Map data cleaning and map updating method, device and equipment, and storage medium

Abstract

The embodiment of the present disclosure discloses a map data cleaning and map updating method, device, equipment and storage medium. The method comprises the following steps: fusing multiple initial vector data corresponding to a to-be-updated area to obtain update vector data corresponding to the to-be-updated area; taking base map data corresponding to the to-be-updated area as a reference to calibrate the update vector data to obtain calibrated vector data; and in response to the data quality of the calibrated vector data failing to meet normal data quality standards, the calibrated vector data is removed. The technical solution can maximize the retention of effective information and improve the effective utilization rate of data.

Description

Map data cleaning and map updating methods, apparatus, equipment, and storage media Technical Field

[0001] This disclosure relates to the field of high-precision map technology, specifically to a map data cleaning and map updating method, apparatus, device, and storage medium. Background Technology

[0002] With social development and the advancement of science and technology, safer and more efficient autonomous driving or advanced driver assistance technologies have been recognized as one of the future directions of the automotive travel industry, and many autonomous driving or advanced driver assistance behaviors rely on high-precision map data.

[0003] To improve the update frequency and reduce the cost of high-precision map data, existing technologies propose using data collection devices mounted on crowdsourced vehicles to compliantly collect data for map creation. However, these crowdsourced vehicles use low-cost data collection devices, resulting in lower data quality compared to data collected by professional equipment. Directly using data collected by these low-cost devices for map updates may lead to a decline in map quality. Therefore, data cleaning is necessary to remove erroneous or invalid data. However, existing data cleaning solutions mostly involve directly deleting substandard raw data, resulting in low data utilization efficiency. Summary of the Invention

[0004] To address the problems in the related technologies, embodiments of this disclosure provide a method, apparatus, device, and storage medium for map data cleaning and map updating.

[0005] Firstly, this disclosure provides a data cleaning method.

[0006] Specifically, the data cleaning method includes:

[0007] The initial vector data corresponding to the region to be updated is fused together to obtain the updated vector data corresponding to the region to be updated.

[0008] The updated vector data is calibrated based on the base map data corresponding to the area to be updated to obtain calibrated vector data.

[0009] If the data quality of the calibration vector data fails to meet the normal data quality standard, the calibration vector data is cleared.

[0010] Secondly, this disclosure provides a map updating method, including:

[0011] Obtain initial vector data corresponding to multiple regions to be updated, with each region to be updated corresponding to multiple passes of initial vector data;

[0012] Using the map data cleaning method described in the first aspect, the calibration vector data of the area to be updated that does not meet the normal data quality standard is removed, and the calibration vector data of the area to be updated that meets the normal data quality standard is obtained.

[0013] Map updates are performed using calibration vector data from the areas to be updated, where the data quality meets normal data quality standards.

[0014] Thirdly, this disclosure provides a map updating device, including:

[0015] The fusion module is configured to fuse multiple initial vector data corresponding to the region to be updated to obtain the updated vector data corresponding to the region to be updated.

[0016] The calibration module is configured to calibrate the update vector data based on the base map data corresponding to the area to be updated, so as to obtain calibration vector data.

[0017] The clearing module is configured to clear the calibration vector data in response to the data quality of the calibration vector data failing to meet the normal data quality standard.

[0018] Fourthly, this disclosure provides a map updating device, comprising:

[0019] The acquisition module is configured to acquire initial vector data corresponding to multiple regions to be updated, with each region to be updated corresponding to multiple passes of initial vector data.

[0020] The cleaning module is configured to use the method described in any one of claims 1 to 9 to remove calibration vector data of the area to be updated that does not meet the normal data quality standard, and to obtain calibration vector data of the area to be updated that meets the normal data quality standard.

[0021] The update module is configured to update the map using calibration vector data of the area to be updated, which meets the normal data quality standard.

[0022] Fifthly, embodiments of this disclosure provide an electronic device including a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method as described in any one of the first or second aspects.

[0023] In a sixth aspect, embodiments of this disclosure provide a computer-readable storage medium having computer instructions stored thereon that, when executed by a processor, implement the method as described in any one of the first or second aspects.

[0024] In a seventh aspect, this disclosure provides a computer program product including computer instructions that, when executed by a processor, implement the method steps as described in any one of the first or second aspects.

[0025] According to the technical solution provided in this disclosure, multiple initial vector data corresponding to the area to be updated can be fused to remove redundant data, making the data more complete and reliable, and obtaining updated vector data corresponding to the area to be updated. Then, the updated vector data is calibrated based on the base map data to obtain calibrated vector data. In this way, the calibrated vector data is improved to the geometric accuracy level of the base map data. It is then determined whether the data quality of the calibrated vector data is normal data quality. If the data quality of the calibrated vector data does not meet the normal data quality standard, the calibrated vector data is cleared to avoid a decrease in map quality after the update. Thus, calibrating the initial vector data after fusion can improve the quality of low-cost data, prevent the clearing of effective data, and improve the effective utilization rate of data.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0027] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments, taken in conjunction with the accompanying drawings. In the drawings:

[0028] Figure 1A shows a flowchart of a map data cleaning method according to an embodiment of the present disclosure;

[0029] Figure 1B shows a schematic diagram of the fusion process according to an embodiment of the present disclosure;

[0030] Figure 1C shows an example diagram of initial vector data in the region to be updated according to an embodiment of the present disclosure;

[0031] Figure 2A shows a flowchart of a map updating method according to an embodiment of the present disclosure;

[0032] Figure 2B shows a flowchart of a map updating method according to an embodiment of the present disclosure;

[0033] Figure 3 illustrates an application scenario of the map data cleaning and map updating method according to an embodiment of the present disclosure.

[0034] Figure 4A shows a structural block diagram of a map data cleaning apparatus according to an embodiment of the present disclosure;

[0035] Figure 4B shows a structural block diagram of a map updating apparatus according to an embodiment of the present disclosure;

[0036] Figure 5 shows a structural block diagram of an electronic device according to an embodiment of the present disclosure;

[0037] Figure 6 shows a schematic diagram of the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure. Detailed Implementation

[0038] In the following, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings to enable those skilled in the art to readily implement them. Furthermore, for clarity, portions unrelated to the description of exemplary embodiments have been omitted from the drawings.

[0039] In this disclosure, it should be understood that terms such as “comprising” or “having” are intended to indicate the presence of features, figures, steps, behaviors, components, parts or combinations thereof disclosed in this specification, and are not intended to exclude the possibility of the presence or addition of one or more other features, figures, steps, behaviors, components, parts or combinations thereof.

[0040] It should also be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0041] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0042] As mentioned above, with social development and the advancement of science and technology, safer and more efficient autonomous driving or advanced driver assistance technologies have been recognized as one of the future directions of the automotive travel industry, and many autonomous driving or advanced driver assistance behaviors rely on high-precision map data.

