A high-precision map updating method, device and server

Abstract

The application discloses a high-precision map updating method, device and server. The method comprises the following steps: matching map elements in first high-precision map data collected with map elements in second high-precision map data corresponding to the first high-precision map data, positioning the map elements in the first high-precision map data according to a matching result, and fusing the map elements in the first high-precision map data after positioning with the map elements in the second high-precision data to obtain third high-precision map data after fusion. The application solves the problem of how to fuse and update high-precision maps, and meets the real-time requirement of high-precision maps on the basis of ensuring the accuracy of high-precision map data.

Description

Technical Field

[0001] This invention relates to the field of geographic information technology, and in particular to a method, apparatus and server for updating high-precision maps. Background Technology

[0002] High-precision maps refer to maps with high precision and detailed definition. Unlike the electronic maps that are currently widely known, their precision often needs to reach the decimeter level, and they can accurately distinguish each lane.

[0003] Unlike traditional electronic maps, high-precision maps primarily serve autonomous driving and advanced driver assistance systems (ADAS), such as driverless cars. Unlike human drivers, machine drivers lack innate visual recognition and logical analysis abilities. For example, humans can easily and accurately use images and GPS to locate themselves and identify obstacles, people, and traffic lights, but these are extremely difficult tasks for current robots.

[0004] Therefore, high-precision maps are an indispensable component of current autonomous driving and driverless vehicle technologies. High-precision maps contain a wealth of driving assistance information, the most important of which is a precise three-dimensional representation of the road network (centimeter-level accuracy). Furthermore, high-precision maps need to have higher real-time performance than traditional maps, ensuring the "newness" and "freshness" of map elements. Because road networks change daily, such as undergoing repairs, worn and repainted road markings, and alterations to traffic signs, these changes need to be reflected in the high-precision maps in a timely manner to ensure the safe operation of driverless vehicles.

[0005] The production and updating of large-scale, high-precision maps has become a major challenge for the mapping industry. Currently, one method for generating high-precision maps relies on crowdsourced vehicles to collect a large number of photos of urban roads. Then, data on various map elements captured in the photos are collected, such as lane lines, ground arrows, poles along the roads, and road signs, to generate high-precision map data.

[0006] Due to the stringent real-time requirements of high-precision maps, they are often updated every few hours, and may be updated several times a day. Ensuring the accuracy of map elements in the generated high-precision maps has become a hot issue. Summary of the Invention

[0007] In view of the above problems, the present invention is proposed to provide a method, apparatus and server for updating high-precision maps that overcomes or at least partially solves the above problems.

[0008] In a first aspect, embodiments of the present invention provide a method for high-precision map updating, comprising:

[0009] For the first high-precision map data collected, the map elements in it are matched with the map elements in the existing second high-precision map data corresponding to the first high-precision map data, and the map elements in the first high-precision map data are located according to the matching results.

[0010] The map features of the first high-precision map data after positioning are merged with the map features of the second high-precision data to obtain the merged third high-precision map data.

[0011] In one embodiment, map features in a first high-precision map are matched with map features in a second high-precision map data, and the map features in the first high-precision map data are located based on the matching results. Specifically, this includes:

[0012] Map elements belonging to the first category in the first high-precision map data are matched with map elements of the corresponding category in the second high-precision map data, and the positions of the map elements in the first high-precision map data are adjusted according to the matching results.

[0013] For the adjusted first high-precision map data, the map features belonging to the second category are matched with the map features of the corresponding category in the second high-precision map data, and the positions of the map features in the adjusted first high-precision map data are adjusted according to the matching results.

[0014] The shape features of map elements in the first category are more distinctive than those in the second category.

[0015] In one embodiment, the first category of map elements includes one or more of the following: ground arrows, poles on both sides of the road, and road signs; the second category of map elements is lane lines.

[0016] In one embodiment, matching map features belonging to a first category in the first high-precision map data with map features of the corresponding category in the second high-precision map data, and adjusting the positions of the map features in the first high-precision map data based on the matching results, includes:

[0017] Within a preset first precision range, the first high-precision map data is sequentially shifted in position according to a preset movement step size, and the total distance between the first or second category map elements in the first high-precision map after each overall shift and the corresponding map elements in the second high-precision map data is calculated.

[0018] The position with the minimum total distance is determined as the position after adjustment of the first high-precision map data.

