Quality evaluation method for crowdsourcing map data and related device

Abstract

The invention discloses a crowdsourcing map data quality evaluation method and a related device, and belongs to the technical field of intelligent driving crowdsourcing maps, and the method comprises the steps: if a to-be-evaluated target space range on a crowdsourcing map has a first sub-range corresponding to reference map data, obtaining a first sub-range corresponding to the reference map data based on the corresponding reference map data; determining an index value of the map data of the first sub-range in the first class of quality evaluation indexes; if a second sub-range not corresponding to the reference map data exists in the target space range, determining an index value of the map data of the second sub-range in a second type of quality evaluation indexes based on multiple and / or multi-source data of the map data of the second sub-range; and performing quality evaluation on the map data of the target space range based on a corresponding evaluation model according to at least one of the two types of index values. According to the method and the device, different conditions of existence and absence of reference map data are fully considered, so that the method and the device can be adapted to an actual scene to realize quick, accurate and effective crowdsourcing map data quality evaluation.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent driving crowdsourcing map, and particularly relates to a crowdsourcing map data quality evaluation method and related device. BACKGROUND

[0002] Crowdsourcing map refers to a lightweight semantic map used on intelligent driving vehicles, the richness and accuracy of elements of which are between ordinary navigation map and traditional high-precision map. Usually, distributed data collection and production methods are used, road semantic data is acquired by multiple vehicles based on vision or other sensors, and these scattered road semantic data is collected, processed and fused to form a crowdsourcing map of the whole corresponding region.

[0003] In the field of intelligent driving, high-precision map is considered an important condition for vehicle enterprises to improve the ability of autonomous driving. However, traditional high-precision map making faces difficulties such as high acquisition and production cost, low update frequency, and long time consumption, and therefore has been regarded as one of the main obstacles restricting the expansion of high-level assisted driving to urban roads. In addition, it is difficult to obtain the qualification of high-precision map, and the map review process is complex, which leads to the generalization challenge of large-scale replication when high-level intelligent driving enters the urban scene. Based on this, the so-called "no map" solution in the field of intelligent driving, that is, not relying on high-precision map, using a lightweight map constructed by crowdsourcing, has become the main way adopted by current intelligent driving solution providers. Crowdsourcing map only needs to retain the necessary elements according to the needs of intelligent driving, which can make up for the limitations of vehicle perception, provide a global perspective, and achieve rapid update and dynamic iteration in the form of low cost.

[0004] In order to ensure that high-quality crowdsourcing map is provided for intelligent driving, the quality of crowdsourcing map data needs to be evaluated. However, there is no uniform evaluation standard for the quality evaluation of crowdsourcing map data at present, and it is difficult to quickly, accurately and effectively realize the quality evaluation of crowdsourcing map data. SUMMARY

[0005] Therefore, the application provides a crowdsourcing map data quality evaluation method and related device for quickly, accurately and effectively realizing systematic evaluation of the quality of crowdsourcing map data, so as to provide a basis for the application of crowdsourcing map to the field of intelligent driving.

[0006] The specific technical solutions are as follows:

[0007] A crowdsourcing map data quality evaluation method comprises the following steps:

[0008] determining a target spatial range of map data quality evaluation on a crowdsourcing map;

[0009] if a first sub-range corresponding to the reference map data exists on the target spatial range, determining an index value of the first sub-range in a first type of quality evaluation index based on the corresponding reference map data;

[0010] if a second sub-range without corresponding reference map data exists on the target spatial range, determining an index value of the second sub-range in a second type of quality evaluation index based on multiple times and / or multi-source data of the map data of the second sub-range;

[0011] based on the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, performing quality evaluation on the map data of the target spatial range based on a corresponding evaluation model to obtain a quality evaluation result of the map data of the target spatial range.

[0012] Optionally, the determining of the target spatial range on the crowd-sourced map to be subjected to map data quality evaluation comprises:

[0013] acquiring indication information of a spatial range on the crowd-sourced map;

[0014] determining a spatial range represented by a road segment or an area indicated by the indication information on the crowd-sourced map as the target spatial range.

[0015] Optionally, the determining of the index value of the first sub-range in the first type of quality evaluation index based on the corresponding reference map data comprises:

[0016] determining index values of different quality evaluation indexes respectively corresponding to different evaluation dimensions of the map data of the first sub-range based on high-precision map ground truth data and / or label-precision map ground truth data;

[0017] wherein the first type of quality evaluation index comprises the different quality evaluation indexes respectively corresponding to the different evaluation dimensions; the different evaluation dimensions comprise at least part of an integrity dimension, an accuracy dimension, and a timeliness dimension; the quality evaluation index in the integrity dimension comprises at least part of geometric integrity, attribute integrity, topological integrity, and semantic object integrity; the quality evaluation index in the accuracy dimension comprises at least part of geometric data precision, attribute data accuracy, and topological accuracy; and the quality evaluation index in the timeliness dimension comprises freshness of the crowd-sourced map data.

[0018] Optionally, the determining of the index value of the second sub-range in the second type of quality evaluation index based on the multiple times and / or multi-source data of the map data of the second sub-range comprises at least one of:

[0019] quantify a degree of spatial aggregation of the multi-time and / or multi-source data of the second sub-range of map data, to obtain a quantification result of the degree of spatial aggregation; and determine a spatial data accuracy of the second sub-range of map data according to the quantification result of the degree of spatial aggregation.

[0020] determine a consistency of attribute data of element attributes of the multi-time and / or multi-source data of the second sub-range of map data at the same position;

[0021] The second type of quality evaluation indicators includes at least one of the spatial data accuracy and the attribute data consistency.

[0022] Optionally, the quality evaluation of the map data of the target spatial range based on the index values of the first type of quality evaluation indicators and / or the index values of the second type of quality evaluation indicators includes:

[0023] If the first sub-range exists on the target spatial range and the second sub-range does not exist, a first sub-quality evaluation result of the map data of the first sub-range is determined based on a first index evaluation model and the index values of the indicators in the first type of quality evaluation indicators; and the quality evaluation result of the map data of the target spatial range is the first sub-quality evaluation result.

[0024] If the second sub-range exists on the target spatial range and the first sub-range does not exist, a second sub-quality evaluation result of the map data of the second sub-range is determined based on a second index evaluation model and the index values of the indicators in the second type of quality evaluation indicators; and the quality evaluation result of the map data of the target spatial range is the second sub-quality evaluation result.

[0025] If the first sub-range and the second sub-range exist on the target spatial range, a first sub-quality evaluation result of the map data of the first sub-range is determined based on the first index evaluation model and the index values of the indicators in the first type of quality evaluation indicators; a second sub-quality evaluation result of the map data of the second sub-range is determined based on the second index evaluation model and the index values of the indicators in the second type of quality evaluation indicators; and the quality evaluation result of the map data of the target spatial range is determined according to the first sub-quality evaluation result and the second sub-quality evaluation result.

[0026] Optionally, the first index evaluation model and the second index evaluation model are both bottom-up hierarchical structures; the bottom layer of the hierarchical structure of the first index evaluation model contains first quality evaluation indexes, the top layer contains first sub-quality evaluation results of the map data in the first sub-range, and the intermediate layers contain intermediate dimension factors obtained by abstracting the factors contained in the lower layers of the intermediate layers;

[0027] the bottom layer of the hierarchical structure of the second index evaluation model contains second quality evaluation indexes, the top layer contains second sub-quality evaluation results of the map data in the second sub-range, and the intermediate layers contain intermediate dimension factors obtained by abstracting the factors contained in the lower layers of the intermediate layers;

[0028] determining the first sub-quality evaluation results of the map data in the first sub-range based on the first index evaluation model and the index values of the first quality evaluation indexes, comprises:

[0029] determining the weights of the factors contained in each layer structure of the first index evaluation model based on the fuzzy analytic hierarchy process;

[0030] determining the first sub-quality evaluation results of the map data in the first sub-range according to the weights of the factors contained in each layer structure of the first index evaluation model and the index values of the first quality evaluation indexes;

[0031] determining the second sub-quality evaluation results of the map data in the second sub-range based on the second index evaluation model and the index values of the second quality evaluation indexes, comprises:

[0032] determining the weights of the factors contained in each layer structure of the second index evaluation model based on the fuzzy analytic hierarchy process;

[0033] determining the second sub-quality evaluation results of the map data in the second sub-range according to the weights of the factors contained in each layer structure of the second index evaluation model and the index values of the second quality evaluation indexes.

