Lane map data fusion method and device

Abstract

The invention discloses a lane map data fusion method and device. The method comprises the following steps: acquiring an initial lane model and lane information thereof; the lane information of the initial lane model comprises the number of lanes and edge line information; obtaining a plurality of lane models according to the obtained observation data, the initial lane model, the number of lanes and the edge line information; based on each lane model and a plurality of pieces of lane section data and a plurality of pieces of observation section data obtained by cutting the observation data, updating the lane model and determining the observation confidence of each lane demarcation point and the real confidence of each lane edge point in the lane model; and based on the observation confidence of each lane demarcation point in each lane model and the real confidence of each lane edge point, calculating to obtain the overall confidence of each lane model, and obtaining the lane model with the maximum overall confidence. Compared with a clustering algorithm, the scheme has higher stability and generalization ability, and a more accurate and stable lane model is obtained by accumulating confidence scores.

Description

Technical Field

[0001] This invention relates to a method and apparatus for fusion of lane map data. Background Technology

[0002] Against the backdrop of the rapid development of modern geographic information systems and mobile internet technology, crowdsourced mapping, as a new model of map construction and updating, is gradually becoming a key approach to solving the problem of lagging traditional map updates. With the widespread adoption of various smart devices and advancements in sensor technology, massive numbers of users can provide real-time data on various types of information, such as geographic location, traffic conditions, and landmark changes. This crowdsourced map data is characterized by diversity and heterogeneity, originating from multiple devices and applications, and its timeliness and accuracy depend on the real-world experience and immediate feedback of users.

[0003] Among these, research on multi-lane map data fusion technology is particularly important. It involves a series of complex algorithms and methods, including data preprocessing, matching and alignment, consistency detection and conflict resolution, quality assessment, and weight allocation. Existing technologies often use clustering algorithms to discover and distinguish the characteristics of similar road segments. The development of these technologies will not only significantly improve the timeliness and completeness of map data but also powerfully promote the development of smart cities and intelligent transportation, playing a significant role in improving public management efficiency, optimizing resource allocation, and enhancing the quality of life for the public. Summary of the Invention

[0004] To obtain a more stable and accurate fused lane model, embodiments of the present invention provide a lane map data fusion method and apparatus.

[0005] In a first aspect, embodiments of the present invention provide a lane map data fusion method, the method comprising:

[0006] Obtain an initial lane model and its lane information; the lane information of the initial lane model includes the number of lanes and edge line information;

[0007] Based on the acquired observation data, the initial lane model, the number of lanes, and the edge line information, multiple lane models are obtained;

[0008] Based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data, the lane model is updated and the observation confidence and the true confidence of each lane boundary point and each lane edge point in the lane model are determined.

[0009] Based on the observed confidence level of each lane boundary point and the true confidence level of each lane edge point in each lane model, the overall confidence level of each lane model is calculated, and the lane model with the highest overall confidence level is obtained.

[0010] In one or more optional embodiments of this application, the step of updating the lane model and determining the observation confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data includes:

[0011] Each lane model and each observation data is divided into multiple lane cross-section data and multiple observation cross-section data, and the initial sub-configuration confidence of each lane boundary point in the multiple lane cross-section data is set.

[0012] For each lane model, for each observation section of each observation data, based on the observation section data and the initial sub-configuration confidence of each lane boundary point, the first maximum observation confidence, the first sub-configuration confidence, and the first true confidence of each observation boundary point are calculated, and the first true confidence and the first parameter confidence of each observation edge point are calculated.

[0013] Based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point, and the first true confidence and first parameter confidence of each observation edge point calculated, the lane model is updated and the observation confidence of each lane boundary point and the true confidence of each lane edge point in the lane model are determined.

[0014] In one or more optional embodiments of this application, for each lane model, for each observation cross-section data of each observation data, based on the observation cross-section data and the initial sub-configuration confidence of each lane boundary point, the first maximum observation confidence, the first sub-configuration confidence, and the first true confidence are calculated, and the first true confidence and the first parameter confidence are calculated for each observation edge point, including:

[0015] Based on the lane cross-section data of each lane in the lane model and the observation cross-section data at the corresponding location, the observation confidence level of each observation boundary point in the observation cross-section data corresponding to each lane boundary point in the lane cross-section data is calculated.

[0016] The maximum value among the multiple observation confidence scores corresponding to each observation boundary point is taken as the first maximum observation confidence score of the observation boundary point;

[0017] For each observation boundary point, the multiple observation confidence scores corresponding to the observation boundary point are multiplied by the initial allocation confidence scores of the lane boundary points corresponding to the observation boundary point to obtain multiple probability factors corresponding to the observation boundary point.

[0018] The maximum value of the normalized multiple probability factors corresponding to the observation boundary point is taken as the first sub-configuration confidence of the observation boundary point.