[0043] To improve the update frequency and reduce the cost of high-precision map data, existing technologies propose using data collection devices mounted on crowdsourced vehicles to compliantly collect data for map creation. However, these crowdsourced vehicles use low-cost data collection devices, resulting in lower data quality compared to data collected by professional equipment. Directly using data collected by these low-cost devices for map updates may lead to a decline in map quality. Therefore, data cleaning is necessary to remove erroneous or invalid data. However, existing data cleaning solutions mostly involve directly deleting substandard raw data, resulting in low data utilization efficiency.

[0044] This disclosure provides a data cleaning method that can fuse low-cost initial vector data and then calibrate it based on the base map data. This improves the quality of the low-cost data. The method then determines whether to perform data cleaning based on the quality of the improved data, thereby preserving the effective information to the greatest extent, avoiding the removal of effective data, and improving the effective utilization rate of the data.

[0045] Figure 1A shows a flowchart of a data cleaning method according to an embodiment of the present disclosure. As shown in Figure 1A, the data cleaning method includes the following steps S101-S103:

[0046] In step S101, the initial vector data from multiple acquisitions of the area to be updated are fused together to obtain the updated vector data corresponding to the area to be updated.

[0047] In step S102, the update vector data is calibrated based on the base map data corresponding to the area to be updated to obtain calibration vector data;

[0048] In step S103, in response to the fact that the data quality of the calibration vector data does not meet the normal data quality standard, the calibration vector data is cleared.

[0049] In one possible implementation, the data cleaning method is applicable to server-side devices such as computers, computing devices, servers, and server clusters that can perform data cleaning.

[0050] In one possible implementation, data cleaning refers to inspecting and processing data to remove erroneous or invalid data and retain as much usable information as possible.

[0051] In one possible implementation, the area to be updated is a road area divided according to a predetermined granularity. For example, it can be a road area with different granularities such as 100m to 100km, and the specific granularity can be determined according to the requirements.

[0052] In one possible implementation, the initial vector data includes various map feature vectors directly generated from the raw data. This raw data includes, but is not limited to, originally acquired positioning data, point cloud data, image data, IMU (Inertial Measurement Unit) data, etc., and can be acquired by data acquisition devices (including positioning devices, cameras, lidar, and IMU sensors, etc.) mounted on crowdsourced vehicles.

[0053] In one possible implementation, the initial vector data corresponding to the area to be updated can be generated by the original data collected by multiple acquisition devices in the area to be updated. The initial vector data can be generated at the acquisition end and then uploaded to the server, or it can be generated by the server after the acquisition end uploads the collected original data to the server. No limitation is made here.

[0054] In one possible implementation, when the area to be updated needs to be updated, the initial vector data of the area to be updated is acquired. Since the collection range of the low-cost collection equipment of the crowdsourced vehicles is limited, in order to improve the integrity and reliability of the data and retain the usable information to the maximum extent, the initial vector data generated by multiple collections can be merged. For example, the same map feature vector in the initial vector data of multiple collections can be merged and stitched into one, the missing parts in the map feature vector can be filled in, and abnormal points can be eliminated (such as convex points in linear feature vectors are abnormal points). In this way, the updated vector data corresponding to the area to be updated is obtained by fusion.

[0055] In one possible implementation, the geometric accuracy of the updated vector data obtained by fusing the initial vector data generated from multiple acquisitions is often not high. In order to improve the effective utilization of the data, retain the usable information to the maximum extent, and avoid data being rejected due to insufficient accuracy, the updated vector data can be calibrated based on the base map data to improve the geometric accuracy level of the low-precision updated vector data. For example, the updated vector data can be matched with the base map data, and the same map feature in the updated vector data can be matched together. The position of each map feature vector in the updated vector data can be calibrated according to the positional deviation between the same map feature matched in the updated vector data and the base map data.

[0056] In one possible implementation, various quality indicators such as data completeness, accuracy, and precision can be used to determine whether the data quality of the calibration vector data meets the normal data quality standard. For example, if the completeness of the calibration vector data exceeds the predetermined normal completeness, the accuracy of the calibration vector data exceeds the predetermined normal accuracy, or the precision of the calibration vector data exceeds the predetermined normal precision, then it is determined that the data quality of the calibration vector data meets the normal data quality standard; otherwise, it is determined that the data quality of the calibration vector data does not meet the normal data quality standard.

[0057] In one possible implementation, if the calibration vector data meets the normal data quality standard, it can be used for map updates to ensure the quality of the updated map. If the calibration vector data does not meet the normal data quality standard, it can be cleared and not used for map updates to avoid a decrease in map quality after the update.

[0058] This implementation method can fuse multiple initial vector data corresponding to the area to be updated, remove redundant data, and make the data more complete and reliable to obtain updated vector data corresponding to the area to be updated. Then, the updated vector data is calibrated based on the base map data to obtain calibrated vector data. In this way, the calibrated vector data is improved to the geometric accuracy level of the base map data. Based on the calibrated vector data, it is determined whether the data quality of the calibrated vector data is normal data quality. If the data quality of the calibrated vector data does not meet the normal data quality standard, the calibrated vector data is cleared to avoid the quality degradation after map update. In this way, calibrating after fusing the initial vector data can improve the quality of low-cost data, avoid the deletion of effective data, improve the effective utilization rate of data, and use the improved calibrated vector data to update the map, making the update result more reliable.

[0059] In addition, this implementation method matches the updated vector data with the base map data on the server side, which can effectively prevent the problem of inconsistent versions of the base map data collected on the vehicle and the base map data in the cloud under the asynchronous update mode, and obtain more reliable matching results.

[0060] In one possible implementation, the method further includes:

[0061] Calculate the completeness of the calibration vector data;

[0062] The accuracy of calculating the relative positional relationships between map feature vectors in the calibration vector data;

[0063] The calibration vector data is vector matched with the base map data to determine the unchanged map feature vectors, and the deviation between the calibration vector data and the unchanged map feature vectors in the base map data is calculated.

[0064] Based on at least one of the completeness, the accuracy, and the deviation, determine whether the data quality of the calibration vector data meets the normal data quality standard.

[0065] In this implementation, the shape and structure of certain map feature vectors in the map follow certain rules. For example, roads have left and right boundaries, and road boundaries always appear in pairs. If a part of a road boundary does not have a corresponding opposite road boundary, that part of the opposite road boundary is incomplete or missing. Or, the shapes of prohibition signs on roads are rectangular or triangular. If the shape of a prohibition sign is irregular, then the prohibition sign is incomplete. The completeness of map feature vectors in the calibration vector data can be determined according to the shape and structure of map features in reality, and then the completeness of the calibration vector data can be calculated. Alternatively, the completeness can also be determined based on the base map data. The updated vector data is matched with the base map data, and the matched map feature vectors are called unchanged map feature vectors. The completeness of the updated vector data and the unchanged map feature vectors in the base map data can be calculated based on the unchanged map feature vectors in the base map data, and the completeness of the unchanged map feature vectors is used as the completeness of the calibration vector data.