[0034] Optionally, determining the weights of the factors contained in each layer structure of the corresponding index evaluation model based on the fuzzy analytic hierarchy process, comprises:

[0035] constructing a fuzzy judgment matrix; the fuzzy judgment matrix is used to provide a quantitative representation of the importance of one factor compared with another factor for the importance comparison between two factors;

[0036] Based on the fuzzy judgment matrix, a pairwise comparison judgment of importance between factors in each layer of the corresponding index evaluation model is made to obtain a fuzzy complementary matrix; the fuzzy complementary matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model;

[0037] Based on a predetermined conversion relationship, the fuzzy complementary matrix is converted into a fuzzy consistent matrix; the fuzzy consistent matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model, and the fuzzy consistent matrix has a thinking consistency compared with the fuzzy complementary matrix;

[0038] Based on the fuzzy consistent matrix, the weight of a factor in each layer of the corresponding index evaluation model is determined; the corresponding index evaluation model is the first index evaluation model or the second index evaluation model.

[0039] Optionally, the method further comprises at least one of the following:

[0040] Based on a dynamic chart, a quality evaluation result of map data of the target space range, and factor information of factors included in each layer structure in the first index evaluation model and / or the second index evaluation model are displayed; the factor information includes the weight and / or the factor value of the factor; the factors included in each layer structure in the first index evaluation model include the first type of quality evaluation index or a processing result of the first type of quality evaluation index; the factors included in each layer structure in the second index evaluation model include the second type of quality evaluation index or a processing result of the second type of quality evaluation index;

[0041] In response to obtaining an information source viewing request for a target factor on the dynamic chart, information sources of the target factor are displayed;

[0042] In response to obtaining an accuracy range adjustment request for an accuracy index, an index value of the accuracy index is updated based on a target accuracy range indicated by the accuracy range adjustment request.

[0043] A quality evaluation device for crowd-sourced map data comprises a first determination module configured to determine a target space range on a crowd-sourced map for which map data quality evaluation is to be performed;

[0044] A second determination module is configured to, in a case where a first sub-range corresponding to reference map data exists in the target space range, determine an index value of the first sub-range in a first type of quality evaluation index based on the corresponding reference map data.

[0045] a third determining module, configured to, in a case that the target space range has a second sub-range without corresponding reference map data, determine, based on multi-time and / or multi-source data of map data of the second sub-range, an index value of the second sub-range in a second type of quality evaluation index;

[0046] a quality evaluation module, configured to, according to the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, perform quality evaluation on the map data of the target space range based on a corresponding evaluation model, to obtain a quality evaluation result of the map data of the target space range.

[0047] An electronic device comprises:

[0048] a memory, configured to store a computer program;

[0049] a processor, configured to realize the quality evaluation method of crowd-sourced map data according to any one of the above by calling and executing the computer program in the memory.

[0050] A computer readable medium, which stores a computer program, and the computer program is executed by a processor to realize the quality evaluation method of crowd-sourced map data according to any one of the above.

[0051] According to the above scheme, it can be known that the present application considers different cases of having reference map data and not having reference map data, and respectively establishes corresponding map data quality evaluation indexes and evaluation models for different cases, proposes a comprehensive evaluation index and evaluation method for the quality of crowd-sourced map data, and can make the quality evaluation of crowd-sourced data adapt to different scenes with or without reference map data. And for the case of having reference map data, the index value of the first type of quality evaluation index is determined based on the corresponding reference map data, for the case of not having reference map data, the index value of the second type of quality evaluation index is determined based on multi-time and / or multi-source data of map data, and the quality of crowd-sourced map data is evaluated based on the quality evaluation index value in the corresponding scene and the corresponding evaluation model, which can make the quality evaluation of crowd-sourced map data fast, accurate and effective in the actual scene. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.

[0053] Figure 1is a flow chart of a quality evaluation method of crowd-sourced map data provided by the present application;

[0054] Figure 2 is an exemplary hierarchical structure of a first index evaluation model provided by the present application;

[0055] Figure 3 is a component structure diagram of a quality evaluation device of crowd-sourced map data provided by the present application;

[0056] Figure 4 is a component structure diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0058] The embodiments of the present application provide a quality evaluation method of crowd-sourced map data and related devices, which are used to realize fast, accurate and effective quality evaluation of crowd-sourced map data by establishing quality evaluation indexes and evaluation models of the crowd-sourced map data. Referring to the method flow diagram shown in Figure 1 The embodiments of the present application provide a quality evaluation method of crowd-sourced map data, which can at least include the following steps 101 to 104, and these steps will be described in detail below.

[0059] Step 101, determining a target spatial range of map data quality evaluation on a crowd-sourced map.

[0060] Among them, the data quality evaluation of the crowd-sourced map can be performed according to a specified road section (including road-related data) or a specified area, or the data quality evaluation of the entire crowd-sourced map can be performed.

[0061] The map data of the crowd-sourced map can include map elements and related element information, such as lane lines, road edges, lane change positions, arrow / stopping line / zebra crossing markings on the road, red light positions, road signboards, etc.

[0062] The crowd-sourced map serves intelligent driving, and is usually provided to an intelligent driving vehicle in units of roads (road segments). Therefore, a required road (road segment) can be selected based on actual needs as a target spatial range to be subjected to map data quality evaluation, and subsequently, data related to the road segment along the road can be found to perform quality evaluation. The data related to the road segment along the road can include, but is not limited to, lane lines, road edges, ground markings, signboards, intersections, etc. related to the road segment. However, the data related to the road segment along the road is not limited to the above. A region including one or more road segments on the crowd-sourced map or a spatial range represented by the entire map can also be selected as the target spatial range to be subjected to map data quality evaluation.

[0063] In implementation, optionally, the indication information of the spatial range on the crowd-sourced map can be acquired, and a spatial range represented by a road segment or a region indicated by the indication information on the crowd-sourced map is determined as the target spatial range on the crowd-sourced map to be subjected to map data quality evaluation.

[0064] The region indicated by the indication information on the crowd-sourced map can be a partial region on the crowd-sourced map or the entire region on the crowd-sourced map (in this case, a spatial range represented by the entire crowd-sourced map is taken as the target spatial range). This is not limited, and is determined according to actual conditions.

[0065] In step 102, if a first sub-range corresponding to corresponding reference map data exists on the target spatial range, the map data of the first sub-range is determined to have an index value of a first type of quality evaluation index based on the corresponding reference map data.

[0066] After the target spatial range on the crowd-sourced map to be subjected to map data quality evaluation is determined, the reference map data of the target spatial range is further detected to determine whether a road segment or a region contained in the target spatial range corresponds to corresponding reference map data.

[0067] The reference map data can include high-precision map ground truth data and / or low-precision map ground truth data.

[0068] Embodiments of the present application establish different evaluation indexes for quality evaluation of map data of a crowd-sourced map for different cases of having reference map data and not having reference map data. For the case of having reference map data, a first type of quality evaluation index is established to perform quality evaluation of map data of a crowd-sourced map by using the established first type of quality evaluation index. For the case of not having reference map data, a second type of quality evaluation index is established to perform quality evaluation of map data of a crowd-sourced map by using the established second type of quality evaluation index.