[0019] Based on the observation cross-sectional data and the first maximum observation confidence of the observation boundary point, the first true confidence of the observation boundary point is calculated.

[0020] Based on the observation cross-sectional data, the first true confidence score and the first parameter confidence score of each observation edge point in the observation cross-sectional data are calculated.

[0021] In one or more optional embodiments of this application, the observation cross-section data further includes trajectory points and the initial true confidence level of each observation edge point;

[0022] The calculation of the first true confidence score and the first parameter confidence score for each observation edge point in the observation cross-section data based on the observation cross-section data includes:

[0023] For each observation edge point, calculate the confidence factor of the observation edge point;

[0024] The maximum value between the confidence factor of the observed edge point and the initial true confidence is taken as the first true confidence of the observed edge point;

[0025] For each observation edge point, calculate the sum of the distances from every two adjacent observation boundary points in the observation cross-sectional data where the observation edge point is located to the trajectory point, and obtain the coordinates of the two adjacent observation boundary points with the smallest sum of distances;

[0026] Calculate the mean of the coordinates of the two adjacent observation boundary points with the smallest sum of distances to obtain the matching lane center point corresponding to the observation edge point;

[0027] The first parameter confidence level of the observed edge point is calculated based on the matching lane center point corresponding to the observed edge point.

[0028] In one or more optional embodiments of this application, the lane model further includes lane width; the observation edge points include observation left edge points and observation right edge points;

[0029] The initial true confidence level of each observed edge point is obtained in the following manner:

[0030] The initial true confidence level of each observed left edge point is calculated based on the coordinates of each observed left edge point, the lane width, and the number of lanes.

[0031] The initial true confidence level of each observed right edge point is calculated based on the coordinates of each observed right edge point, the lane width, and the number of lanes.

[0032] In one or more optional embodiments of this application, the step of updating the lane model and determining the observation confidence of each lane boundary point and the true confidence of each lane edge point in the lane model based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point, and the first true confidence and first parameter confidence of each observation edge point, includes:

[0033] Based on the first maximum observation confidence, first subconfiguration confidence, and first true confidence of each observation boundary point of all the aforementioned observation cross-sectional data, determine the maximum observation confidence, subconfiguration confidence, and true confidence of each observation boundary point respectively;

[0034] Based on the first true confidence and the first parameter confidence of each observation edge point of all the observation cross-sectional data, the true confidence and parameter confidence of each observation edge point are determined respectively;

[0035] The lane model is updated based on the configuration confidence and true confidence of each observation boundary point, and the true confidence and parameter confidence of each observation edge point, and the true confidence of each lane edge point in the lane model is obtained.

[0036] The observation confidence level of each lane boundary point in the lane model is determined based on the first maximum observation confidence level of each observation boundary point of all the observation cross-sectional data.

[0037] In one or some optional embodiments of this application, after each lane model and each observation data is cut into multiple lane cross-section data and multiple observation cross-section data respectively, the method further includes setting the initial true confidence of each lane edge point and the initial true confidence of each lane boundary point in the multiple lane cross-section data.

[0038] The process of updating the lane model based on the configuration confidence and true confidence of each observation boundary point, and the true confidence and parameter confidence of each observation edge point, and obtaining the true confidence of each lane edge point in the lane model, includes:

[0039] The updated confidence level of each observation boundary point is obtained by multiplying the distributed confidence level of each observation boundary point with the true confidence level.

[0040] The updated confidence of each observed edge point is obtained by multiplying the true confidence of each observed edge point by the parameter confidence.

[0041] If the updated confidence of each observation boundary point or observation edge point is greater than the initial true confidence of the observation boundary point or observation edge point, then the lane boundary point or lane edge point in the lane model is updated to the observation boundary point or observation edge point, and the true confidence of the lane boundary point or lane edge point is updated to the updated confidence of the observation boundary point or observation edge point.

[0042] In one or more optional embodiments of this application, obtaining multiple lane models based on the acquired multiple observation data, the initial lane model, the number of lanes, and the edge line information includes:

[0043] The initial lane model is used as a lane model;

[0044] The coordinates of the leftmost and rightmost observation boundaries are calculated based on the multiple observation data.

[0045] The coordinates of the left and right observation edges are calculated based on the multiple observation data and the coordinates of the leftmost and rightmost observation boundaries.

[0046] The total road width, the number of estimated lanes, and the probability of each estimated lane number are calculated based on the coordinates of the observed left edge line and the observed right edge line.

[0047] When the probability that the estimated number of lanes equals the number of lanes plus one is greater than a preset threshold, determine whether the left edge line exists in the edge line information:

[0048] If not, add a new dividing line and a new edge line to the left of the initial lane model to obtain a lane model;

[0049] When the probability that the estimated number of lanes equals the number of lanes plus one is greater than the preset threshold, determine whether the right edge line exists in the edge line information:

[0050] If not, add a new dividing line and a new edge line to the right of the initial lane model to obtain a new lane model.