[0066] In this implementation, the relative positional relationships between certain map feature vectors in the map also follow certain rules. For example, lane lines in a road are used to divide lanes, so the distance between two adjacent lane lines should be the distance of one lane. If the distance between two lane lines is not the distance of a normal lane, it indicates that the relative positional relationship between these two lanes is inaccurate; or, lane lines should be located between the left and right road boundaries, and so on. The accuracy of the relative positional relationships between map feature vectors in the calibration vector data can be determined according to the relative positional relationships of map features in reality, and then the accuracy rate of the relative positional relationships between map feature vectors in the calibration vector data can be calculated.

[0067] In this embodiment, the calibration vector data corresponding to the area to be updated is compared with the existing base map data corresponding to the area to be updated. This includes both unchanged map feature vectors and potentially changed map feature vectors. The calibration vector data and the base map data can be vector matched. If the map feature vector has not changed, then both the calibration vector data and the base map data contain the same map feature vector. The same map feature vector in the calibration vector data and the base map data can be matched. The matched map feature vector can be called the unchanged map feature vector. The deviation between the calibration vector data and the unchanged map feature vector in the base map data can be calculated.

[0068] In this embodiment, the data quality of the calibration vector data can be determined based on one of the completeness, accuracy, and deviation. For example, if the completeness does not exceed a predetermined completeness threshold, the data quality of the calibration vector data can be determined to be below the normal data quality standard; if the accuracy does not exceed a predetermined accuracy threshold, the data quality of the calibration vector data can be determined to be below the normal data quality standard; and if the jitter of the deviation exceeds a preset jitter threshold, the data quality of the calibration vector data can be determined to be below the normal data quality standard. Alternatively, the data quality of the calibration vector data can be determined by comprehensively calculating two or more of the completeness, accuracy, and deviation. It should be noted that the steps of calculating completeness, accuracy, and deviation are not sequential.

[0069] In one possible implementation, calibrating the update vector data based on the base map data corresponding to the area to be updated to obtain calibration vector data includes:

[0070] The updated vector data is matched with the base map data to determine the unchanged map feature vectors;

[0071] Calculate the relative positional relationship between the updated vector data and the unchanged map feature vectors in the base map data;

[0072] Based on the relative positional relationships between the unchanged map feature vectors, the updated vector data is corrected using the base map data as a reference to obtain calibrated vector data.

[0073] In this embodiment, vector matching can be performed between the updated vector data and the base map data. Vectors containing the same map feature in the updated vector data and the base map data can be matched. These matched map feature vectors are referred to as unchanged map feature vectors. The relative positional relationships between the updated vector data and multiple unchanged map feature vectors in the base map data can be calculated, and then an average or weighted average can be obtained to determine the relative positional relationships between the unchanged map feature vectors. These relative positional relationships can be positional deviation vectors. Each map feature vector in the updated vector data can be translated towards the base map data corresponding to the area to be updated according to this positional deviation vector. This allows the updated vector data to be calibrated against the base map data, resulting in calibrated vector data.

[0074] This implementation method can compare the newly acquired updated vector data with the base map data (i.e., the old map) to obtain the unchanged map feature vectors. Based on the relative positional relationship between the updated vector data and the unchanged map feature vectors in the base map data, the updated vector data is calibrated, which can improve the vector accuracy of the changed map feature vectors in the area to be updated, and improve the accuracy of the changed areas that need to be updated in the area to be updated.

[0075] In one possible implementation, fusing multiple initial vector data passes corresponding to the region to be updated to obtain updated vector data corresponding to the region to be updated includes:

[0076] Obtain multiple initial vector data corresponding to the region to be updated;

[0077] The initial vector data is split into batches to obtain multiple sets of initial vector data, wherein the relative positional relationship of each map feature vector in each set of initial vector data satisfies a predetermined feature relative positional rule.

[0078] Using a set of initial vector data as a reference, the other sets of initial vector data are corrected to obtain calibration data for the other sets of initial vector data;

[0079] The calibration data of the first set of initial vector data and the other sets of initial vector data are fused to obtain the updated vector data in the region to be updated.

[0080] In this implementation, the acquisition quality of the data acquisition device is uncontrollable, and the acquired data is initial vector data produced from multiple runs of raw data. This initial vector data suffers from quality issues such as redundancy, incompleteness, and distortion, making it impossible for existing map update schemes to further improve data quality. However, this implementation, regardless of the quality of the acquired initial vector data or the number of acquisition runs, can first uniformly split the data batches into multiple groups of initial vector data according to the relative relationships between vectors, then eliminate inconsistencies between these groups, and finally fuse them to obtain updated vector data.

[0081] In this implementation, batch splitting refers to splitting a set of data according to certain characteristics of that data as classification criteria, resulting in each group of data being relatively consistent. This group of initial vector data differs from the concept of a single pass of data collection. For example, for an area to be updated, one pass of original data collection might collect data from the left side of the area, with the right side obscured, while another pass collects data from the right side, with the left side obscured. If the relative relationships of the map feature vectors on the left and right sides in the initial vector data produced by these two passes are correct, then these two passes of initial vector data can be grouped into one set of initial vector data.

[0082] In this implementation, the initial vector data of the area to be updated can be generated from raw data collected multiple times by multiple acquisition devices within the area. Since the acquisition quality and trajectory differ between each acquisition device, errors exist between the initial vector data from each acquisition. Therefore, it is necessary to divide the initial vector data from multiple acquisitions within the area to be updated into multiple groups according to the characteristics of the vectors. The relative positional relationships of the map feature vectors within each group of initial vector data satisfy predetermined relative positional rules for features. There is a systematic deviation between the multiple groups of initial vector data. Here, the predetermined relative positional rules for features refer to the distance between adjacent lane line vectors within the group of initial vector data being the distance of one lane, lane line vectors being in the middle of road boundary vectors, signs being on one side of road boundary vectors, and so on.

[0083] In this embodiment, in order to eliminate inconsistencies between multiple sets of initial vector data, one set of initial vector data can be used as a reference to correct the other sets of initial vector data, thereby obtaining calibration data for the other sets of initial vector data.