[0069] On this basis, if it is determined through detection of reference map data of the target space range to be evaluated that there is a first sub-range corresponding to the reference map data on the target space range, then based on the corresponding reference map data, the map data of the first sub-range is determined in the index value of the first type of quality evaluation index, so as to subsequently perform quality evaluation on the map data of the target space range based on the index value of the first type of quality evaluation index of the map data of the first sub-range.

[0070] The first type of quality evaluation index includes different quality evaluation indexes corresponding to different evaluation dimensions, and the different evaluation dimensions include at least part of the integrity dimension, the accuracy dimension, and the timeliness dimension.

[0071] That is, for the case of having reference map data (such as high-precision map true value data or standard-precision map true value data), the quality of the map data can be evaluated from the integrity, accuracy, and timeliness dimensions, and each dimension is refined into several evaluation indexes to form the first type of quality evaluation index for the case of having reference map data for crowdsourcing map data.

[0072] The integrity represents the completeness of the data to be evaluated, and the evaluation indexes in the integrity dimension can include but are not limited to at least part of the geometric integrity, the attribute integrity, the topological integrity, and the semantic object integrity.

[0073] The geometric integrity refers to the coverage degree of the crowdsourcing map data in the research range (such as the target space range); the attribute integrity refers to the proportion of valid attributes of the crowdsourcing map data (such as lane lines, road edges, and ground markings); for different types of data, the importance of the attribute fields is different, so the selected attribute fields can be flexibly changed in actual application; the topological integrity is used to evaluate whether the association relationship between the crowdsourcing map elements is complete, for example, whether a lane has a preceding lane and a subsequent lane, and whether a road intersection is associated with a zebra crossing, a signal lamp, etc.; the semantic object integrity is used to evaluate whether the crowdsourcing map includes the necessary semantic objects, and the semantic objects in the crowdsourcing map mainly refer to signboards, signal lamps, and other geometric data containing semantic information.

[0074] The accuracy represents the difference between the data to be evaluated and the true value. The evaluation indexes in the accuracy dimension can include but are not limited to the geometric data precision, the attribute data accuracy, and the topological accuracy.

[0075] The geometric data precision is used to judge the lateral and longitudinal deviation of the crowd-sourced map element coordinates and the corresponding true value. The attribute data accuracy judges the consistency degree of the crowd-sourced element attribute and the true value. The attributes related to the lane in the crowd-sourced map mainly include: the belonging road, the road type, the lane line color, the lane line type, the lane number, the lane driving direction, etc. The true value does not have the lane line color and type attributes. The attributes of the road facility data are different according to the type, and the general attributes include: the belonging road, the facility type. The topology accuracy is used to judge the accuracy of the topology connection between the crowd-sourced map elements, such as whether the predecessor and successor lanes of a lane are correct.

[0076] The present situation refers to the current state or update degree of the crowd-sourced map, which is represented by freshness, and the value is 0-1. The reference time is used for evaluation. The closer the last update time of the element in the crowd-sourced map to the reference time, the closer the value to 1, indicating that the data is fresher.

[0077] The following further provides the index value determination method of each first type quality evaluation index under each dimension.

[0078] I. Integrity

[0079] 11) Geometric data integrity (geometric integrity)

[0080] Optionally, referring to Table 1 below, the geometric data integrity can be calculated by calculating the ratio of the identified element length (number) in the evaluation range (such as the first sub-range in the target space range corresponding to the corresponding reference map data) and the prior reference element length (number). The source of the prior reference element length / number is a high-precision map or a standard-precision map.

[0081] Table 1

[0082]

[0083] L rb represents the data length (number) of the element in the evaluation range, L true represents the data length (number) of the element in the evaluation range, L

[0084] 12) Attribute integrity

[0085] The attribute integrity can be calculated by counting the proportion of the number of elements with complete attributes (the attribute fields selected in actual use can be flexibly changed) in the evaluation range based on the prior reference data. The attribute integrity can include but is not limited to the attribute integrity of lane line data and road facility data.

[0086] For lane line data, the attribute completeness can be calculated by counting the proportion of the number of lane lines with non-empty and non-undecided attributes (lane_type, lane_color, pos, etc.) in the total number of lane lines based on the prior reference data of lane line attributes.

[0087] For road facility data, the attribute completeness can be calculated by counting the proportion of the number of road facility elements with non-empty and non-undecided attributes (roadfactype, etc.) in the total number of road facility elements based on the prior reference data of road facility data, as shown in Table 2.

[0088] Table 2

[0089]

[0090] where P notNull represents the number of elements with non-empty / undecided attributes in the range to be evaluated, P all represents the total number of prior elements in the range to be evaluated.

[0091] 13) Topological completeness

[0092] As shown in Table 3, the topological completeness can be calculated by traversing the connectivity between any two points in the range to be evaluated through path planning, and counting the proportion of the number of connected point pairs based on the corresponding prior reference data.

[0093] Table 3

[0094]

[0095] k conn represents the number of connected point pairs in the range to be evaluated, k all represents all point pairs in the corresponding prior reference data of the range to be evaluated.

[0096] 14) Semantic object completeness

[0097] As shown in Table 4, the semantic object completeness can be calculated by counting the ratio of the number of semantic elements in the range to be evaluated to the corresponding reference true value.

[0098] Table 4

[0099]

[0100] where N rb represents the number of semantic objects (signs, traffic lights, etc.) in the range to be evaluated, N true represents the number of semantic objects (signs, traffic lights, etc.) in the range to be evaluated.

[0101] II. Accuracy

[0102] 21) Geometric data accuracy

[0103] Geometric data accuracy can also be referred to as geometric accuracy or spatial accuracy.

[0104] In the crowd-sourced map, geometric data includes both point elements and line elements, and the calculation method of the spatial accuracy of point elements is different from that of line elements.

[0105] Referring to Table 5, the spatial accuracy of point elements can be obtained by calculating the horizontal accuracy and the vertical accuracy of point elements to obtain the comprehensive point spatial distance, and then by calculating the proportion of the comprehensive point spatial distance within a set threshold range. The horizontal accuracy of elements such as road signs is more important than the vertical accuracy.

[0106] Table 5

[0107]

[0108] wherein x represents the difference between the perpendicular distances of two points to the same lane line; y represents the other right angle side of the right triangle formed by the line connecting the two points and the horizontal distance; w1 and w2 represent the weights of the horizontal accuracy and the vertical accuracy, respectively. The first point of the two points refers to a certain feature point in the crowd-sourced map within the range to be evaluated, and the second point refers to the corresponding feature point in the prior reference data such as the true value.

[0109] The set threshold range refers to the accuracy evaluation range of point elements, such as 0.5 m, 1 m, etc. The value of the accuracy evaluation range is not absolute, and the setting basis mainly comes from the demand of intelligent driving algorithm for the crowd-sourced map, such as requiring the accuracy to be within 0.5 m or within 1 m, etc. For example, it is finally calculated that 95% of the comprehensive point spatial distances of point elements fall within the 1 m threshold range, or 80% of the comprehensive point spatial distances of point elements fall within the 0.5 m threshold range, etc., and the corresponding spatial accuracy of point elements can be obtained accordingly.

[0110] For line elements, the proportion of element data falling within a certain threshold buffer range can be used for calculation.

[0111] For the case of using high-precision map true value as reference map data, the buffer superposition method can be specifically used, that is, a buffer zone is made outside the true value based on the high-precision map element true value (for example, the buffer zone is determined by adding / subtracting a certain amount from the true value), and the ratio of the length / number of line elements falling into the buffer zone to the total length / number of line elements is calculated to obtain the spatial data accuracy (geometric accuracy) of the line elements. The specific method is shown in Table 6.