[0019] In one embodiment, the step of matching map features belonging to the second category in the adjusted first high-precision map data with map features of the corresponding category in the second high-precision map data, and then adjusting the positions of the map features in the adjusted first high-precision map data again based on the matching results, includes:

[0020] Within the preset second precision range, according to the preset movement step size, the adjusted first high precision map data is sequentially translated as a whole, and the total distance between the second category map elements in the adjusted first high precision map after each overall translation and the corresponding map elements in the second high precision map data is calculated.

[0021] The position with the minimum total distance is determined as the position after further adjustment of the adjusted first high-precision map data.

[0022] In one embodiment, fusing map features from the first high-precision map data after positioning with map features from the second high-precision map data includes:

[0023] The position data of shape points on map features in the first high-precision map data after positioning are weighted and calculated with the position data of shape points on corresponding map features in the second high-precision map data to determine the position data of each shape point of each map feature in the third high-precision map data.

[0024] In one embodiment, the step of fusing map features of the first high-precision map data after positioning with map features of the second high-precision data further includes: recording the fusion time and / or fusion number of map features in the third high-precision map data; the fusion number is the sum of the fusion numbers of the second high-precision map data and the first high-precision map data;

[0025] After obtaining the fused third high-precision map data, the process also includes:

[0026] Based on the fusion time and / or fusion number of map elements in the third high-precision map data, extract and output the map elements that reach the preset fusion time threshold and / or fusion number threshold.

[0027] In one embodiment, after recording the fusion time and / or number of fusions for each map element in the third high-precision map data, the method further includes:

[0028] Based on the fusion time of map features in the recorded third high-precision map data, find map features that have not been fused for a preset time.

[0029] Subtract a preset value from the number of times the found map elements have been merged;

[0030] Delete map features in the current third-highest precision map data whose fusion count is lower than a preset threshold.

[0031] Secondly, embodiments of the present invention provide an apparatus for high-precision map updating, comprising:

[0032] The positioning module is used to match the map features in the collected first high-precision map data with the map features in the existing second high-precision map data corresponding to the first high-precision map data, and to locate the first high-precision map data according to the matching result.

[0033] The fusion module is used to fuse the map features of the first high-precision map data after positioning with the map features of the second high-precision data to obtain the fused third high-precision map data.

[0034] Thirdly, embodiments of the present invention provide a high-precision map, wherein the data of the high-precision map is obtained through the aforementioned high-precision map update method.

[0035] Fourthly, embodiments of the present invention provide a high-precision map server, including: a memory, a processor, and computer instructions stored in the memory and executable on the processor, wherein the instructions, when executed by the processor, can realize the aforementioned method for updating the high-precision map.

[0036] Fifthly, embodiments of the present invention provide a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the aforementioned method for updating a high-precision map.

[0037] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:

[0038] The technical solution of the high-precision map update method, apparatus and server provided in the embodiments of the present invention involves matching and locating map elements in the currently collected first high-precision map data with map elements in the existing second high-precision map corresponding to the first high-precision map data to improve the accuracy of the data. Then, the map elements of the located first high-precision map are merged with the map elements of the second high-precision map data to update the merged map and obtain the latest high-precision map (i.e., the third high-precision map data). The present invention solves the problem of how to merge and update high-precision maps, and meets the real-time requirements of high-precision maps while ensuring the accuracy of high-precision map data.

[0039] Furthermore, in the above-mentioned technical solution provided by the embodiments of the present invention, when matching and locating the map elements of the currently collected first high-precision map with the map elements of the existing second high-precision map, a two-stage matching scheme is adopted. That is, the first matching is based on the first category of map elements to perform a coarse matching of the overall map elements, and the second matching is based on the result of the first coarse matching to perform a fine matching of the overall elements based on the second category of map elements. The position detection accuracy of the first category of map elements is lower than that of the second category of map elements. Therefore, the two-stage matching scheme can effectively eliminate the deviations caused by the inertial navigation of the map data collection vehicle (such as a crowdsourced vehicle) and the vehicle sensors when perceiving map elements, and ensure the accuracy of the data.

[0040] Furthermore, in the above-mentioned technical solution provided by the embodiments of the present invention, after fusing the map elements of the first high-precision map data after positioning with the map elements of the second high-precision map, a step of recording the fusion time and / or the number of fusions of the map elements is also performed. In this way, the real-time degree of the map elements of the fused high-precision map, that is, the degree of "newness" and "freshness", can be determined based on the recorded fusion time and / or the number of fusions, and map elements that do not conform to the real environment due to false detection, element changes, etc. can be removed.