[0084] In this embodiment, after eliminating inconsistencies between multiple sets of initial vector data, the calibration data of one set of initial vector data and the other sets of initial vector data can be fused to obtain the updated vector data corresponding to the area to be updated. In this field, the vector fusion process mainly includes redundancy elimination, vector completion, and vector smoothing. Redundancy elimination refers to determining whether map feature vectors in each set of vector data belong to the same map feature based on their position and direction, merging and stitching the vector data of the same map feature into a single map feature vector, and removing redundant vectors. Vector completion refers to connecting and completing missing vectors between two map feature vectors. Vector smoothing refers to eliminating abnormal points in map feature vectors, such as protruding points in road boundary vectors or lane line vectors. For example, Figure 1B shows a schematic diagram of the fusion process according to an embodiment of this disclosure. As shown in Figure 1B, after fusing the calibration data of the two sets of initial vector data in each area to be updated, higher quality updated vector data can be obtained.

[0085] This implementation method first divides the initial vector data corresponding to the area to be updated into multiple sets of initial vector data according to their relative positional relationships. Then, using one set of initial vector data as a reference, other sets of initial vector data are corrected to this set of initial vector data to eliminate inconsistencies between multiple sets of initial vector data. Finally, the updated vector data is obtained by fusion. In this way, by splitting and fusing the data in batches, the quality problems existing in the initial vector data can be effectively handled. It is compatible with the defects of missing original observation descriptions when updating based on third-party data. It can also improve the data quality and filter out redundant data for the acquired low-quality initial vector data, thereby further improving the effective utilization rate of the data.

[0086] In one possible implementation, the step of batch splitting the initial vector data to obtain multiple sets of initial vector data includes:

[0087] Based on the relative positions between map feature vectors in the initial vector data, two map feature vectors whose relative positions are within a predetermined ghosting range are identified as a candidate ghosting vector pair.

[0088] In response to the fact that the number of candidate ghost vector pairs exceeds one pair, map feature vectors that do not belong to the same candidate ghost vector pair and whose relative positions within and between groups satisfy the predetermined feature relative position rules are grouped together to obtain multiple groups of ghost vectors.

[0089] Based on their relative positions to the map feature vectors in the multiple sets of ghost vectors, the map feature vectors that are not determined as candidate ghost vector pairs are assigned to the multiple sets of ghost vectors to obtain the multiple sets of initial vector data.

[0090] In this implementation, the relative positions between map element vectors can be relative horizontal distance and / or relative elevation distance. The relative horizontal distance refers to the distance between the latitude and longitude positions of two map element vectors, and the relative elevation distance refers to the distance between the elevations of two map element vectors. Vectors of the same map element collected in different trips will not overlap due to collection errors; instead, their relative horizontal and relative elevation distances will be relatively close, resulting in ghosting. Therefore, the ghosting range corresponding to each map element vector can be preset. For example, for lane line vectors, the positive value of the relative horizontal distance between adjacent lane line vectors can be set as follows: The normal range is greater than or equal to 2.5m. Since the purpose of determining the ghosting is to eliminate inconsistencies between ghostings, there is a certain calculation error (e.g., less than or equal to 0.2m) when eliminating inconsistencies. If the relative horizontal distance is too small, it can be considered that the vectors of the two map elements are consistent and no inconsistency elimination is needed. Therefore, the ghosting range of the relative horizontal distance can be set to greater than 0.2m and less than or equal to 2.5m, and the range of the relative horizontal distance less than or equal to 0.2m is a redundant range. The normal range of the relative elevation distance between adjacent lane line vectors is greater than or equal to 0.5m, and the ghosting range of the relative elevation distance is less than 0.5m. Multiple map element vectors whose relative positions are within the predetermined ghosting range can be considered as a candidate ghosting vector pair. For example, Figure 1C shows an example diagram of the initial vector data in the area to be updated according to an embodiment of the present disclosure. As shown in Figure 1C, the initial vector data of the area to be updated includes lane line vector 11, lane line vector 12, lane line vector 13, lane line vector 14, lane line vector 15, lane line vector 16, and lane line vector 17. Among them, the relative horizontal distance between lane line vector 11 and lane line vector 12 is less than or equal to 0.2m, and the relationship between the two is redundant. The relative horizontal distance between lane line vector 12 and lane line vector 13 is greater than 0.2m and less than or equal to 2.5m, so lane line vector 12 and lane line vector 13 can be identified as a candidate pair of ghosting vectors. Similarly, lane line vector 14 and lane line vector 15 are a candidate pair of ghosting vectors, and lane line vector 16 and lane line vector 17 are a candidate pair of ghosting vectors, for a total of three pairs of candidate ghosting vectors.

[0091] In this implementation, if there are no candidate ghost vector pairs, it is assumed that there is no vector ghosting, and the initial vector data can be directly determined as a set of initial vector data and directly fused to obtain updated vector data. If there is only one candidate ghost vector pair, grouping calibration can be skipped, and it can be directly fused as a set of initial vector data to obtain updated vector data. If the number of candidate ghost vector pairs exceeds one pair, it indicates that there are multiple candidate ghost vector pairs formed by multiple map feature vectors collected in different passes. In this case, ghosting can be grouped to eliminate the consistency between different groups before fusion to obtain updated vector data.

[0092] In this implementation, the two map feature vectors in each candidate ghosting vector pair are ghostings caused by map feature vectors collected in different passes. Therefore, the map feature vectors in the same candidate ghosting vector pair can be split into different groups. Simultaneously, considering the relative positions of map feature vectors within and between groups, map feature vectors that do not belong to the same candidate ghosting vector pair and whose relative positions within and between groups satisfy predetermined relative position rules can be grouped together. Satisfying predetermined relative position rules within a group means that the relative positions between map feature vectors within the group must conform to normal map feature distribution rules. For example, the distance between adjacent lane line vectors within the same group is the distance of one lane, the lane line vector is in the middle of the road boundary vector, and the sign is on one side of the road boundary vector, etc. Satisfying predetermined relative position rules between groups means that for the corresponding map feature vectors located in a local road segment within two sets of ghosting vectors, the angle between any two relative offset directions is less than a predetermined angle threshold. For example, taking the lane line vectors shown in Figure 1C as an example, since lane line vectors 12 and 13 are a candidate pair of ghosted vectors, lane line vectors 14 and 15 are a candidate pair of ghosted vectors, and lane line vectors 16 and 17 are a candidate pair of ghosted vectors, lane line vectors 12 and 13 are in different groups, lane line vectors 14 and 15 are in different groups, and lane line vectors 16 and 17 are in different groups. That is, lane line vectors 12 and 13 are the corresponding map element vectors in the two sets of ghosted vectors. The distance between lane line vectors 14, 15 and 12 is one lane distance. When they are grouped with lane line vector 12, the relative position within the group meets the requirements. However, in any local road segment, as shown in local road segment 18 in Figure 1C, lane line vector 12 is relative to lane line vectors... The relative offset direction of lane vector 13 is x1, and the relative offset direction of lane vector 13 relative to lane vector 12 is y1; the relative offset direction of lane vector 14 relative to lane vector 15 is x2, and the angle between x2 and y1 is greater than a preset angle threshold (e.g., 90 degrees), while the angle between x2 and x1 is less than the preset angle threshold. The relative offset direction of lane vector 15 relative to lane vector 14 is y2, and the angle between y2 and x1 is greater than the preset angle threshold, while the angle between y2 and y1 is less than the preset angle threshold. Therefore, in order to satisfy the requirement that the angle between any two relative offset directions is less than the predetermined angle threshold, lane vector 12 and lane vector 14 can be grouped together, and lane vector 13 and lane vector 15 can be grouped together.Similarly, lane line vector 16 can be grouped with lane line vectors 12 and 14 into one group of overlapping vectors, and lane line vector 17 can be grouped with lane line vectors 13 and 15 into another group of overlapping vectors. Lane line vectors 11 and 12 are redundant. Since lane line vector 11 is not determined as a map feature vector in the candidate overlapping vector pair, it can be assigned to lane line vector 12, which is redundant with it. This results in two sets of initial vector data: one set includes lane line vectors 11, 12, 14, and 16, and the other set includes lane line vectors 13, 15, and 17.

