a recent data list; A clustering module, configured to cluster the historical data set of the first semantic object and the second semantic object in the new data to obtain a plurality of clusters including clusters of the first semantic object and / or the second semantic object; The updating module is used to aggregate the clusters formed by clustering and perform differential update on the map data with the aggregated semantic objects.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the map data differential updating method according to any one of claims 1 to 10 is implemented.

13. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the map data differential updating method according to any one of claims 1 to 10 is implemented.

Images