[0051] In one or more optional embodiments of this application, the lane information further includes the total number of dividing lines;

[0052] The initial true confidence level of each lane edge point, and the initial true confidence level and initial partial confidence level of each lane boundary point in the multiple lane cross-section data are set in the following manner:

[0053] The initial true confidence of lane boundary points and lane edge points on newly added boundary lines and newly added edge lines in all the lane cross-section data is set to the preset new true confidence value, and the initial true confidence of other lane boundary points and lane edge points in the lane cross-section data is set to the preset initial true confidence value.

[0054] For each lane cross-section model, determine whether there is a lane dividing point on the newly added dividing line in the lane cross-section model:

[0055] If so, the initial configuration confidence of all lane dividing points in the lane cross-section model is set to the reciprocal of the total number of dividing lines plus one.

[0056] If not, then set the initial configuration confidence of all lane dividing points in the lane cross-section model to the reciprocal of the total number of dividing lines.

[0057] In a second aspect, embodiments of the present invention provide a lane map data fusion device, the device comprising:

[0058] The first acquisition module is used to acquire an initial lane model and its lane information; the lane information of the initial lane model includes the number of lanes and edge line information.

[0059] The second acquisition module is used to obtain multiple lane models based on the acquired multiple observation data, the initial lane model, the number of lanes, and the edge line information;

[0060] The first calculation module is used to update the lane model and determine the observation confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data.

[0061] The second calculation module is used to calculate the overall confidence of each lane model based on the observed confidence of each lane boundary point and the true confidence of each lane edge point in each lane model, and to obtain the lane model with the highest overall confidence.

[0062] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the lane map data fusion method as described above.

[0063] Fourthly, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the lane map data fusion method as described above.

[0064] Fifthly, embodiments of the present invention provide a computer program product containing instructions that, when the computer program product is run on a computer device, cause the computer device to execute the lane map data fusion method described above.

[0065] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:

[0066] The lane map data fusion method provided in this invention obtains multiple lane models based on the acquired initial lane model, lane information, and multiple observation data. Then, each lane model and observation data is segmented into multiple lane cross-sectional data and multiple observation cross-sectional data. The lane model is updated based on the lane cross-sectional data and observation cross-sectional data. The observed confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model are determined, and the overall confidence level of each lane model is calculated. The lane model with the highest overall confidence level is selected as the fused lane model. By segmenting the complex lane model into multiple continuous cross-sectional data, and thus meticulously analyzing and quantifying each cross-sectional data, this method exhibits stronger stability and generalization ability compared to clustering algorithms when facing data noise, local missing data, or abnormal fluctuations. Finally, by accumulating the confidence scores, a more accurate and stable lane model is obtained.

[0067] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0068] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. Attached Figure Description

[0069] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0070] Figure 1 This is a schematic diagram illustrating the steps of lane map data fusion provided in an embodiment of the present invention;

[0071] Figure 2 This is a schematic diagram illustrating the steps of updating the lane model provided in an embodiment of the present invention;

[0072] Figure 3 This is a schematic diagram illustrating the steps for calculating multiple confidence levels in observation cross-sectional data provided in an embodiment of the present invention;

[0073] Figure 4 A schematic diagram illustrating an example of a lane model and observation section data provided in an embodiment of the present invention;

[0074] Figure 5 This is a schematic diagram of the lane map data fusion device provided in an embodiment of this application. Detailed Implementation

[0075] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0076] The inventors discovered that in existing technologies, map fusion algorithms often directly use clustering algorithms or determine a set of high-precision map feature data as a baseline lane model. However, the quality of the map feature data obtained in practice is difficult to control and is prone to dispersion, leading to unstable clustering results. Consequently, the quality of the final generated map data is unstable and its accuracy is reduced. Based on this, the inventors conducted further research and development, resulting in this invention, which provides a lane map data fusion method and apparatus.

[0077] Example 1

[0078] This invention provides a lane map data fusion method, referring to... Figure 1 As shown, the method includes:

[0079] S101: Obtain the initial lane model and its lane information; the lane information of the initial lane model includes the number of lanes and edge line information;

[0080] S102: Based on the acquired observation data, initial lane model, number of lanes, and edge line information, multiple lane models are obtained;

[0081] S103: Based on the multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data, update the lane model and determine the observation confidence of each lane boundary point and the true confidence of each lane edge point in the lane model.

[0082] S104: Based on the observed confidence level of each lane boundary point and the true confidence level of each lane edge point in each lane model, the overall confidence level of each lane model is calculated, and the lane model with the highest overall confidence level is obtained.