[0041] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0042] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0043] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0044] Figure 1 This is a schematic diagram of map elements in a high-precision map according to an embodiment of the present invention;

[0045] Figure 2 This is a flowchart of a high-precision map update method in an embodiment of the present invention;

[0046] Figure 3 This is a flowchart illustrating the process of matching and locating map elements in the first high-precision map data with existing map elements in the second high-precision map data in an embodiment of the present invention.

[0047] Figure 4This is a schematic diagram illustrating the overall translation range of each map element in the coarse matching process of this invention.

[0048] Figure 5A and 5B This is a schematic diagram comparing coarse matching and fine matching before and after in an embodiment of the present invention;

[0049] Figure 6 This is a schematic diagram of map element fusion in an embodiment of the present invention;

[0050] Figure 7 This is the implementation process of the fading memory mechanism for fusion time in this embodiment of the invention;

[0051] Figure 8 This is a structural block diagram of the high-precision map updating device in an embodiment of the present invention. Detailed Implementation

[0052] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0053] Before describing the technical solutions for the high-precision map updating method, apparatus, and server provided in the embodiments of the present invention, the process of collecting high-precision map data will be briefly explained first.

[0054] In the process of collecting high-precision map data, autonomous vehicles equipped with various sensors (inertial navigation, cameras, LiDAR, millimeter-wave radar, etc.) are mainly used to perform autonomous driving tasks in urban areas. While assisting the vehicles in completing autonomous driving, these sensors can also collect the map element data needed to create high-precision maps. Inertial navigation can obtain the absolute position information of the vehicle, while cameras (or LiDAR, millimeter-wave radar, etc.) can measure the relative positional relationship between the surrounding map elements and the vehicle. Combining these two methods yields the map element data required to create high-precision maps.

[0055] Reference Figure 1 As shown, high-precision maps typically include the following map elements:

[0056] Lane lines: Road markings that separate lanes, which can be represented by a series of consecutive points.

[0057] Ground arrows: Ground markings used to indicate vehicle lanes. They are represented by the four corner points of an enclosing rectangle, and a type value is used to distinguish different types such as straight ahead, left turn, and right turn. Figure 1 The number within the rectangle is the type value.

[0058] Pole: A pole-like object on either side of a road. It can be any man-made pole-like object on either side of the road, represented by a point projected onto the ground.

[0059] Sign: Road sign. Represented by the four corner points of an outer rectangle. Figure 1 The sign is represented by a top-down view, using a line projected onto the ground.

[0060] Of course, in this embodiment of the invention, the map elements include, but are not limited to, the above four types, and can be expanded to include more map elements such as road edges and zebra crossings.

[0061] To address the problem of fusion and updating existing high-precision map data, this invention provides a method for high-precision map updating, referring to... Figure 2 As shown, it includes:

[0062] S21. For the collected first high-precision map data, match the map elements in it with the map elements in the existing second high-precision map data corresponding to the first high-precision map data, and locate the first high-precision map data according to the matching result.

[0063] S22. The map elements of the first high-precision map data after positioning are fused with the map elements of the second high-precision data to obtain the fused third high-precision map data.

[0064] It should be noted that the first high-precision map is the high-precision map data identified from the photos collected by crowdsourcing, and it is also the high-precision map data to be merged. The second high-precision map is the existing merged high-precision map data. The third high-precision map data refers to the high-precision map data after the first and second high-precision map data are merged. The terms "first", "second" and "third" are only used to distinguish different high-precision maps, and do not indicate the order or priority of the maps.

[0065] In the specific implementation of this invention, the above steps S21 to S22 will be repeatedly executed during the production of high-precision maps. Each time new high-precision map data is generated, it needs to be merged with existing high-precision map data to generate new merged high-precision map data as the basis for the next fusion update.

[0066] The first high-precision map data has a corresponding relationship with the existing second high-precision map data. This correspondence can be a correspondence in the geographical scope of the map, with both having the same scope, such as both representing the same road or the same block area.

[0067] In one possible implementation, during the crowdsourced data collection process, the vehicle's inertial navigation system provides its absolute position, while cameras or lidar measure the relative positions of map features around the vehicle. Combining these two methods yields the map feature data required for a high-precision map. Since the error of the vehicle's inertial navigation system can be considered constant over a relatively short period or distance, it changes with time and distance. If all the crowdsourced data is used as a whole to create and update the high-precision map, the change in inertial navigation error will inevitably affect the accuracy of the high-precision map data. Therefore, the high-precision map data can be segmented and processed using the road between two intersections as the basic unit. In this way, the ranges of the first, second, and third high-precision map data can represent the geographical area corresponding to this basic unit.