[0112] Table 6

[0113]

[0114] wherein, represents the length / number of line element data intersecting with (i.e. falling into) the buffer, and represents the total length / number of line elements.

[0115] For the case of using the high-precision map ground truth as the reference map data, the corresponding high-precision map element ground truth can be buffered based on the high-precision map element ground truth, and the spatial data precision (geometric precision) of the line element can be calculated by calculating the ratio of the length (number) of the line element falling into the buffer to the total length (number) of the line element.

[0116] 22) Attribute data accuracy

[0117] The attribute data accuracy can be calculated by calculating the proportion of the attribute-correct elements in the to-be-evaluated range, such as the first spatial sub-range, wherein the proportion of the lane line is calculated by length, and the proportion of the facility point element is calculated by number, and the formula is as follows: A i = L same / L all , wherein, L same represents the number of attribute-correct elements in the to-be-evaluated range, L all represents the number of prior elements in the to-be-evaluated range, A i represents the attribute data accuracy.

[0118] In the implementation, optionally, for the attribute data accuracy, the consistency comparison result of the crowd-sourced map element attribute and the ground truth can also be used to calculate the weighted summary, wherein the lane line color and type need high-precision ground truth as a reference, and such attributes can be given a higher weight, and other road and facility level attributes only need high-precision ground truth, and can be given a relatively low weight.

[0119] 23) Topological accuracy

[0120] The topological accuracy of the to-be-evaluated range can be obtained by traversing the element topological connection in the to-be-evaluated range on the crowd-sourced map, and calculating the proportion of the number of correct topological relationships in the total number of prior topological relationships in the to-be-evaluated range.

[0121] Three, present

[0122] Freshness can be used to represent the present.

[0123] Specifically, the latest update time of the elements in the to-be-evaluated range can be normalized to 0-1 according to the interval, the freshness of a single element is called local freshness, and the overall freshness of the to-be-evaluated range can be calculated based on the local freshness of all elements in the range. The calculation method is shown in Table 7.

[0124] Table 7

[0125]

[0126] wherein, L i denotes the length of a single element (point data value is 1), C i denotes the freshness.

[0127] An exemplary correspondence between the latest update time of an element and the element freshness is provided as follows.

[0128] Table 8

[0129] Time range of last update Freshness value 1 day 1 1-3 days 0.95 3-7 days 0.9 7-30 days 0.8 30-90 days 0.7 90-180 days 0.6 180-365 days 0.5 >365 days 0.2

[0130] Step 103, if there is a second sub-range in the target space range that does not correspond to reference map data, determining the index value of the second sub-range in the second type of quality evaluation indicator based on the multiple and / or multi-source data of the map data of the second sub-range.

[0131] The second type of quality evaluation indicator in the case of no reference map data includes, but is not limited to, spatial data accuracy and attribute data consistency.

[0132] Crowdsourcing map is derived from the fusion of multiple and / or multi-source data. In the case of no reference map data (such as high-precision / label-precision map true value data), the quality of crowdsourcing map elements depends on the aggregation degree of original data. The more aggregated the original data is, that is, the more aggregated the multiple / multi-source data is in the same center, the closer the crowdsourcing data is to the true value. Therefore, in the case of no reference map data, the quality of map data of crowdsourcing map can be evaluated by the aggregation and consistency of crowdsourcing source data itself.

[0133] The evaluation indicators include, but are not limited to, spatial data accuracy and attribute data consistency. Spatial data accuracy can be calculated by quantifying the spatial aggregation degree of original data, and attribute data consistency refers to whether the element attributes of crowdsourcing source data at the same position are consistent. The more consistent, the better the attribute quality is.

[0134] The following further provides the determination method of the index value of each second type of quality evaluation indicator.

[0135] I. Spatial data accuracy

[0136] For the second sub-range in the target space range that does not correspond to reference map data, the spatial aggregation degree of the multiple and / or multi-source data of the map data of the second sub-range can be quantified to obtain the quantification result of the spatial aggregation degree, and the spatial data accuracy of the map data of the second sub-range is determined according to the quantification result of the spatial aggregation degree.

[0137] Specifically, theoretically, the clustering center of the multiple / multi-source map source data of the crowdsourcing map is the fused crowdsourcing map data. Therefore, when determining the spatial clustering degree of the map source data in the to-be-evaluated range (such as the second sub-range), a buffer zone can be established for the fused crowdsourcing map data, if the distance between the original data does not exceed a threshold value represented by the buffer zone, it can be considered as clustering, and the data proportion of the multiple / multi-source map source data of a single element in the corresponding buffer zone can be calculated, the data proportion is used to represent the clustering degree of the multiple / multi-source map source data of the element, the higher the proportion, the more clustering, and the source data clustering degree of a single element is represented as local spatial data accuracy C i .

[0138] The buffer zone essentially refers to a distance threshold of the distance between data, wherein the distance refers to the actual spatial distance (such as meters), and the threshold of the buffer distance is determined according to the accuracy of the perception result and the requirement for precision. For example, if there is a straight arrow in the middle of a lane, theoretically, the multiple / multi-source perception results of the arrow should be clustered near the actual position of the arrow (which can be set to not more than half a lane, such as 2m).

[0139] On this basis, the spatial data accuracy of the map data in the to-be-evaluated range can be determined based on the local spatial data accuracy C i of each element in the to-be-evaluated range, so as to realize the conversion of the local spatial data accuracy to the global level. An exemplary calculation formula is as follows:

[0140]

[0141] Wherein, L represents the length (number) of the multiple / multi-source data of all elements in the to-be-evaluated region, L i represents the length (number) of the multiple / multi-source data of a single element in the to-be-evaluated region. M represents the number of elements in the to-be-evaluated region.

[0142] II. Attribute data consistency

[0143] Specifically, the attribute data consistency of the element attributes of the multiple and / or multi-source data of the map data in the to-be-evaluated range (such as the second sub-range) at the same position can be determined.

[0144] Further, for each element, a buffer zone can be set with reference to the attribute value of the fused crowdsourcing map data of the element, the proportion of the attribute quantity in the multiple and / or multi-source data of the element falling within the clustering center buffer zone is counted, the proportion is used to represent the attribute data consistency of the element, and the attribute data consistency of a single element is referred to as local attribute data consistency. On this basis, the local attribute data consistency of each element can be converted to the global level in a similar manner as the spatial data accuracy, and the overall attribute data consistency of the to-be-evaluated range is obtained.

[0145] In step 104, according to the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, the map data of the target spatial range is evaluated based on the corresponding evaluation model to obtain the quality evaluation result of the map data of the target spatial range.

[0146] In this step, specifically, if the first sub-range exists on the target spatial range and the second sub-range does not exist, based on the first index evaluation model and the index value of each index in the first type of quality evaluation index, a first sub-quality evaluation result of the map data of the first sub-range is determined as the quality evaluation result of the map data of the target spatial range.

[0147] If the second sub-range exists on the target spatial range and the first sub-range does not exist, based on the second index evaluation model and the index value of each index in the second type of quality evaluation index, a second sub-quality evaluation result of the map data of the second sub-range is determined as the quality evaluation result of the map data of the target spatial range.

[0148] If the first sub-range and the second sub-range exist on the target spatial range, based on the first index evaluation model and the index value of each index in the first type of quality evaluation index, a first sub-quality evaluation result of the map data of the first sub-range is determined; based on the second index evaluation model and the index value of each index in the second type of quality evaluation index, a second sub-quality evaluation result of the map data of the second sub-range is determined; and according to the first sub-quality evaluation result and the second sub-quality evaluation result, a quality evaluation result of the map data of the target spatial range is determined.