[0083] The lane map data fusion method provided in this invention obtains multiple lane models based on the acquired initial lane model, lane information, and multiple observation data. Each lane model and observation data is then segmented into multiple lane cross-sectional data and multiple observation cross-sectional data. The lane model is updated based on the lane cross-sectional data and observation cross-sectional data. The observed confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model are determined, and the overall confidence level of each lane model is calculated. The lane model with the highest overall confidence level is selected as the fused lane model. By segmenting the complex lane model into multiple continuous cross-sectional data, and thus meticulously analyzing and quantifying each cross-sectional data, this method exhibits stronger stability and generalization ability compared to clustering algorithms when facing data noise, local missing data, or abnormal fluctuations. Finally, by accumulating the confidence scores, a more accurate and stable lane model is obtained.

[0084] The lane map data fusion method provided in this invention offers a new solution for the dynamic updating and maintenance of high-precision maps in future intelligent transportation systems. It has profound significance and wide application value for the precise positioning of autonomous vehicles, path planning, and even the construction and development of smart cities.

[0085] In an optional embodiment, obtaining the initial lane model and its lane information in step S101 specifically includes the following steps S1011 to S1012:

[0086] S1011: Obtain the initial lane model.

[0087] Specifically, the initial lane model can be obtained in the following two ways.

[0088] The first method for obtaining the initial lane model is as follows: Obtain map feature data, and based on the edge lines and boundary lines in the map feature data, divide the road into multiple lanes according to the relative positional relationship of the boundary lines. Calculate the number of lanes and the width of each lane, and then adjust the calculated lane widths in conjunction with the route design specifications. Combined with the map feature data, an initial lane model is constructed.

[0089] The second method for obtaining the initial lane model is to acquire map feature data with multiple overlapping trajectory points, use a clustering algorithm to cluster the points on the boundary lines and edge lines in the multiple map feature data, determine the distribution of the boundary lines and edge lines, and obtain the initial lane model.

[0090] Those skilled in the art can implement the above two methods for obtaining the initial lane model based on the detailed description of the prior art, and no specific limitations are made in the embodiments of this application.

[0091] S1012: Obtain lane information based on the initial lane model.

[0092] The lane information mentioned above can include the number of lanes, the total number of dividing lines, and edge line information. Among them, the edge line information includes the existence information of the left and right edge lines, that is, whether the left and right edge lines exist in the initial lane model.

[0093] In an optional embodiment, step S102 above, based on the acquired multiple observation data, the initial lane model, the number of lanes, and edge line information, obtains multiple lane models, specifically including the following steps S1021 to S1028:

[0094] S1021: Treat the initial lane model as a lane model.

[0095] S1022: The coordinates of the leftmost and rightmost observation boundaries are calculated based on multiple observation data.

[0096] Specifically, the leftmost and rightmost dividing lines observed in multiple observation data can be used as the leftmost observation dividing line and the rightmost observation dividing line.

[0097] S1023: The coordinates of the left and right edge lines of the observation are calculated based on multiple observation data, the coordinates of the leftmost and rightmost observation boundary lines.

[0098] Specifically, this can be achieved by comparing the coordinates of the leftmost observation boundary line, obtained by moving it to the left by a preset interval, with the coordinates of the leftmost edge line among multiple observation data sets, and using the leftmost coordinate as the observation left edge line. Similarly, the coordinates of the rightmost observation boundary line, obtained by moving it to the right by a preset interval, are compared with the coordinates of the rightmost edge line among multiple observation data sets, and the rightmost coordinate as the observation right edge line.

[0099] For example, if we set the coordinate values ​​from smallest to largest to represent the position of the edge line from left to right, then the operation of step S1023 above can be expressed as:

[0100] Observation left edge line = Min(leftmost observation boundary line - preset interval distance, left edge line);

[0101] Observation right edge line = Max(rightmost observation boundary line + preset interval distance, right edge line);

[0102] The left and right edge lines represent the coordinates of the leftmost and rightmost edge lines in the multiple observation data mentioned above. The preset interval distance can be set to 0.2 meters (m).

[0103] S1024: Calculate the total road width, multiple estimated lane numbers, and the probability of each estimated lane number based on the coordinates of the observed left and right edge lines.

[0104] Specifically, since the left and right edge lines are parallel line segments, the distance between the two points can be calculated based on the coordinates of the endpoints on the same side of these two line segments, thus obtaining the distance between the left and right edge lines, which is the total width of the road.

[0105] Based on the total road width and the width range of a single lane in the route design specifications, multiple estimated lane numbers are calculated.

[0106] By combining analysis and statistics of historical data, prior knowledge—the distribution of road lanes within the area where the initial lane model is located—is obtained, thereby estimating the probability of each estimated lane number. Then, the probabilities of all estimated lane numbers are normalized so that the sum of the probabilities of all estimated lane numbers equals 1.