[0068] Based on the aforementioned high-precision map production process, the crowdsourced acquisition of map elements involves: inertial navigation providing the vehicle's absolute position, and sensing sensors (cameras, LiDAR, millimeter-wave radar, etc.) providing the relative positions of map elements from the vehicle within the high-precision map data. This acquisition method results in two types of data bias: one is the bias in the vehicle's inertial navigation providing the absolute position, and the other is the bias in the relative positions of map elements detected by the sensors on the vehicle.

[0069] The absolute position information provided by the inertial navigation system (INS) is subject to systematic bias due to the working principles of GPS and other positioning systems, resulting in overall deviations in the positions of the collected map elements. For example, when traveling the same road at different times, the position data provided by the INS may differ, exhibiting a deviation at the meter level. Therefore, in step S21 of this embodiment, the present solution eliminates the overall deviation of road data through map element matching. Matching positioning shifts map elements in the newly collected high-precision map data to the most suitable position in the final high-precision map to be merged and updated, thereby achieving the purpose of eliminating overall deviations.

[0070] Specifically, in the actual implementation of step S21 above, refer to Figure 3 As shown, it may include the following steps:

[0071] S31. Match the map elements belonging to the first category in the first high-precision map data with the map elements of the corresponding category in the second high-precision map data, and adjust the position of the map elements in the first high-precision map data according to the matching result.

[0072] S32. For the adjusted first high-precision map data, match the map elements belonging to the second category with the map elements of the corresponding category in the second high-precision map data, and adjust the positions of the map elements in the adjusted first high-precision map data again based on the matching results.

[0073] In the above process, the shape features of the first category of map elements have a higher distinguishability than those of the second category of map elements.

[0074] The selection of map features into the first and second categories is primarily based on the distinguishability of their shape features, in other words, their degree of "unlikeliness" in being confused. Therefore, map features in the first category are suitable for coarse matching, which first eliminates the overall system's positional error, and then fine-tunes the positional error based on map features in the second category that are closer to the vehicle and have smaller relative positional errors.

[0075] For example, with Figure 1 Taking the map elements shown as an example, the first category of map elements could be ground arrows, poles along roads, and road signs, while the second category could be lane lines. The first category of map elements is farther from vehicles, but their shapes vary more and their shape features are more distinct than those of the second category, lane lines. Therefore, when a computer program identifies and matches these elements, it's unlikely to identify two similar map elements of the same category as the same element. For example, within a very small area, it's unlikely there are two streetlights with exactly the same shape. Conversely, if lane lines are used for coarse matching, because the shape features of lane lines are not very distinct when collecting them, the differentiation is low, and lane lines are not unique. It's impossible to accurately find the location of the currently collected lane line and determine which lane line in the existing map data has a precise correspondence. The sensor may easily confuse the current lane with adjacent lanes, leading to lane line misidentification. Therefore, it's necessary to first determine the absolute position of the current lane line by comparing it with elements in the road that have relatively distinct and easily identifiable shape features, i.e., the first category of map elements, such as utility poles, signs, and ground arrows.

[0076] The coarse matching in S31 above can be achieved in the following way:

[0077] Within a preset first precision range, the position of the first high-precision map data is shifted as a whole according to a preset movement step size, and the total distance between the first category map elements in the first high-precision map and the corresponding map elements in the second high-precision map data is calculated after each overall shift; the position with the smallest total distance calculated each time is determined as the adjusted position of the first high-precision map data.

[0078] Examples are provided in conjunction with the diagrams; please refer to them. Figure 4As shown, taking the center point of the geographical range of the first high-precision map data currently collected as the origin (coordinate (0.0)), the positions of all map elements in the first high-precision map data are moved sequentially in a step of 0.5 meters in all directions (within ±10m in the x-axis and y-axis directions).

[0079] When calculating distances, for example, we can first perform feature matching on the maps, that is, find corresponding map features in the first and second high-precision maps respectively, forming matching pairs; the matching criterion is to find similar map features within a 2-meter radius of the feature in the first high-precision map. Figure 5A As shown, for ease of explanation, the map elements in the first high-precision map data are called new map elements, and the corresponding map elements in the existing high-precision map data are called original map elements. Then, the following pairs of map elements are formed: new pole 1 and original pole 1, new pole 2 and original pole 2, new ground arrow 3 and original ground arrow 3, new ground arrow 4 and original ground arrow 4, and new sign 1 and original sign 2.