[0149] The first index evaluation model and the second index evaluation model are both bottom-up ladder hierarchy. The bottom layer of the ladder hierarchy of the first index evaluation model contains the first type of quality evaluation index, the top layer contains the first sub-quality evaluation result of the map data of the first sub-range, and the intermediate layer contains the intermediate dimension factor obtained by abstracting the factors contained in the lower layer of the intermediate layer.

[0150] The bottom layer of the ladder hierarchy of the second index evaluation model contains the second type of quality evaluation index, the top layer contains the second sub-quality evaluation result of the map data of the second sub-range, and the intermediate layer contains the intermediate dimension factor obtained by abstracting the factors contained in the lower layer of the intermediate layer.

[0151] In implementation, optionally, the first and second index evaluation models can be implemented by using fuzzy analytic hierarchy process. The basic idea of fuzzy analytic hierarchy process is to decompose the problem itself according to the nature of the multi-objective evaluation problem and the overall goal, and form a bottom-up ladder structure. For the present application, the multi-objective refers to the evaluation of the accuracy, completeness, and timeliness of the crowd-sourced map data, and the overall goal is the comprehensive evaluation of the quality of the crowd-sourced map data. The quality of the problem is the quality of each evaluation dimension, such as the accuracy, completeness, and timeliness of the crowd-sourced data. When the exact index of each dimension is known, the quality of each dimension can be evaluated.

[0152] Referring to Figure 2 An exemplary ladder structure of the first index evaluation model is provided, and the ladder structure of the second index evaluation model is similar to this.

[0153] According to the weight of each factor included in each layer structure of the first index evaluation model and the index value of each index in the first type of quality evaluation index, a first sub-quality evaluation result of the map data in the first sub-range is determined. Specifically, the weight of each factor included in each layer structure of the first index evaluation model is determined based on fuzzy analytic hierarchy process, and then the first sub-quality evaluation result of the map data in the first sub-range is determined according to the weight of each factor included in each layer structure of the first index evaluation model and the index value of each index in the first type of quality evaluation index.

[0154] Similarly, according to the second index evaluation model and the index value of each index in the second type of quality evaluation index, a second sub-quality evaluation result of the map data in the second sub-range is determined. Specifically, the weight of each factor included in each layer structure of the second index evaluation model is determined based on fuzzy analytic hierarchy process, and then the second sub-quality evaluation result of the map data in the second sub-range is determined according to the weight of each factor included in each layer structure of the second index evaluation model and the index value of each index in the second type of quality evaluation index.

[0155] Optionally, the process of determining the weight of each factor included in each layer structure of the corresponding index evaluation model based on fuzzy analytic hierarchy process can be further implemented as follows:

[0156] a. Construct a fuzzy judgment matrix. The fuzzy judgment matrix is used to provide a quantitative representation of the importance of one factor compared to another factor for the importance comparison between two factors;

[0157] b. based on the fuzzy judgment matrix, making pairwise comparison judgment of importance between factors in each layer of the corresponding index evaluation model to obtain a fuzzy complementary matrix; the fuzzy complementary matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model;

[0158] c. based on a predetermined conversion relationship, converting the fuzzy complementary matrix into a fuzzy consistent matrix; the fuzzy consistent matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model, and the fuzzy consistent matrix has a thinking consistency compared with the fuzzy complementary matrix;

[0159] d. based on the fuzzy consistent matrix, determining the weight of the factors in each layer of the corresponding index evaluation model; the corresponding index evaluation model is the first index evaluation model or the second index evaluation model.

[0160] The following further takes the first index evaluation model as an example to specifically describe the implementation process of determining the weight of the factors included in each layer structure of the first index evaluation model based on the fuzzy analytic hierarchy process.

[0161] In the process of determining the weight of the factors included in each layer structure of the first index evaluation model based on the fuzzy analytic hierarchy process, firstly, in the fuzzy analytic hierarchy process, pairwise comparison judgment of importance between factors is made for each layer of the first index evaluation model, and before making the pairwise comparison judgment of importance between factors, a fuzzy judgment matrix is constructed based on quantitative representation of the importance of one factor compared with another factor. Referring to Table 9, a fuzzy judgment matrix based on a 0.1-0.9 scale method and its meaning are provided.

[0162] Table 9

[0163]

[0164] Then, based on the fuzzy judgment matrix, pairwise comparison judgment of importance between factors is made for each layer of the first index evaluation model to obtain a fuzzy complementary matrix. Taking lane line data as an example, Table 10 provides an example of a fuzzy complementary matrix between some factors.

[0165] Table 10

[0166] Lane line data quality Accuracy Completeness Freshness C i ]] Accuracy 0.5 0.7 0.7 1.9 Completeness 0.3 0.5 0.6 1.4 Freshness 0.3 0.4 0.5 1.2

[0167] With reference to Table 9, in Table 10, Ci represents the sum of the ith row, and the numbers other than Ci are used to represent the importance of one factor in a factor pair compared to another factor. For example, the "0.5" in the 2nd column of the 2nd row in Table 10 indicates that "accuracy" is equally important compared to "accuracy"; the "0.7" in the 3rd column of the 2nd row indicates that "accuracy" is more important than "completeness" (reflecting "ambiguity"), and the value is added to the value in the 2nd column of the 3rd row to equal 1 (reflecting "complementarity"); the meanings of the remaining table data in Table 10 other than Ci are the same.

[0168] Thereafter, a fuzzy consistent matrix can be determined based on the fuzzy complementary matrix, and the conversion relationship between the fuzzy complementary matrix and the fuzzy consistent matrix is as follows:

[0169]

[0170] where f ij represents the value in the jth column of the ith row in Table 10, c i represents the sum of the ith row in Table 10, c ij represents the value in the jth column of the ith row in Table 11, and n represents the number of rows (columns).

[0171] The following provides an example of the fuzzy consistent matrix determined based on the fuzzy complementary matrix of Table 10.

[0172] Table 11

[0173]

[0174] The fuzzy consistent matrix is as shown in Table 11, where Si represents the geometric mean of the ith row in Table 11, and each value other than Si represents the importance of one factor (e.g., "accuracy") in a factor pair relative to another factor, which is a test of the fuzzy complementary matrix and is also an intermediate result for calculating the subsequent optimal weight.

[0175] The property of the fuzzy consistent matrix specifically refers to the consistency of thinking in determining the importance of one factor compared to another factor. For example, if "accuracy" is more important than "completeness", then the index values in the row belonging to "accuracy" in the matrix should all be greater than the index values in the row belonging to "completeness".

[0176] On this basis, the weight w i of each factor can be further determined based on the importance of one factor relative to another factor in the fuzzy consistent matrix, and the calculation formula is as follows:

[0177]

[0178] where n is the order of the matrix, represents the index weight.

[0179] The weights of the final factors (such as accuracy, integrity, and freshness) are respectively: For example, for the examples of Tables 10-11, the weights of the accuracy, integrity, and freshness can be determined by the above-mentioned manner as

[0180] Based on the above-mentioned manner, the weights of the factors included in each layer structure of the first index evaluation model can be finally determined, and on this basis, the first sub-quality evaluation result of the map data of the first sub-range can be determined according to the weights of the factors included in each layer structure of the first index evaluation model and the index values of each index in the first quality evaluation index. For example, the index values of each index in the first quality evaluation index are weighted and operated with the corresponding index (factor) weights to obtain the factor values of the upper layer factors, and then the factor values of the upper layer factors are weighted and operated with the corresponding factor weights respectively, until the top layer of the first index evaluation model is calculated, so as to obtain the first sub-quality evaluation result of the map data of the first sub-range.

[0181] In the case where the target space range only includes the first sub-range and does not include the second sub-range, the first sub-quality evaluation result of the map data of the first sub-range is also the quality evaluation result of the map data of the target space range.