[0107] For example, assuming the calculated total road width is 10m, and the width of a single lane in the route design specifications ranges from 3m to 4m, the estimated number of lanes can be calculated as 10m / 3m to 10m / 4m. This calculation rounds down, resulting in an estimated number of lanes of 2 or 3. Then, based on prior knowledge, the probability of each estimated number of lanes is estimated and normalized, yielding the probability of 2 lanes P(2|10) = 0.8 and the probability of 3 lanes P(3|10) = 0.2.

[0108] S1025: When the probability that the estimated number of lanes equals the number of lanes plus one is greater than the preset threshold, determine whether the left edge line exists in the edge line information: if yes, proceed to step S1027; if no, proceed to step S1026.

[0109] S1026: Add a new dividing line and a new edge line to the left of the initial lane model to obtain a lane model.

[0110] Specifically, according to the route design specifications, a new dividing line and a new edge line can be added at an appropriate position on the left side of the initial lane model. For example, a new dividing line can be added 3m to the left of the leftmost dividing line of the initial lane model, and a new edge line can be added 0.2m to the left of the new dividing line to obtain a lane model.

[0111] S1027: When the probability that the estimated number of lanes equals the number of lanes plus one is greater than the preset threshold, determine whether the right edge line exists in the edge line information: if yes, proceed to step S103; if no, proceed to step S1028.

[0112] S1028: Add a new dividing line and a new edge line to the right side of the initial lane model to obtain a lane model.

[0113] In an optional embodiment, in step S103 above, based on the multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data, the lane model is updated and the observation confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model are determined, with reference to... Figure 2 As shown, the specific steps include S1031 to S1033:

[0114] S1031: Divide each lane model and each observation data into multiple lane cross-section data and multiple observation cross-section data respectively, and set the initial sub-configuration confidence of each lane boundary point in the multiple lane cross-section data.

[0115] Specifically, S1031 may include the following steps S10311 to S10312:

[0116] S10311: Determine the cutting interval based on the sampling accuracy of the observation data, and cut each lane model and each observation data into multiple lane cross-section data and multiple observation cross-section data according to the cutting interval.

[0117] In the embodiments of this application, the actual lane model or observation data often involves complex geometry. Those skilled in the art can determine the cutting spacing according to the actual needs and the detailed description of the prior art to realize the cutting of the lane model and observation data, so that there is a one-to-one correspondence between multiple lane cross-section data and multiple observation cross-section data.

[0118] S10312: Set the initial confidence level of each lane boundary point in multiple lane cross-section data, and set the initial true confidence level of each lane edge point and lane boundary point.

[0119] Furthermore, S10312 may specifically include the following steps S103121 to S103124:

[0120] S103121: Set the initial true confidence of lane boundary points and lane edge points on newly added boundary lines and newly added edge lines in all lane cross-section data to the preset newly added true confidence value, and set the initial true confidence of other lane boundary points and lane edge points in lane cross-section data to the preset initial true confidence value.

[0121] For example, the initial true confidence of lane dividing points and lane edge points on newly added dividing lines and newly added edge lines is set to 0.1, and the initial true confidence of other lane dividing points and lane edge points is set to 0.7.

[0122] S103122: For each lane cross-section model, determine whether there is a lane dividing point on the newly added dividing line in the lane cross-section model: if yes, proceed to step S103123; if no, proceed to step S103124.

[0123] S103123: Set the initial subconfiguration confidence of all lane boundary points in the lane cross-section model to the reciprocal of the number of lanes plus one.

[0124] S103124: Set the initial sub-configuration confidence of all lane boundary points in the lane cross-section model to the reciprocal of the number of lanes.

[0125] Specifically, if a lane boundary point exists on a newly added boundary line in a lane cross-section model, then the initial allocation confidence of all lane boundary points in the lane cross-section model is 1 / (total number of boundary lines + 1). If a lane boundary point does not exist on a newly added boundary line in a lane cross-section model, then the initial allocation confidence of all lane boundary points in the lane cross-section model is 1 / total number of boundary lines.

[0126] For example, if the total number of boundary lines is 3, then the initial sub-configuration confidence of lane boundary points in lane models with newly added boundary lines is 1 / 4, and the initial sub-configuration confidence of lane boundary points in lane models with newly added boundary lines at the edges is 1 / 3.

[0127] S1032: For each lane model, for each observation section of each observation data, based on the observation section data and the initial sub-configuration confidence of each lane boundary point, calculate the first maximum observation confidence, the first sub-configuration confidence, and the first true confidence of each observation boundary point, and calculate the first true confidence and the first parameter confidence of each observation edge point.

[0128] Specifically, refer to Figure 3 As shown, S1032 specifically includes the following steps S10321 to S10326:

[0129] S10321: Based on the lane cross-section data of each lane model and the observation cross-section data at the corresponding location, calculate the observation confidence of each observation boundary point in the observation cross-section data corresponding to each lane boundary point in the lane cross-section data.