[0080] For example, during calculations, since the error provided by the inertial navigation system is within ±10m, the first high-precision map data elements are translated as a whole, with a step size of 0.5 meters, within the first precision range of ±10m. Starting from the center of the first high-precision map data, the entire first high-precision map data is translated 0.5 meters in the positive horizontal direction. Then, the distances between the map elements of the first high-precision map data and the corresponding elements of the second high-precision map data are calculated and summed. For example... Figure 5A As shown, calculate the distance between the new pole 1 and the original pole 1, the distance between the new pole 2 and the original pole 2, the distance between the new ground arrow 3 and the original ground arrow 3, the distance between the new ground arrow 4 (which is obscured by the original ground arrow 4 and not shown) and the original ground arrow 4, and the distance between the new sign 5 and the original sign 5, and then add them together to get the total distance S1.

[0081] Then, the new map elements are shifted 0.5 meters in the positive direction of the horizontal axis according to the step size. Using a similar method, the distances between the new map elements and the original merged map elements are obtained in 5 sets, and then added together to obtain S2.

[0082] This process is repeated 40*40 times in both the positive and negative directions of the x and y axes, resulting in a total of 1600 distances from S1 to S1600.

[0083] When the sum of the distances between the first high-precision map data and the second high-precision map data is minimized, that is, when the first high-precision map and the second high-precision map are closest, it is the position where the coarse matching ends.

[0084] After matching the first category of map features, the positions of each map feature in the first high-precision map are roughly close to the positions of the corresponding map features in the second high-precision map. However, because the map features of this category are far from the vehicle, their detection accuracy is not as good as that of lane lines. However, in order to overcome the error of the vehicle sensor, further fine matching is needed to eliminate the error. Specifically, the fine matching process in step S32 above can be carried out in the following way:

[0085] The map elements belonging to the second category in the adjusted first high-precision map data are matched with the map elements of the corresponding category in the second high-precision map data, and the positions of the map elements in the adjusted first high-precision map data are adjusted again based on the matching results.

[0086] Since the basic location of the first high-precision map has been determined through the matching steps S31 and S32, it is now only necessary to perform another round of matching on its location based on the lane lines.

[0087] Similar to coarse matching, in fine matching, the lane lines in the first high-precision map feature are used as a reference to find lane lines in the second high-precision map within a 0.5-meter range as matching pairs. For example... Figure 5B As shown, two matching pairs are formed: new lane line 1 and original lane line 1, and new lane line 2 and original lane line 2.

[0088] Within a preset second precision range, such as 1 meter, and a preset movement step size of 0.1 meters, the position of the first high-precision map data is translated as a whole. The distance between the new lane line 1 and the original lane line 1, as well as the distance between the new lane line 2 and the original lane line 2, is calculated. This process is repeated 20*20=400 times to obtain 400 distance sums.

[0089] When calculating the distance between lane lines, for example, sampling points can be selected on the new lane line, the distance from each sampling point to the original lane line can be calculated, and then the average of the distances from each sampling point to the original lane line can be calculated to determine the distance between the new lane line and the original lane line. Of course, the embodiments of the present invention are not limited to using other calculation methods.

[0090] The minimum value among the 400 distances and results is selected as the final best matching position for the first high-precision map.

[0091] After the above steps, the first high-precision map has been shifted to the optimal position of the second high-precision map, thereby eliminating the overall deviation caused by the data collected by the inertial navigation equipment and the deviation during sensor detection.

[0092] At this point, the map elements in the first high-precision map and the second high-precision map are very close to each other. Following the method described in step S22 above, the map elements in the first high-precision map data after matching and positioning are fused with the map elements in the second high-precision map data in terms of shape. In specific implementation, this can be achieved in the following way: the position data of the shape points on each map element in the first high-precision map data after positioning are weighted and calculated with the position data of the shape points on the corresponding map elements in the second high-precision map data to obtain the position data of each shape point of each map element in the third high-precision map data.