[0093] It should be noted that when dividing ghost vectors, if a group of ghost vectors includes a single feature vector, the feature vector can be merged into the group of ghost vectors containing the feature vector that is closest to it. For example, if lane line vector 11 and lane line vector 13 are the same candidate ghost vector pair, then lane line vector 11 and lane line vector 13 are split into different groups. In this case, lane line vector 11 is a ghost vector in its own group. At this time, lane line vector 11 can be assigned to the group of ghost vectors containing the lane line vector 12 that is closest to it.

[0094] In one possible implementation, the step of using a set of initial vector data as a reference to correct other sets of initial vector data to obtain calibration data for the other sets of initial vector data includes:

[0095] Calculate the data quality of the multiple sets of initial vector data, and obtain the set of initial vector data with the highest data quality;

[0096] Calculate the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data;

[0097] Based on the relative positional relationship, the other sets of initial vector data are corrected using the set of initial vector data with the highest data quality as a reference, thereby obtaining calibration data for the other sets of initial vector data.

[0098] In this embodiment, in order to improve data quality, the other sets of initial vector data can be corrected based on the set of initial vector data with the highest data quality, and the other sets of initial vector data can be corrected to the set of initial vector data to obtain calibration data of the other sets of initial vector data.

[0099] In this implementation, to obtain the set of initial vector data with the highest data quality, the data quality of multiple sets of initial vector data can be calculated. For example, the smoothness of the linear feature vectors in multiple sets of initial vector data can be calculated, and / or the data volume of multiple sets of initial vector data can be calculated. The higher the smoothness and the more data volume, the higher the data quality of the initial vector data. In this way, the set of initial vector data with the highest data quality can be calculated.

[0100] In this embodiment, in order to eliminate inconsistencies between multiple sets of data, the relative positional relationship, such as positional offset coordinates, between the initial vector data with the highest data quality and other initial vector data can be calculated. In this way, the other initial vector data can be translated according to the calculated positional offset coordinates, and the other initial vector data can be translated to the position of the initial vector data with the highest data quality, so as to obtain the calibration data of the other initial vector data.

[0101] In one possible implementation, calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data includes:

[0102] Calculate the relative positional relationship between each set of initial vector data and base map data;

[0103] Based on the relative positional relationship of the set of initial vector data with the highest data quality and the relative positional relationship of the other sets of initial vector data, calculate the initial relative positional relationship between the set of initial vector data with the highest data quality and the other sets of initial vector data.

[0104] Based on the initial relative positional relationship, vector matching is performed on the set of initial vector data with the highest data quality and other sets of initial vector data to obtain the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data.

[0105] In this embodiment, each set of initial vector data can be vector matched with the base map data. Based on the matched map feature vectors, the relative positional relationship R between each set of initial vector data and the base map data is calculated. Based on the relative positional relationship R1 corresponding to the set of initial vector data with the highest data quality and the relative positional relationship R2 corresponding to other sets of initial vector data, the initial relative positional relationship R1_2 between the set of initial vector data with the highest data quality and other sets of initial vector data can be obtained. For example, if the relative positional relationship is a positional deviation coordinate, then R1_2 = R1 - R2; if the relative positional relationship is a positional deviation ratio, then R1_2 = R1 / R2.

[0106] In this embodiment, other groups of initial vector data can be translated and calibrated according to the initial relative position relationship to calibrate the other groups of initial vector data to the group of initial vector data with the highest data quality, thus obtaining the other groups of initial vector data with initial calibration. Then, the other groups of initial vector data with initial calibration are vector matched with the group of initial vector data with the highest data quality to obtain the calibrated relative position relationship between the other groups of initial vector data with initial calibration and the group of initial vector data with the highest data quality. The initial relative position relationship is adjusted using the calibrated relative position relationship to obtain the final relative position relationship between the group of initial vector data with the highest data quality and other groups of initial vector data.

[0107] This implementation method can use base map data to calculate the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data. Since the base map data is relatively more complete and reliable, the reliability of the calculated relative positional relationship is relatively high.

[0108] In one possible implementation, calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data includes:

[0109] Match the initial vector data of each group to calculate the direct relative positional relationship between the initial vector data of each group.

[0110] Based on the direct relative positional relationship between the first and second groups of initial vector data and the direct relative positional relationship between the second and third groups of initial vector data, the cross-validation relative positional relationship between the first and third groups of initial vector data is calculated, wherein the first, second, and third groups of initial vector data are any three groups of initial vector data.

[0111] If the difference between the cross-validation relative positional relationship between any two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is greater than a preset difference, then the initial vector data in the area to be updated is cleared.

[0112] If the difference between the cross-validation relative positional relationship between any two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is less than or equal to a preset difference, then the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data is obtained from the direct relative positional relationships between the sets of initial vector data.

[0113] In this implementation, if there is no base map data in the area to be updated, vector matching can be performed between any two initial vector data. Based on the map feature vectors matched between the two sets of initial vector data, the direct relative positional relationship between the two sets of initial vector data is calculated. Since this direct relative positional relationship is calculated by directly performing vector matching between the two sets of initial vector data without the aid of base map data, the calculated relative positional relationship is inaccurate and cross-validation is required.