[0182] The process of determining the weights of the factors included in each layer structure of the second index evaluation model based on the fuzzy analytic hierarchy process, and determining the second sub-quality evaluation result of the map data of the second sub-range according to the weights of the factors included in each layer structure of the second index evaluation model and the index values of each index in the second quality evaluation index is the same as the above-mentioned process of determining the weights of the factors included in each layer structure of the first index evaluation model and determining the first sub-quality evaluation result of the map data of the first sub-range, and will not be repeated here.

[0183] In the case where the target space range only includes the second sub-range and does not include the first sub-range, the second sub-quality evaluation result of the map data of the second sub-range is also the quality evaluation result of the map data of the target space range.

[0184] In the case where the target space range includes both the first sub-range and the second sub-range, the quality evaluation result of the map data of the target space range can be further determined according to the first sub-quality evaluation result and the second sub-quality evaluation result. For example, the quality evaluation result of the map data of the target space range can be obtained by weighted operation on the first sub-quality evaluation result and the second sub-quality evaluation result.

[0185] In an implementation, for the case of having reference map data, such as for the case of the target spatial range containing a first sub-range corresponding to the respective reference map data, if the reference map data includes both the ground truth data of the high-definition map and the ground truth data of the standard-definition map, one of them can be selected as the finally adopted reference map data as needed, or both of them can be selected as the finally adopted reference map data. If both the ground truth data of the high-definition map and the ground truth data of the standard-definition map are adopted as the reference map data, the map data of the crowd-sourced map can be evaluated in quality by performing a weighted operation on the quality evaluation results determined based on the two kinds of ground truth data.

[0186] In summary, the method provided by the embodiments of the present application fully considers different cases of having reference map data and not having reference map data, and respectively establishes corresponding map data quality evaluation indexes and evaluation models for different cases, proposes a comprehensive evaluation index and evaluation method for the quality of crowd-sourced map data, and enables the crowd-sourced data quality evaluation to adapt to different scenarios with or without reference map data. For the case of having reference map data, the index value of the first type of quality evaluation index is determined based on the corresponding reference map data, for the case of not having reference map data, the index value of the second type of quality evaluation index is determined based on multiple and / or multiple source data of the map data, and the quality of the crowd-sourced map data is evaluated based on the quality evaluation index value in the corresponding scenario in combination with the corresponding evaluation model, which enables the crowd-sourced map data quality evaluation to be quickly, accurately and effectively adapted to the actual scenario.

[0187] In an optional embodiment, the method for evaluating the quality of crowd-sourced map data provided by the present application can further include at least one of the following processes:

[0188] 1) Based on a dynamic chart, display the quality evaluation result of the map data of the target spatial range, and the factor information of the factors contained in each layer structure in the first index evaluation model and / or the second index evaluation model. The factor information includes the weight and / or factor value of the factor.

[0189] In an implementation, the reference map data of the specified target spatial range can be automatically matched in the device system, the corresponding indexes are dynamically adapted according to different actual scenarios of high-precision ground truth, standard-precision ground truth and no ground truth, the quality evaluation result of the map data of the target spatial range is calculated based on the corresponding evaluation model, and then the evaluation result is displayed in the form of a dynamic chart.

[0190] The dynamic chart can present the quality evaluation result (i.e., the total result) of the map data of the target space range, and in addition, can present the weight and / or factor value of each layer structure contained in the corresponding first index evaluation model and / or second index evaluation model, so that the user can comprehensively understand the quality of the crowdsourcing map data from the total evaluation result and the perspective of each level, each dimension factor / index.

[0191] 2) In response to obtaining the information source viewing request of the target factor on the dynamic chart, the information source of the target factor is displayed.

[0192] The user can also query the information source of the target factor by clicking the accuracy, integrity and timeliness of the target factor on the chart as needed. For example, clicking the accuracy can view that it is calculated by the spatial accuracy, attribute accuracy and topological accuracy indexes according to the corresponding weight.

[0193] 3) In response to obtaining the accuracy range adjustment request of the accuracy index, the index value of the accuracy index is updated based on the target accuracy range indicated by the accuracy range adjustment request.

[0194] The user can also dynamically select the accuracy range of the accuracy index to be evaluated. For example, assuming that the element accuracy within 0.5m or the element accuracy within 1m needs to be selected and evaluated, the corresponding accuracy range is clicked. In response to obtaining the accuracy range adjustment request of the accuracy index, the system can automatically update the index value of the accuracy index based on the target accuracy range such as 0.5m or 1m indicated by the accuracy range adjustment request. At the same time, the evaluation result of the high-level factor (such as the accuracy dimension level, the quality evaluation level of the crowdsourcing map) based on the accuracy index can also be updated.

[0195] The embodiment can display the multi-level and multi-dimensional evaluation content of the quality evaluation result of the crowdsourcing map data based on the dynamic chart, which can facilitate the user to comprehensively understand the quality of the crowdsourcing map data from the total evaluation result and the perspective of each level and each dimension factor / index, and can trace the data source and the effectiveness of the generation strategy based on this. In response to obtaining the accuracy range adjustment request of the accuracy index, the index value of the accuracy index is updated based on the target accuracy range indicated by the accuracy range adjustment request, which can meet the different needs of different users for the accuracy range, and further improves the user experience.

[0196] An application example of the method of the present application is provided below.

[0197] The example based on the method of the application realizes a web system for systematically evaluating crowd-sourced map data, mainly including data overview, data positioning, evaluation result presentation, evaluation result interaction and precision index switching, and the functions of the modules are as follows:

[0198] Data overview: used for presenting the overall data quality of the crowd-sourced map, including the total evaluation result of the crowd-sourced map data quality and the chart of the index / factor information of the evaluation system, so as to quickly and comprehensively master the evaluation situation of the current crowd-sourced map data quality.

[0199] Data positioning: used for positioning to the area where the crowd-sourced data is located through a heat map, and a range can be selected or a road can be clicked to determine the spatial range to be evaluated on the crowd-sourced map.

[0200] Evaluation result presentation: the system automatically matches the true value reference data of the selected spatial range, dynamically adapts the corresponding type of index according to the actual situation of high-precision true value, standard-precision true value or no true value, calculates the quality evaluation result, and then presents the quality evaluation result in the form of a dynamic chart. The chart can present the total evaluation result of the crowd-sourced map data quality and the weight and index / factor value of each level index / factor in the corresponding index system (index evaluation model).

[0201] Evaluation result interaction: the accuracy, completeness and timeliness systems can be clicked to query the source of each system result. For example, clicking on the accuracy can view that it is calculated by the spatial precision and attribute accuracy index according to the specified weight.

[0202] Precision index switching: the precision range of the precision index to be evaluated can be dynamically selected. For example, if the precision of 0.5m range or the precision of 1m range needs to be evaluated, the corresponding precision value is clicked to update the evaluation result.

[0203] In summary, the quality evaluation method of the crowd-sourced map data provided in the embodiments of the application fully considers the quality evaluation of the crowd-sourced map data under the conditions of high-precision map true value, standard-precision map true value and no true value, proposes a comprehensive index and evaluation method, so that the crowd-sourced data quality evaluation can adapt to different scenes. In addition, the quality of the crowd-sourced map data is evaluated from the dimensions of accuracy, completeness and timeliness. In addition to obtaining the overall quality evaluation result, the situation of a local range or a certain level index can also be dynamically obtained, so that the crowd-sourced map quality evaluation is more comprehensive, accurate and effective.

[0204] Corresponding to the above method, the embodiments of the application also provide a quality evaluation device for crowd-sourced map data, which is shown in the component structure diagram of Figure 3 The device comprises:

[0205] The first determining module 301 is configured to determine a target spatial range on a crowdsourcing map to be subjected to map data quality evaluation.