[0130] Specifically, the observation confidence level of each observation boundary point in the observation cross-section data corresponding to each lane boundary point in the lane cross-section data can be calculated using the following formula:

[0131]

[0132] Where P(E_j|H_i) represents the observation confidence level of the lane boundary point corresponding to the observation boundary point j, N represents the total number of boundary lines, and ε is the preset observation bias, which can be set to 0.1.

[0133] For example, refer to Figure 4As shown, the lane model contains three dividing lines A1 to A3, trajectory points, and seven observation dividing points E_1 to E_7 in the observation cross-section data. Since the lane cross-section data corresponding to this observation cross-section data contains three lane dividing points, three observation confidence scores can be calculated for each observation dividing point. For example, the first observation dividing point E_1 can be calculated to have three observation confidence scores: P(E_1|H_1), P(E_1|H_2), and P(E_1|H_3). Therefore, a total of 21 observation confidence scores can be calculated for this observation cross-section data.

[0134] S10322: Take the maximum value among the multiple observation confidence scores corresponding to each observation boundary point as the first maximum observation confidence score of the observation boundary point.

[0135] Specifically, the maximum value among the multiple observation confidence scores corresponding to each observation boundary point obtained in step S10321 can be used as the first maximum observation confidence score of that observation boundary point.

[0136] For example, if the three observation confidence scores P(E_1|H_1), P(E_1|H_2), and P(E_1|H_3) corresponding to the observation boundary point E_1 are 0.98534194, 0.009755861, and 0.004902199 respectively, then the maximum value P(E_1|H_1) is taken as the first maximum observation confidence score of the observation boundary point E_1.

[0137] S10323: For each observation boundary point, multiply the multiple observation confidence scores corresponding to the observation boundary point with the initial allocation confidence scores of the lane boundary points corresponding to the observation boundary point to obtain multiple probability factors corresponding to the observation boundary point.

[0138] Specifically, it can be done by multiplying the multiple observation confidence levels corresponding to each observation boundary point obtained in step S10321 with the initial sub-configuration confidence level of the corresponding lane boundary point, based on the following formula, to obtain multiple probability factors corresponding to the observation boundary points:

[0139] PF(H_i|E_j)=P(E_j|H_i)×P(H_i);

[0140] Wherein, PF(H_i|E_j) represents the probability factor of lane boundary point H_i corresponding to observation boundary point E_j, P(E_j|H_i) represents the observation confidence of lane boundary point corresponding to observation boundary point j, and P(H_i) represents the initial allocation confidence of lane boundary point H_i.

[0141] For example, the three observation confidence scores P(E_1|H_1), P(E_1|H_2), and P(E_1|H_3) corresponding to the observation boundary point E_1 are multiplied by the initial allocation confidence scores of lane boundary points H_1, H_2, and H_3, respectively, to obtain the three probability factors PF(E_1|H_1), PF(E_1|H_2), and P(E_1|H_3) corresponding to the observation boundary point E_1. F (E_1|H_3).

[0142] S10324: The maximum value of the normalized multiple probability factors corresponding to the observation boundary point is taken as the first sub-configuration confidence of the observation boundary point.

[0143] Specifically, it can be done by normalizing the multiple probability factors corresponding to each observation boundary point obtained in step S10323 above based on the following formula, and taking the maximum value among them as the first subconfiguration confidence of the observation boundary point:

[0144]

[0145] Where P(H_i|E_j) is the first subconfiguration confidence of the observation boundary point E_j, P F (H_i|E_j) represents the probability factor of the lane boundary point H_i corresponding to the observed boundary point E_j.

[0146] S10325: Based on the observation cross-section data and the first maximum observation confidence of the observation boundary point, the first true confidence of the observation boundary point is calculated.

[0147] Specifically, it can be done by first calculating the Gaussian kernel function of the observation boundary point based on the trajectory points, using the following formula:

[0148] P(E_j|trajectory point) = exp(-(distance between observation boundary point j and trajectory point)) 2 / 2σ 2 );

[0149] Where P(E_j|trajectory point) is the Gaussian kernel function between the observation boundary point j and the trajectory point, and σ is a preset adjustment coefficient. The trajectory point can be referenced. Figure 4 As shown.

[0150] Then, the first maximum observation confidence of the observation boundary point obtained in step S10322 above is multiplied by the Gaussian kernel function to obtain the first true confidence of the observation boundary point.

[0151] S10326: Based on the observation cross-section data, calculate the first true confidence and the first parameter confidence of each observation edge point in the observation cross-section data.

[0152] Specifically, S10326 may include the following steps S103261 to S103265:

[0153] S103261: For each observation edge point, calculate the confidence factor of the observation edge point.

[0154] For each observation edge point, the confidence factor is calculated using the following formula:

[0155] P(E_b) = 1 / (distance between observation edge point b and trajectory point + ε);

[0156] Where P(E_b) represents the confidence factor of the observed edge point b, and ε is the preset observation bias, which can be set to 0.1.