[0114] In this implementation, during cross-validation, any three sets of initial vector data can be selected from each set of initial vector data for cross-calculation. These three sets of initial vector data can be denoted as: the first set, the second set, and the third set of initial vector data. The direct relative positional relationship between the first and second sets of initial vector data, and the direct relative positional relationship between the second and third sets of initial vector data, can be subtracted or divided to obtain the relative positional relationship for cross-validation between the first and third sets of initial vector data. If there are two sets of initial vector data, and the difference between the relative positional relationship for cross-validation between these two sets of initial vector data and the direct relative positional relationship between these two sets of initial vector data is greater than a preset difference, then it is considered that the differences between multiple sets of data within the area to be updated cannot be effectively eliminated, and it is considered an abnormal area. These data cannot be used for map updates, so the initial vector data within the area to be updated needs to be cleared. If, for any two sets of initial vector data, the difference between the cross-validation relative positional relationship between the two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is less than or equal to a preset difference, then the differences between the sets of initial vector data can be eliminated, and these data can be used. In this case, the direct relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data can be obtained from the direct relative positional relationships between the sets of initial vector data. This direct relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data can be used as the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data.

[0115] In one possible implementation, obtaining multiple passes of initial vector data corresponding to the region to be updated includes:

[0116] Obtain initial vector data;

[0117] The initial vector data is matched with the base map to determine the location area of ​​the initial vector data in the base map;

[0118] If the number of data collection passes for the area to be updated exceeds the preset number of passes or the time without updating exceeds the preset time, then based on the location area of ​​the initial vector data in the base map, multiple passes of initial vector data corresponding to the area to be updated are obtained.

[0119] In this embodiment, after the acquisition device at the acquisition end acquires the raw data, it can generate initial vector data and upload it to the server. Alternatively, it can directly upload the acquired raw data to the server, which will then generate the initial vector data. After obtaining this initial vector data, the server can perform position matching between the initial vector data and the base map, mapping the initial vector data to the spatial reference of the base map to determine the positional region of these initial vector data in the base map.

[0120] In this implementation, for the area to be updated, if the number of data collection trips within the area exceeds a preset number, it indicates that there is enough data collected in the area to be updated and it can be updated. Alternatively, if the period of no update in the area exceeds a preset period, map updates should be performed on the area to be updated to ensure map freshness, even if the number of data collection trips is small. When the area to be updated needs to be updated, the initial vector data within the area to be updated can be aggregated into a batch based on the location of the initial vector data on the base map, and data cleaning can be performed. The cleaned and retained calibration vector data with normal data quality can then be used for map updates.

[0121] Figure 2A shows a flowchart of a map updating method according to an embodiment of the present disclosure. As shown in Figure 2A, the map updating method includes the following steps S201-S203:

[0122] In step S201, initial vector data corresponding to multiple regions to be updated are obtained, and each region to be updated corresponds to multiple passes of initial vector data;

[0123] In step S202, the above-described map data cleaning method is used to remove calibration vector data of the area to be updated that does not meet the normal data quality standard, and to obtain calibration vector data of the area to be updated that meets the normal data quality standard.

[0124] In step S203, the map is updated using the calibration vector data of the area to be updated, which meets the normal data quality standard.

[0125] In one possible implementation, the map update method is applicable to server-side devices such as computers, computing devices, servers, and server clusters that can perform map updates.

[0126] In one possible implementation, the area to be updated is a road segment divided according to a predetermined granularity, and the initial vector data includes various map feature vectors directly generated from the original data. The multiple initial vector data corresponding to the area to be updated can be generated from the original data collected multiple times within the area by multiple acquisition devices. This initial vector data can be generated at the acquisition end and then uploaded to the server, or it can be generated by the server after the acquisition end uploads the collected original data; no limitation is imposed here.

[0127] In one possible implementation, after obtaining initial vector data corresponding to multiple areas to be updated, for each area to be updated, the data cleaning method described in any of the above embodiments can be used to fuse the multiple initial vector data corresponding to the area to be updated to obtain updated vector data corresponding to the area to be updated; the updated vector data is calibrated based on the base map data corresponding to the area to be updated to obtain calibrated vector data; for the calibrated vector data of the area to be updated whose data quality does not meet the normal data quality standard, the calibrated vector data of the area to be updated can be cleared; for the calibrated vector data of the area to be updated whose data quality meets the normal data quality standard, the calibrated vector data of the area to be updated whose data quality meets the normal data quality standard can be used for map updating. For example, Figure 2B shows a flowchart of a map updating method according to an embodiment of the present disclosure. As shown in Figure 2B, the data quality of the calibrated vector data of area 21 to be updated does not meet the normal data quality standard and can be cleared; the data quality of the calibrated vector data of other areas to be updated meets the normal data quality standard, so the calibrated vector data of other areas to be updated can be used for map updating.

[0128] This embodiment can use the above-described data cleaning method to clean the data of multiple areas to be updated, removing calibration vector data from areas whose data quality does not meet the normal data quality standard, and obtaining calibration vector data from areas whose data quality meets the normal data quality standard for map updates. This avoids a decrease in map quality after updates. Moreover, after the above-described data cleaning method improves the quality of low-cost data, it determines whether to perform data cleaning based on the data quality of the improved data, avoiding the removal of effective data, improving the effective utilization rate of data, and using the improved calibration vector data for map updates makes the map update results more reliable.

[0129] Figure 3 illustrates an application scenario of the map data cleaning and map updating method according to an embodiment of this disclosure. As shown in Figure 3, the data processing server 301 can obtain initial vector data of the area to be updated from the collection end of the collection vehicle 302, and clean the data using the data cleaning method described above, removing data from areas with abnormal data quality, retaining the calibration vector data with normal data quality, and providing it to the map production server 303. The map production server 303 can update the map accordingly and provide the updated map to the navigation server 304. The navigation server 304 can provide navigation data to the location service terminal 305 based on the map data to perform navigation, route planning, and other services.

[0130] Figure 4A shows a structural block diagram of a map data cleaning apparatus according to an embodiment of the present disclosure. This apparatus can be implemented as part or all of an electronic device through software, hardware, or a combination of both. As shown in Figure 4A, the map data cleaning apparatus includes:

[0131] The fusion module 401 is configured to fuse multiple initial vector data corresponding to the region to be updated to obtain updated vector data within the region to be updated.

[0132] The calibration module 402 is configured to calibrate the update vector data based on the base map data corresponding to the area to be updated, so as to obtain calibration vector data;

[0133] The clearing module 403 is configured to clear the calibration vector data in response to the data quality of the calibration vector data failing to meet the normal data quality standard.

[0134] In one possible implementation, the device further includes:

[0135] The calculation module is configured to calculate the completeness of the calibration vector data; calculate the accuracy of the relative positional relationships between map feature vectors in the calibration vector data; perform vector matching between the updated vector data and the base map to determine the unchanged map feature vectors; and calculate the deviation between the updated vector data and the unchanged map feature vectors in the base map.

[0136] The judgment module is configured to determine whether the data quality of the calibration vector data is normal based on at least one of the completeness, the accuracy, and the deviation.