[0093] Reference Figure 6 As shown, the leftmost box represents a map feature in the second high-precision map data (original fused data), the rightmost box represents the corresponding map feature in the first high-precision map data (new data), and the box between them represents the corresponding map feature in the third high-precision map data after shape fusion (new fused data). Figure 6 As can be seen, the location of the merged map features lies between the two, and is closer in position to the map features in the existing second high-precision map data. In other words, during shape fusion, the weight values ​​of the positional data of each shape point of the map features in the second high-precision map data are greater than the weight values ​​of the positional data of the corresponding shape points of the map features in the first high-precision map data. For example... Figure 6 As shown, point A in the original fused data, point B in the new data, and point C in the newly fused data are three corresponding shape points. The position data of point A and point B are weighted and calculated to obtain the position data of point C. The calculation of other shape points is similar and will not be described in detail here.

[0094] The weight values ​​for the aforementioned weighted calculation can be calculated using the number of times the first high-precision map data and the number of times the second high-precision map data are fused. Based on this, in step S22 above, the following operations can also be performed: record the fusion time and / or fusion count of each map element in the third high-precision map data (indicating the number of times the map element currently participates in fusion); the fusion count of each map element in the third high-precision map data is the sum of the fusion counts of the second high-precision map data and the first high-precision map data. Accordingly, after obtaining the fused third high-precision map data, based on the fusion time and / or fusion count of each map element in the third high-precision map data, extract and output the map elements that reach the preset fusion time threshold and / or fusion count threshold to obtain the final high-precision map data.

[0095] The number of times the first high-precision map data is fused can be set to 1. The number of times the second high-precision map data is fused is determined based on the recorded actual number of fusions. In other words, the more times the second high-precision map data is fused, the closer the shape and position of the newly fused data will be to the second high-precision map data (because the more times it is fused, the more accurate the existing map data is considered to be).

[0096] In one embodiment, as described above, for each map feature in the fused third high-precision map data, a corresponding fusion time and / or fusion count is recorded. Another advantage of this design is that when one or more map features in the high-precision map data experience false detection or feature updates—for example, due to obstruction by other vehicles causing some features to be missed in the image, or due to various objective factors such as road construction causing traffic signs to be removed—these possible errors can be eliminated through the fusion time and / or fusion count.

[0097] Specifically, the fusion time can be represented by a timestamp, which identifies the last time the element participated in the fusion process. Let's take... Figure 6 Taking the example shown, the attributes of map elements in the second high-precision map include the corresponding fusion count and fusion time: fusion count Times = 2, Time Stamp = 1497250729.98; the attributes of map elements in the first high-precision map include: fusion count Times = 1, Time Stamp = 1496815040.17. After the fusion update, the fusion count of the third high-precision map element C is the sum of the fusion counts of the first and second high-precision map elements. The attributes of the corresponding map element in the third high-precision map include: fusion count Times = 3, Time Stamp = 1497250729.98.

[0098] After the above processing, if a map element not present in the existing second high-precision map data is added to the first high-precision map data, the map element will be added to the merged high-precision map data. If a map element in the second high-precision map data is removed or its position is changed, the map element will still be retained in the merged high-precision map data after the above matching process. In order to ensure the smooth metabolism of map elements, in this embodiment of the invention, a fading memory mechanism of fusion time can also be used to update the merged map elements, thereby ensuring the accuracy of the map elements themselves.

[0099] For specific methods of integrating the fading memory mechanism over time, refer to... Figure 7 The process shown includes the following steps:

[0100] S71. Based on the fusion time of map elements in the recorded third high-precision map data, find map elements that have not been fused for a preset time.

[0101] S72. Subtract a preset value from the number of times the found map elements have been merged;

[0102] S73. Delete map elements in the current third high-precision map data whose fusion count is lower than the preset threshold.

[0103] For example.

[0104] The fusion time of each element in the updated third high-precision map data is used to determine which map elements did not participate in the fusion calculation (e.g., did not participate in the fusion calculation for 1 day). If so, the fusion count of that map element is penalized (e.g., the fusion count is reduced by 1), and map elements in the current third high-precision map whose fusion count is already 0 are deleted.

[0105] When a map feature's fusion count is found to be 0, that feature is removed from the merged map. For example, a roadside sign exists in previously collected map data, and its fusion count is 30. However, if this sign is not found in newly collected map data, its fusion count decreases by 1 to 29. If the sign was not collected due to a false positive, it remains in the merged map. If it is subsequently collected, its fusion count increases by 1, becoming 30 again. This process ensures the sign remains in the merged map, reflecting the actual situation. However, if the sign is indeed removed, crowdsourced vehicles will continue to fail to collect data related to it in subsequent data collections. This causes its fusion count to continuously decrease with each high-precision map feature fusion until it reaches 0, at which point it is removed from the merged map. Only after continuous confirmation of its non-existence is it truly deleted from the map data, ensuring the accuracy of map feature updates.