[0157] S103262: The maximum value between the confidence factor of the observed edge point and the initial true confidence is taken as the first true confidence of the observed edge point.

[0158] The observed edge points include the observed left edge point and the observed right edge point. The initial true confidence of the observed edge points is obtained in the following way.

[0159] The initial true confidence level for each observed left edge point is calculated based on its coordinates, lane width, and number of lanes. Taking the coordinate values ​​from smallest to largest, representing the edge point's position from left to right, as an example, the expression for calculating the initial true confidence level for each observed left edge point is:

[0160] P(observation of left edge point) = 1 / (Min(coordinates of observation of left edge point, lane width * number of lanes) + δ);

[0161] Where P (observation of the left edge point) represents the initial true confidence level of the observation of the left edge point, and δ is a preset adjustment parameter.

[0162] The initial true confidence level for each observed right edge point is calculated based on its coordinates, lane width, and number of lanes. Taking the coordinate values ​​from smallest to largest, representing the edge point's position from left to right, as an example, the expression for calculating the initial true confidence level for each observed right edge point is:

[0163] P(observation of the right edge point) = 1 / (Min(coordinates of the observation of the right edge point, lane width * number of lanes) + δ);

[0164] Where P (observation right edge point) represents the initial true confidence level of the observation right edge point, and δ is a preset adjustment parameter.

[0165] S103263: For each observation edge point, calculate the sum of the distances from every two adjacent observation boundary points to the trajectory point in the observation cross section data where the observation edge point is located, and obtain the coordinates of the two adjacent observation boundary points with the smallest sum of distances.

[0166] Specifically, for each observation edge point, the sum of the distances from every two adjacent observation boundary points in the observation cross-section data to the trajectory point can be calculated, and the lane between the two observation boundary points with the smallest sum of distances can be determined as the matching lane, thus obtaining the coordinates of the two adjacent observation boundary points.

[0167] S103264: Calculate the mean of the coordinates of the two adjacent observation boundary points with the smallest sum of distances to obtain the matching lane center point corresponding to the observation edge point.

[0168] S103265: Calculate the first parameter confidence level of the observed edge point based on the center point of the matching lane corresponding to the observed edge point.

[0169] Specifically, the confidence level of the first parameter of the observed edge point can be calculated based on the center point of the matched lane using the following formula:

[0170] P p (E_b) = 1 / (distance between observation edge point b and matching lane center point + ε);

[0171] Among them, P p (E_b) represents the confidence level of the first parameter of the observed edge point, and ε is the preset observation bias.

[0172] S1033: Based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point, and the first true confidence and first parameter confidence of each observation edge point calculated, update the lane model and determine the observation confidence of each lane boundary point and the true confidence of each lane edge point in the lane model. Specifically, this includes the following steps S10331 to S10334.

[0173] S10331: Based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point of all observation cross-sectional data, determine the maximum observation confidence, sub-configuration confidence, and true confidence of each observation boundary point respectively.

[0174] Specifically, the maximum value among the first maximum observation confidence scores of each observation boundary point of all observation cross-section data can be used as the maximum observation confidence score of that observation boundary point. Similarly, the maximum value among the first subconfiguration confidence scores of each observation boundary point of all observation cross-section data can be used as the subconfiguration confidence score of that observation boundary point, and the maximum value among the first true confidence scores of each observation boundary point of all observation cross-section data can be used as the true confidence score of that observation boundary point.

[0175] S10332: Based on the first true confidence and the first parameter confidence of each observation edge point in all observation cross-sectional data, determine the true confidence and parameter confidence of each observation edge point respectively.

[0176] Specifically, the maximum value of the first true confidence score of each observation edge point in all observation cross-sectional data can be used as the true confidence score of that observation edge point. Similarly, the maximum value of the first parameter confidence score of each observation edge point in all observation cross-sectional data can be used as the parameter confidence score of that observation edge point.

[0177] S10333: Update the lane model based on the configuration confidence and true confidence of each observation boundary point, the true confidence and parameter confidence of each observation edge point, and obtain the true confidence of each lane edge point in the lane model.

[0178] Specifically, it can be done by first multiplying the confidence level of each observation boundary point by the true confidence level to obtain the updated confidence level of each observation boundary point;

[0179] Then, the true confidence of each observed edge point is multiplied by the parameter confidence to obtain the updated confidence of each observed edge point;

[0180] Finally, if the updated confidence of each observation boundary point or observation edge point is greater than the initial true confidence of the observation boundary point or observation edge point, then the lane boundary point or lane edge point in the lane model is updated to the observation boundary point or observation edge point, and the true confidence of the lane boundary point or lane edge point is updated to the updated confidence of the observation boundary point or observation edge point.

[0181] S10334: Determine the observation confidence of each lane boundary point in the lane model based on the maximum observation confidence of each observation boundary point of all observation cross-section data.