[0137] In one possible implementation, the calibration module is configured as follows:

[0138] The updated vector data is matched with the base map data to determine the unchanged map feature vectors;

[0139] Calculate the relative positional relationship between the updated vector data and the unchanged map feature vectors in the base map data;

[0140] Based on the relative positional relationships between the unchanged map feature vectors, the updated vector data is corrected using the base map data as a reference to obtain calibrated vector data.

[0141] In one possible implementation, the fusion module is configured as follows:

[0142] Obtain multiple initial vector data corresponding to the region to be updated;

[0143] The initial vector data is split into batches to obtain multiple sets of initial vector data, wherein the relative positional relationship of each map feature vector in each set of initial vector data is correct;

[0144] Using a set of initial vector data as a reference, other sets of initial vector data are corrected to obtain calibration data for other sets of initial vector data;

[0145] The highest quality initial vector data set is fused with the calibration data of the other initial vector data sets to obtain the updated vector data corresponding to the region to be updated.

[0146] In one possible implementation, the fusion module performs batch splitting of the initial vector data to obtain multiple sets of initial vector data, which is configured as follows:

[0147] Based on the relative positional relationship between map feature vectors in the initial vector data, two map feature vectors whose relative positional relationship is within a predetermined ghosting range are identified as a candidate ghosting vector pair.

[0148] In response to the fact that the number of candidate ghost vector pairs exceeds one pair, map feature vectors that do not belong to the same candidate ghost vector pair and whose relative positions within and between groups satisfy the predetermined feature relative position rules are grouped together to obtain multiple groups of ghost vectors.

[0149] Based on the relative positional relationship with the map feature vectors in the multiple sets of ghost vectors, the map feature vectors that are not determined to be candidate ghost vectors are divided into the multiple sets of ghost vectors to obtain the multiple sets of initial vector data.

[0150] In one possible implementation, the portion of the fusion module that uses a set of initial vector data as a reference to correct other sets of initial vector data to obtain calibration data of the other sets of initial vector data is configured as follows:

[0151] Calculate the data quality of the multiple sets of initial vector data, and obtain the set of initial vector data with the highest data quality;

[0152] Calculate the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data;

[0153] Based on the relative positional relationship, the other sets of initial vector data are corrected using the set of initial vector data with the highest data quality as a reference, thereby obtaining calibration data for the other sets of initial vector data.

[0154] In one possible implementation, the part of the fusion module that calculates the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data is configured as follows:

[0155] Match each set of initial vector data with the base map data, and calculate the relative positional relationship between each set of initial vector data and the base map data;

[0156] Based on the relative positional relationship of the set of initial vector data with the highest data quality and the relative positional relationship of the other sets of initial vector data, calculate the initial relative positional relationship between the set of initial vector data with the highest data quality and the other sets of initial vector data.

[0157] Based on the initial relative positional relationship, vector matching is performed on the set of initial vector data with the highest data quality and other sets of initial vector data to calculate the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data.

[0158] In one possible implementation, the part of the fusion module that calculates the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data is configured as follows:

[0159] Match the initial vector data of each group to calculate the direct relative positional relationship between the initial vector data of each group.

[0160] Based on the direct relative positional relationship between the first and second groups of initial vector data and the direct relative positional relationship between the second and third groups of initial vector data, the cross-validation relative positional relationship between the first and third groups of initial vector data is calculated, wherein the first, second, and third groups of initial vector data are any three groups of initial vector data.

[0161] If the difference between the relative positional relationship of cross-validation between two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is greater than a preset difference, then the initial vector data in the area to be updated is cleared.

[0162] If the difference between the cross-validation relative positional relationship between any two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is less than or equal to a preset difference, then the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data is obtained from the direct relative positional relationships between the sets of initial vector data.

[0163] In one possible implementation, the part of the fusion module that acquires the multiple passes of initial vector data corresponding to the region to be updated is configured as follows:

[0164] Obtain initial vector data;

[0165] The initial vector data is matched with the base map data to determine the location region of the initial vector data in the base map.

[0166] If the number of data collection passes for the area to be updated exceeds a preset number or the time without updating exceeds a preset time, then based on the location area of ​​the initial vector data in the base map, multiple passes of initial vector data corresponding to the area to be updated are obtained.

[0167] Figure 4B shows a structural block diagram of a map updating device according to an embodiment of the present disclosure. This device can be implemented as part or all of an electronic device through software, hardware, or a combination of both. As shown in Figure 4B, the map updating device includes:

[0168] The acquisition module 404 is configured to acquire initial vector data corresponding to multiple regions to be updated, with each region to be updated corresponding to multiple passes of initial vector data.

[0169] The cleaning module 405 is configured to use the map data cleaning method in the above embodiments to remove calibration vector data of the area to be updated that does not meet the normal data quality standard, and to obtain calibration vector data of the area to be updated that meets the normal data quality standard.

[0170] The update module 406 is configured to update the map using calibration vector data of the area to be updated, which meets the normal data quality standard.

[0171] The technical terms and features mentioned in this device implementation are the same or similar. For the explanation and description of the technical terms and features involved in this device, please refer to the explanation of the above method implementation, which will not be repeated here.

[0172] This disclosure also discloses an electronic device, and FIG5 shows a structural block diagram of the electronic device according to an embodiment of the present disclosure.

[0173] As shown in FIG5, the electronic device 500 includes a memory 501 and a processor 502, wherein the memory 501 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 502 to implement the method according to the embodiments of the present disclosure.

[0174] Figure 6 shows a schematic diagram of the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure.

[0175] As shown in Figure 6, the computer system 600 includes a processing unit 601, which can execute various processes described in the above embodiments according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage section 608 into a random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the computer system 600. The processing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0176] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed. The processing unit 601 can be implemented as a CPU, GPU, TPU, FPGA, NPU, etc.

[0177] In particular, according to embodiments of this disclosure, the methods described above can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising computer instructions that, when executed by a processor, implement the steps of the methods described above. In such embodiments, the computer program product can be downloaded and installed from a network via communication section 609, and / or installed from removable media 611.

[0178] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0179] The units or modules described in the embodiments of this disclosure can be implemented in software or programmable hardware. The described units or modules can also be located in a processor, and the names of these units or modules do not necessarily constitute a limitation on the unit or module itself.

[0180] In another aspect, this disclosure also provides a computer-readable storage medium, which may be a computer-readable storage medium included in the electronic device or computer system described above; or it may be a standalone computer-readable storage medium not assembled into a device. The computer-readable storage medium stores one or more programs, which are used by one or more processors to perform the methods described in this disclosure.