[0206] The second determining module 302 is configured to, in a case where a first sub-range corresponding to reference map data exists on the target spatial range, determine, based on the corresponding reference map data, an index value of the map data in the first sub-range in a first type of quality evaluation index.

[0207] The third determining module 303 is configured to, in a case where a second sub-range without corresponding reference map data exists on the target spatial range, determine, based on multiple times and / or multiple source data of the map data in the second sub-range, an index value of the map data in the second sub-range in a second type of quality evaluation index.

[0208] The quality evaluation module 304 is configured to, based on a corresponding evaluation model, perform quality evaluation on the map data in the target spatial range according to the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, to obtain a quality evaluation result of the map data in the target spatial range.

[0209] In an optional implementation, the first determining module 301 is specifically configured to:

[0210] Obtain indication information of a spatial range on the crowdsourcing map;

[0211] Determine, as the target spatial range, a spatial range represented by a road segment or an area indicated by the indication information on the crowdsourcing map.

[0212] In an optional implementation, the second determining module 302 is specifically configured to:

[0213] Determine, based on high-precision map ground truth data and / or high-precision map ground truth data, an index value of a different quality evaluation index corresponding to the map data in the first sub-range in a different evaluation dimension respectively;

[0214] The first type of quality evaluation index includes different quality evaluation indexes corresponding to the different evaluation dimensions respectively; the different evaluation dimensions include at least part of an integrity dimension, an accuracy dimension, and a timeliness dimension; the quality evaluation index in the integrity dimension includes at least part of geometric integrity, attribute integrity, topological integrity, and semantic object integrity; the quality evaluation index in the accuracy dimension includes at least part of geometric data precision, attribute data accuracy, and topological accuracy; and the quality evaluation index in the timeliness dimension includes freshness of the crowdsourcing map data.

[0215] In an optional implementation, the third determining module 303 is specifically configured to:

[0216] quantifying a degree of spatial aggregation of the multi-time and / or multi-source data of the second sub-range of map data, to obtain a quantification result of the degree of spatial aggregation; and determining a spatial data accuracy of the second sub-range of map data according to the quantification result of the degree of spatial aggregation;

[0217] determining a consistency of attribute data of element attributes of the multi-time and / or multi-source data of the second sub-range of map data at the same position;

[0218] The second type of quality evaluation index includes at least one of the spatial data accuracy and the attribute data consistency.

[0219] In an optional embodiment, the quality evaluation module 304 is specifically configured to:

[0220] If the first sub-range exists on the target spatial range and the second sub-range does not exist, a first sub-quality evaluation result of the map data of the first sub-range is determined as the quality evaluation result of the map data of the target spatial range based on the first index evaluation model and the index values of the first type of quality evaluation index;

[0221] If the second sub-range exists on the target spatial range and the first sub-range does not exist, a second sub-quality evaluation result of the map data of the second sub-range is determined as the quality evaluation result of the map data of the target spatial range based on the second index evaluation model and the index values of the second type of quality evaluation index;

[0222] If the first sub-range and the second sub-range exist on the target spatial range, a first sub-quality evaluation result of the map data of the first sub-range is determined based on the first index evaluation model and the index values of the first type of quality evaluation index; a second sub-quality evaluation result of the map data of the second sub-range is determined based on the second index evaluation model and the index values of the second type of quality evaluation index; and the quality evaluation result of the map data of the target spatial range is determined according to the first sub-quality evaluation result and the second sub-quality evaluation result.

[0223] In an optional embodiment, the first index evaluation model and the second index evaluation model are both bottom-up ladder structures, the bottom layer of the ladder structure of the first index evaluation model contains the first type of quality evaluation index, the top layer contains the first sub-quality evaluation result of the map data of the first sub-range, and the intermediate layer contains intermediate dimension factors obtained by abstracting the factors contained in the lower layer of the intermediate layer.

[0224] The bottom layer of the hierarchical structure of the second index evaluation model comprises second quality evaluation indexes, the top layer comprises a second sub-quality evaluation result of the map data in the second sub-range, and the middle layer comprises intermediate dimension factors obtained by abstracting the factors in the lower layer of the middle layer;

[0225] The quality evaluation module 304 is specifically used for determining a first sub-quality evaluation result of the map data in the first sub-range based on the first index evaluation model and the index values of the indexes in the first quality evaluation indexes.

[0226] The quality evaluation module 304 is specifically used for determining a first sub-quality evaluation result of the map data in the first sub-range based on the first index evaluation model and the index values of the indexes in the first quality evaluation indexes.

[0227] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0228] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0229] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0230] In an optional implementation, the quality evaluation module 304 is specifically used for determining the weights of the factors included in each layer structure of the corresponding index evaluation model based on the fuzzy analytic hierarchy process.

[0231] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0232] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0233] The quality evaluation module 304 is specifically used for determining a second sub-quality evaluation result of the map data in the second sub-range based on the second index evaluation model and the index values of the indexes in the second quality evaluation indexes.

[0234] determine weights of factors in each layer of the corresponding index evaluation model based on the fuzzy consistent matrix; the corresponding index evaluation model is the first index evaluation model or the second index evaluation model.

[0235] In an optional implementation, the apparatus further includes at least one of the following:

[0236] The first display module is configured to display, based on the dynamic chart, a quality evaluation result of map data of the target space range, and factor information of factors included in each layer structure in the first index evaluation model and / or the second index evaluation model; the factor information includes weights and / or factor values of the factors; the factors included in each layer structure in the first index evaluation model include the first type of quality evaluation index or a processing result of the first type of quality evaluation index; the factors included in each layer structure in the second index evaluation model include the second type of quality evaluation index or a processing result of the second type of quality evaluation index.

[0237] The second display module is configured to display, in response to obtaining an information source viewing request for a target factor on the dynamic chart, an information source of the target factor.

[0238] The updating module is configured to update, in response to obtaining an accuracy range adjustment request for an accuracy index, an index value of the accuracy index based on a target accuracy range indicated by the accuracy range adjustment request.

[0239] Embodiments of the present application also disclose an electronic device, and a component structure of the electronic device, as shown in Figure 4 The electronic device includes at least:

[0240] The memory 10 is configured to store a computer program.

[0241] The processor 20 is configured to realize the quality evaluation method of crowd-sourced map data by calling and executing the computer program in the memory, and the quality evaluation method is provided in any of the above method embodiments.

[0242] The processor 20 can be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a neural network processor (NPU), a deep learning processor (DPU), or other programmable logic devices.

[0243] In addition, the electronic device can further include a communication interface, a communication bus, and the like. The memory, the processor, and the communication interface perform communication with each other through the communication bus.

[0244] The communication interface is used for communication between the electronic device and other devices. The communication bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, or the like, and can be divided into an address bus, a data bus, a control bus, and the like.

[0245] In addition, the present application also provides a computer readable medium having a computer program stored thereon, the computer program comprising program codes for executing the quality evaluation method of crowd-sourced map data as disclosed in any of the above method embodiments, when executed by a processor, for implementing the quality evaluation method of crowd-sourced map data as disclosed in any of the above method embodiments.

[0246] It should be noted that each embodiment in the present specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between embodiments can be referred to each other.

[0247] For the convenience of description, the above system or device is described as various modules or units described respectively in terms of functions. Of course, in the implementation of the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0248] From the above description of the embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software and the necessary general hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments of the present application.

[0249] Finally, it needs to be pointed out that, in this article, the relationship terms such as first, second, third and target, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0250] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can also be made, which should be considered as the protection scope of the present application.

Claims

1. A method for quality evaluation of crowd-sourced map data, characterized in that, The method comprises: determining a target spatial range on a crowdsourcing map to be evaluated for map data quality; if there is a first sub-range corresponding to reference map data on the target spatial range, determining an index value of a first type of quality evaluation index of the map data in the first sub-range based on the corresponding reference map data; if there is a second sub-range without corresponding reference map data on the target spatial range, determining an index value of a second type of quality evaluation index of the map data in the second sub-range based on multiple and / or multi-source data of the map data in the second sub-range; performing quality evaluation on the map data in the target spatial range based on a corresponding evaluation model according to the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, to obtain a quality evaluation result of the map data in the target spatial range.