[0106] Finally, map elements that meet the requirements for fusion time and / or fusion number are extracted from the third high-precision map, thereby satisfying the requirements of "newness" and "accuracy" of map elements in high-precision map data.

[0107] Based on the same inventive concept, embodiments of the present invention also provide a device for high-precision map updating, a high-precision map, and a high-precision map server. Since the principles by which these devices and servers solve the problem are similar to the aforementioned high-precision map updating method, the implementation of these devices and servers can refer to the implementation of the aforementioned method, and repeated details will not be repeated.

[0108] This invention provides a device for high-precision map updating, referring to... Figure 8 As shown, it includes:

[0109] The positioning module 81 is used to match the map features in the collected first high-precision map data with the map features in the existing second high-precision map data corresponding to the first high-precision map data, and to locate the first high-precision map data according to the matching result.

[0110] The fusion module 82 is used to fuse the map features of the first high-precision map data after positioning with the map features of the second high-precision data to obtain the fused third high-precision map data.

[0111] In one embodiment, the positioning module 81 is specifically used to match map elements belonging to a first category in the first high-precision map data with map elements of the corresponding category in the second high-precision map data, and adjust the positions of all map elements in the first high-precision map data according to the matching results; for the adjusted first high-precision map data, match map elements belonging to a second category with map elements of the corresponding category in the second high-precision map data, and adjust the positions of map elements in the adjusted first high-precision map data again according to the matching results; the shape feature distinguishability of the map elements of the first category is higher than that of the map elements of the second category.

[0112] In one embodiment, the first category of map elements includes one or more of the following: ground arrows, poles on both sides of the road, and road signs; the second category of map elements is lane lines.

[0113] In one embodiment, the positioning module 81 is specifically used to perform overall translation of the position of the first high-precision map data sequentially within a preset first precision range and according to a preset movement step size, and calculate the total distance between the map elements of the first category or the second category in the first high-precision map after each overall translation and the corresponding map elements in the second high-precision map data; determine the position with the smallest total distance as the adjusted position of the first high-precision map data.

[0114] In one embodiment, the positioning module 81 is specifically used to perform overall translation of the adjusted first high-precision map data within a preset second precision range and according to a preset movement step size, and to calculate the total distance between all map elements of the second category in the adjusted first high-precision map after each overall translation and the corresponding map elements in the second high-precision map data; and to determine the position with the smallest total distance as the position of the adjusted first high-precision map data after further adjustment.

[0115] In one embodiment, the fusion module 82 is specifically used to perform a weighted operation on the position data of shape points on each map element in the first high-precision map data after positioning, and the position data of shape points on the corresponding map element in the second high-precision map data, to obtain the position data of each shape point of the map element in the third high-precision map data.

[0116] In one embodiment, the aforementioned high-precision map updating apparatus refers to Figure 8 As shown, it also includes: extraction module 83;

[0117] The fusion module 82 is also used to record the fusion time and / or fusion number of each map element in the third high-precision map data; the fusion number is the sum of the fusion numbers of the second high-precision map data and the first high-precision map data;

[0118] Accordingly, the extraction module 83 is used to extract and output the map elements that reach the preset fusion time threshold and / or fusion number threshold based on the fusion time and / or fusion number of each map element in the third high-precision map data.

[0119] This invention also provides a high-precision map, the data of which is obtained through the aforementioned high-precision map update method.

[0120] This invention also provides a high-precision map server, comprising: a memory, a processor, and computer instructions stored in the memory and executable on the processor, characterized in that the instructions, when executed by the processor, can realize the aforementioned high-precision map update method.

[0121] This invention also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the aforementioned method for updating high-precision maps.

[0122] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0123] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0124] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0125] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0126] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for high-precision map updating, characterized in that, include: For the first high-precision map data collected, the map elements in it are matched with the map elements in the existing second high-precision map data corresponding to the first high-precision map data, and the map elements in the first high-precision map data are located according to the matching results. The map features of the first high-precision map data after positioning are merged with the map features of the second high-precision map data to obtain the merged third high-precision map data. Specifically, this involves matching map features in the first high-precision map with map features in the second high-precision map data, and locating the map features in the first high-precision map data based on the matching results. This includes: Map elements belonging to the first category in the first high-precision map data are matched with map elements of the corresponding category in the second high-precision map data, and the positions of the map elements in the first high-precision map data are adjusted according to the matching results. For the adjusted first high-precision map data, the map features belonging to the second category are matched with the map features of the corresponding category in the second high-precision map data, and the positions of the map features in the adjusted first high-precision map data are adjusted according to the matching results. The shape features of map elements in the first category are more distinctive than those in the second category.