[0181] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A map data cleaning method, comprising: Multiple initial vector data corresponding to the area to be updated are fused to obtain updated vector data corresponding to the area to be updated. The multiple initial vector data corresponding to the area to be updated are generated by multiple acquisition devices collecting raw data in multiple passes in the area to be updated. The updated vector data is calibrated based on the base map data corresponding to the area to be updated to obtain calibrated vector data. If the data quality of the calibrated vector data does not meet the normal data quality standard, the calibrated vector data is cleared.

2. The method according to claim 1, wherein, The method further includes: calculating the completeness of the calibration vector data; calculating the accuracy of the relative positional relationship between map feature vectors in the calibration vector data; performing vector matching between the calibration vector data and the base map data to determine the unchanged map feature vectors, and calculating the deviation between the calibration vector data and the unchanged map feature vectors in the base map data; and determining whether the data quality of the calibration vector data meets the normal data quality standard based on at least one of the completeness, the accuracy, and the deviation.

3. The method according to claim 1, wherein, The step of calibrating the updated vector data based on the base map data corresponding to the area to be updated to obtain calibrated vector data includes: performing vector matching between the updated vector data and the base map data to determine the unchanged map feature vectors; calculating the relative positional relationship between the updated vector data and the unchanged map feature vectors in the base map data; and correcting the updated vector data based on the relative positional relationship between the unchanged map feature vectors, using the base map data as a reference, to obtain calibrated vector data.

4. The method according to claim 1, wherein, The step of fusing multiple initial vector data sets corresponding to the area to be updated to obtain updated vector data for the area to be updated includes: acquiring multiple initial vector data sets corresponding to the area to be updated; batch splitting the multiple initial vector data sets corresponding to the area to be updated to obtain multiple sets of initial vector data sets, wherein the relative positions of each map feature vector within each set of initial vector data sets satisfy a predetermined feature relative position rule; using one set of initial vector data sets as a reference, correcting other sets of initial vector data sets to obtain calibration data for the other sets of initial vector data sets; and fusing the first set of initial vector data sets and the calibration data for the other sets of initial vector data sets to obtain updated vector data for the area to be updated.

5. The method according to claim 4, wherein, The step of batch splitting the multiple initial vector data corresponding to the area to be updated to obtain multiple sets of initial vector data includes: determining two map feature vectors whose relative positions are within a predetermined ghosting range as a candidate ghosting vector pair based on the relative positions between map feature vectors in the multiple initial vector data; responding to the fact that the number of candidate ghosting vector pairs exceeds one pair, grouping map feature vectors that do not belong to the same candidate ghosting vector pair and whose relative positions within and between groups satisfy predetermined feature relative position rules into a group to obtain multiple sets of ghosting vectors; and assigning map feature vectors that are not determined to be candidate ghosting vector pairs to the multiple sets of ghosting vectors according to their relative positions with the map feature vectors in the multiple sets of ghosting vectors to obtain the multiple sets of initial vector data.

6. The method according to claim 4, wherein, The step of using a set of initial vector data as a reference to correct other sets of initial vector data to obtain calibration data for the other sets of initial vector data includes: calculating the data quality of the multiple sets of initial vector data to obtain the set of initial vector data with the highest data quality; calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data; and, based on the relative positional relationship, correcting the other sets of initial vector data using the set of initial vector data with the highest data quality as a reference to obtain calibration data for the other sets of initial vector data.

7. The method according to claim 6, wherein, The step of calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data includes: matching each set of initial vector data with the base map data, and calculating the relative positional relationship between each set of initial vector data and the base map data; calculating the initial relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data based on the relative positional relationship corresponding to the set of initial vector data with the highest data quality and the relative positional relationship corresponding to the other sets of initial vector data; and performing vector matching on the set of initial vector data with the highest data quality and other sets of initial vector data based on the initial relative positional relationship, and calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data.

8. The method according to claim 6, wherein, The step of calculating the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data includes: matching the initial vector data sets with each other to calculate the direct relative positional relationship between the initial vector data sets; calculating the cross-validation relative positional relationship between the first and second sets of initial vector data and the second and third sets of initial vector data based on the direct relative positional relationship between the initial vector data sets, wherein the first, second, and third sets of initial vector data are any three sets of initial vector data sets; if the difference between the cross-validation relative positional relationship between two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is greater than a preset difference, then the multiple passes of initial vector data corresponding to the area to be updated are cleared; if the difference between the cross-validation relative positional relationship between any two sets of initial vector data and the direct relative positional relationship between the two sets of initial vector data is less than or equal to a preset difference, then the relative positional relationship between the set of initial vector data with the highest data quality and other sets of initial vector data is obtained from the direct relative positional relationship between the initial vector data sets.

9. The method according to claim 4, wherein, The step of obtaining multiple initial vector data corresponding to the area to be updated includes: obtaining initial vector data; matching the initial vector data with the base map data to determine the location area of ​​the initial vector data in the base map; if the number of collection passes for the area to be updated exceeds a preset number of passes or the time without updating exceeds a preset time, then obtaining multiple initial vector data corresponding to the area to be updated based on the location area of ​​the initial vector data in the base map.

10. A map updating method, comprising: Acquire initial vector data corresponding to multiple areas to be updated, each area to be updated corresponding to multiple passes of initial vector data, the multiple passes of initial vector data corresponding to the area to be updated being generated by multiple acquisition devices collecting raw data in multiple passes in the area to be updated; using the method of any one of claims 1 to 9, remove the calibration vector data of the areas to be updated whose data quality does not meet the normal data quality standard, and acquire the calibration vector data of the areas to be updated whose data quality meets the normal data quality standard; use the calibration vector data of the areas to be updated whose data quality meets the normal data quality standard for map updating.

11. A map data cleaning apparatus, comprising: The fusion module is configured to fuse multiple initial vector data corresponding to the region to be updated to obtain the updated vector data corresponding to the region to be updated. The calibration module is configured to calibrate the updated vector data based on the base map data corresponding to the area to be updated, to obtain calibration vector data; the clearing module is configured to clear the calibration vector data in response to the data quality of the calibration vector data failing to meet the normal data quality standard.

12. A map updating device, comprising: The acquisition module is configured to acquire initial vector data corresponding to multiple regions to be updated, with each region to be updated corresponding to multiple passes of initial vector data. The cleaning module is configured to use the method described in any one of claims 1 to 9 to remove calibration vector data of the area to be updated that does not meet the normal data quality standard, and to obtain calibration vector data of the area to be updated that meets the normal data quality standard. The update module is configured to update the map using calibration vector data of the area to be updated, which meets the normal data quality standard.

13. An electronic device comprising a memory and a processor; wherein, The memory is used to store one or more computer instructions, which are executed by the processor to implement the method according to any one of claims 1 to 10.

14. A computer-readable storage medium having stored thereon computer instructions, wherein, When executed by a processor, the computer instructions implement the method described in any one of claims 1-10.

Images