2. The method of claim 1, wherein, The method comprises: obtaining indication information of a spatial range on the crowdsourcing map; determining a spatial range represented by a road segment or an area indicated by the indication information on the crowdsourcing map as the target spatial range.

3. The method of claim 1, wherein, The method comprises: determining an index value of a different quality evaluation index corresponding to each of different evaluation dimensions of the map data in the first sub-range based on high-precision map ground truth data and / or medium-precision map ground truth data; wherein the first type of quality evaluation index comprises different quality evaluation indexes corresponding to different evaluation dimensions; the different evaluation dimensions comprise at least part of integrity dimension, accuracy dimension, and timeliness dimension; the quality evaluation index in the integrity dimension comprises at least part of geometric integrity, attribute integrity, topological integrity, and semantic object integrity; the quality evaluation index in the accuracy dimension comprises at least part of geometric data precision, attribute data accuracy, and topological accuracy; and the quality evaluation index in the timeliness dimension comprises freshness of the crowdsourcing map data.

4. The method of claim 1, wherein, The method comprises at least one of: quantifying a spatial aggregation degree of the multiple and / or multi-source data of the map data in the second sub-range to obtain a quantification result of the spatial aggregation degree; determining spatial data accuracy of the map data in the second sub-range according to the quantification result of the spatial aggregation degree; determining attribute data consistency of an attribute of an element at the same position of the multiple and / or multi-source data of the map data in the second sub-range; wherein the second type of quality evaluation index comprises at least one of the spatial data accuracy and the attribute data consistency.

5. The method of claim 1, wherein, The index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index are used to evaluate the quality of the map data of the target spatial range based on a corresponding evaluation model, and a quality evaluation result of the map data of the target spatial range is obtained. If the first sub-range exists on the target spatial range and the second sub-range does not exist, a first sub-quality evaluation result of the map data of the first sub-range is determined based on a first index evaluation model and the index values of the respective indexes in the first type of quality evaluation index; and the quality evaluation result of the map data of the target spatial range is the first sub-quality evaluation result. If the second sub-range exists on the target spatial range and the first sub-range does not exist, a second sub-quality evaluation result of the map data of the second sub-range is determined based on a second index evaluation model and the index values of the respective indexes in the second type of quality evaluation index; and the quality evaluation result of the map data of the target spatial range is the second sub-quality evaluation result. If the first sub-range and the second sub-range exist in the target space range, a first sub-quality evaluation result of map data of the first sub-range is determined based on the first index evaluation model and index values of each index in the first type of quality evaluation index; a second sub-quality evaluation result of map data of the second sub-range is determined based on the second index evaluation model and index values of each index in the second type of quality evaluation index; and the first sub-quality evaluation result and the second sub-quality evaluation result are combined to obtain a final quality evaluation result of the target space range. The second sub-quality evaluation result determines the quality evaluation of the map data of the target spatial range The first index evaluation model and the second index evaluation model are both bottom-up ladder structures; the bottom layer of the ladder structure of the first index evaluation model contains the first type of quality evaluation index, the top layer contains the first sub-quality evaluation result of the map data of the first sub-range, and the intermediate layer contains intermediate dimension factors obtained by abstracting the factors contained in the lower layer of the intermediate layer. Results.

6. The method of claim 5, wherein, The bottom layer of the ladder structure of the second index evaluation model contains the second type of quality evaluation index, the top layer contains the second sub-quality evaluation result of the map data of the second sub-range, and the intermediate layer contains intermediate dimension factors obtained by abstracting the factors contained in the lower layer of the intermediate layer. Based on the first index evaluation model and the index values of the respective indexes in the first type of quality evaluation index, the first sub-quality evaluation result of the map data of the first sub-range is determined, including: Based on fuzzy analytic hierarchy process, the weights of the factors contained in each layer structure of the first index evaluation model are determined; According to the weights of the factors contained in each layer structure of the first index evaluation model and the index values of the respective indexes in the first type of quality evaluation index, the first sub-quality evaluation result of the map data of the first sub-range is determined. Based on the second index evaluation model and the index values of the respective indexes in the second type of quality evaluation index, the second sub-quality evaluation result of the map data of the second sub-range is determined, including: Based on fuzzy analytic hierarchy process, the weights of the factors contained in each layer structure of the second index evaluation model are determined; According to the weights of the factors contained in each layer structure of the second index evaluation model and the index values of the respective indexes in the second type of quality evaluation index, the second sub-quality evaluation result of the map data of the second sub-range is determined. ​ 7. The method of claim 6, wherein, Determine the weight of each layer structure of the corresponding index evaluation model based on the fuzzy analytic hierarchy process, including: Construct a fuzzy judgment matrix; the fuzzy judgment matrix is used for importance comparison between two factors provides a quantitative representation of the importance of one factor compared to another factor; Based on the fuzzy judgment matrix, make pairwise comparison judgment of the importance between factors in each layer of the corresponding index evaluation model to obtain a fuzzy complementary matrix; the fuzzy complementary matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model; Based on a predetermined conversion relationship, convert the fuzzy complementary matrix into a fuzzy consistent matrix; the fuzzy consistent matrix is used to represent the importance of one factor compared with another factor in the corresponding layer of the corresponding index evaluation model, and the fuzzy consistent matrix has a thinking consistency compared with the fuzzy complementary matrix; Based on the fuzzy consistent matrix, determine the weight of the factors in each layer of the corresponding index evaluation model; the corresponding index evaluation model is the first index evaluation model or the second index evaluation model. 8.The method of claim 6 or 7, wherein, Further include at least one of the following: Based on a dynamic chart, display the quality evaluation result of the map data of the target space range, and the factor information of the factors contained in each layer structure of the first index evaluation model and / or the second index evaluation model; the factor information includes the weight and / or the factor value of the factors; the factors contained in each layer structure of the first index evaluation model include the first type of quality evaluation index or the processing result of the first type of quality evaluation index; the factors contained in each layer structure of the second index evaluation model include the second type of quality evaluation index or the processing result of the second type of quality evaluation index; In response to obtaining a request for viewing the information source of a target factor on the dynamic chart, display the information source of the target factor; In response to obtaining a request for adjusting the accuracy range of an accuracy index, update the index value of the accuracy index based on the target accuracy range indicated by the request for adjusting the accuracy range. 9.A quality evaluation device of crowd-sourced map data, characterized by comprising: Include: A first determination module is configured to determine a target space range on a crowdsourcing map for which map data quality evaluation is to be performed; A second determination module is configured to, in a case where a first sub-range corresponding to reference map data exists on the target space range, determine the index value of the first sub-range in the first type of quality evaluation index based on the corresponding reference map data; A third determination module is configured to, in a case where a second sub-range not corresponding to reference map data exists on the target space range, determine the index value of the second sub-range in the second type of quality evaluation index based on the corresponding reference map data In the second sub-range case, the index value of the second sub-range map data in the second type of quality evaluation index is determined based on the multiple and / or multi-source data of the second sub-range map data. A quality evaluation module is configured to perform quality evaluation on the map data of the target space range based on the corresponding evaluation model according to the index value of the first type of quality evaluation index and / or the index value of the second type of quality evaluation index, and obtain a quality evaluation result of the map data of the target space range.

10. An electronic device, comprising: Include: A memory is configured to store a computer program; A processor is configured to realize the quality evaluation method of the crowdsourcing map data by calling and executing the computer program in the memory, as claimed in any one of claims 1-8.

Images