2. The method as described in claim 1, characterized in that, The first category of map elements includes one or more of the following: ground arrows, poles on both sides of the road, and road signs; the second category of map elements is lane lines.

3. The method as described in claim 1 or 2, characterized in that, The step of matching map features belonging to the first category in the first high-precision map data with map features of the corresponding category in the second high-precision map data, and adjusting the positions of the map features in the first high-precision map data according to the matching results, includes: Within a preset first precision range, the first high-precision map data is sequentially shifted in position according to a preset movement step size, and the total distance between the first or second category map elements in the first high-precision map after each overall shift and the corresponding map elements in the second high-precision map data is calculated. The position with the minimum total distance is determined as the position after adjustment of the first high-precision map data.

4. The method as described in claim 1 or 2, characterized in that, The process of matching map features belonging to the second category in the adjusted first high-precision map data with map features of the corresponding category in the second high-precision map data, and then adjusting the positions of the map features in the adjusted first high-precision map data again based on the matching results, includes: Within the preset second precision range, according to the preset movement step size, the adjusted first high precision map data is sequentially translated as a whole, and the total distance between the second category map elements in the adjusted first high precision map after each overall translation and the corresponding map elements in the second high precision map data is calculated. The position with the minimum total distance is determined as the position after further adjustment of the adjusted first high-precision map data.

5. The method as described in claim 1 or 2, characterized in that, The map features of the first high-precision map data after positioning are merged with the map features of the second high-precision map data, including: The position data of shape points on map features in the first high-precision map data after positioning are weighted and calculated with the position data of shape points on corresponding map features in the second high-precision map data to determine the position data of each shape point of each map feature in the third high-precision map data.

6. The method as described in claim 1 or 2, characterized in that, The step of fusing map features of the first high-precision map data after positioning with map features of the second high-precision map data further includes: recording the fusion time and / or fusion number of map features in the third high-precision map data; the fusion number is the sum of the fusion numbers of the second high-precision map data and the first high-precision map data; After obtaining the fused third high-precision map data, the process also includes: Based on the fusion time and / or fusion number of map elements in the third high-precision map data, extract and output the map elements that reach the preset fusion time threshold and / or fusion number threshold.

7. The method as described in claim 6, characterized in that, After recording the fusion time and / or number of fusions for each map element in the third high-precision map data, the method further includes: Based on the fusion time of map features in the recorded third high-precision map data, find map features that have not been fused for a preset time. Subtract a preset value from the number of times the found map elements have been merged; Delete map features in the current third-highest precision map data whose fusion count is lower than a preset threshold.

8. A device for high-precision map updating, characterized in that, include: The positioning module is used to match the map features in the collected first high-precision map data with the map features in the existing second high-precision map data corresponding to the first high-precision map data, and to locate the first high-precision map data according to the matching result. The fusion module is used to fuse the map features of the first high-precision map data after positioning with the map features of the second high-precision map data to obtain the fused third high-precision map data. Specifically, this involves matching map features in the first high-precision map with map features in the second high-precision map data, and locating the map features in the first high-precision map data based on the matching results. This includes: Map elements belonging to the first category in the first high-precision map data are matched with map elements of the corresponding category in the second high-precision map data, and the positions of the map elements in the first high-precision map data are adjusted according to the matching results. For the adjusted first high-precision map data, the map features belonging to the second category are matched with the map features of the corresponding category in the second high-precision map data, and the positions of the map features in the adjusted first high-precision map data are adjusted according to the matching results. The shape features of map elements in the first category are more distinctive than those in the second category.

9. A high-precision map, characterized in that, The data for the high-precision map is obtained through the high-precision map update method as described in any one of claims 1-7.

10. A high-precision map server, characterized in that, include: A memory, a processor, and computer instructions stored in the memory and executable on the processor, characterized in that, when executed by the processor, the instructions are capable of implementing the high-precision map updating method as described in any one of claims 1-7.

11. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, this instruction implements the high-precision map update method as described in any one of claims 1-7.

Images