[0182] Specifically, through the calculation in step S1032 above, each lane boundary point in the lane model may match multiple observation boundary points, meaning that the maximum observation confidence of multiple observation boundary points corresponds to the same lane boundary point. The maximum value among the maximum observation confidence of the multiple observation boundary points matched by each lane boundary point is taken as the observation confidence of that lane boundary point.

[0183] In step S104 above, the overall confidence of each lane model is calculated based on the observed confidence of each lane boundary point and the true confidence of each lane edge point in each lane model, and the lane model with the highest overall confidence is obtained.

[0184] Specifically, for each lane model, the observed confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model are summed to obtain the overall confidence level of the lane model, and the lane model with the highest overall confidence level is taken as the fused lane model.

[0185] In this embodiment of the application, after obtaining the fused lane model, multiple new observation data can be acquired, and the above steps S102 to S104 can be re-executed based on the multiple new observation data and the fused lane model.

[0186] In this process, the initial confidence levels set in S102 to S104 need to be replaced with the updated confidence levels obtained in step S103 above.

[0187] Meanwhile, in step S10333 above, the updated confidence of the observation boundary points and observation edge points in the new observation data will no longer be compared with the initial true confidence, but with the updated confidence of the observation boundary points and observation edge points in the previous batch of observation data, and the updated confidence of the observation boundary points and observation edge points in the previous batch of observation data needs to be multiplied by a decay coefficient (e.g., 0.9).

[0188] In this embodiment of the application, after step S103, the Kalman filter algorithm can be used to smooth the dividing lines and edge lines in the lane model respectively, which can effectively reduce noise interference and optimize the line shape.

[0189] In this embodiment of the application, it should be noted that the determination of the coordinates of edge points in the observed cross-sectional data includes the following three cases:

[0190] The first case: The edge line is a three-dimensional object in the actual road (such as a flower bed by the roadside), and it appears as a cluster of many edge points in the observation section data. Then, the point in the middle of these edge points is taken as the edge point in the observation section data.

[0191] The second scenario: If the edge line exists in the actual road and is a straight line, then the point closest to the outermost dividing point is taken as the edge point.

[0192] The third scenario: If there is no edge line in the actual road, then the outermost dividing point is taken as the edge point.

[0193] Example 2

[0194] Based on the same inventive concept, embodiments of the present invention also provide a lane map data fusion device, referring to... Figure 5 As shown, the device includes:

[0195] The first acquisition module 101 is used to acquire an initial lane model and its lane information; the lane information of the initial lane model includes the number of lanes and edge line information;

[0196] The second acquisition module 102 is used to obtain multiple lane models based on the acquired multiple observation data, the initial lane model, the number of lanes, and the edge line information;

[0197] The first calculation module 103 is used to update the lane model and determine the observation confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data.

[0198] The second calculation module 104 is used to calculate the overall confidence of each lane model based on the observed confidence of each lane boundary point and the true confidence of each lane edge point in each lane model, and to obtain the lane model with the highest overall confidence.

[0199] Example 3

[0200] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the lane map data fusion method as described in Embodiment 1 above.

[0201] Example 4

[0202] Based on the same inventive concept, this embodiment of the invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the lane map data fusion method as described in Embodiment 1 above.

[0203] Example 5

[0204] Based on the same inventive concept, this embodiment of the invention also provides a computer program product containing instructions that, when the computer program product is run on a computer device, cause the computer device to execute the lane map data fusion method as described in Embodiment 1 above.

[0205] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0206] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0207] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0208] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0209] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for fusion of lane map data, characterized in that, include: Obtain the initial lane model and its lane information; The lane information of the initial lane model includes the number of lanes and edge line information; Based on the acquired observation data, the initial lane model, the number of lanes, and the edge line information, multiple lane models are obtained; Based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data, the lane model is updated and the observation confidence and the true confidence of each lane boundary point and each lane edge point in the lane model are determined. Based on the observed confidence level of each lane boundary point and the true confidence level of each lane edge point in each lane model, the overall confidence level of each lane model is calculated, and the lane model with the highest overall confidence level is obtained.

2. The method as described in claim 1, characterized in that, The process of updating the lane model and determining the observation confidence level and the true confidence level of each lane boundary point and each lane edge point in the lane model, based on multiple lane cross-section data and multiple observation cross-section data obtained from each lane model and observation data, includes: Each lane model and each observation data is divided into multiple lane cross-section data and multiple observation cross-section data, and the initial sub-configuration confidence of each lane boundary point in the multiple lane cross-section data is set. For each lane model, for each observation section of each observation data, based on the observation section data and the initial sub-configuration confidence of each lane boundary point, the first maximum observation confidence, the first sub-configuration confidence, and the first true confidence of each observation boundary point are calculated, and the first true confidence and the first parameter confidence of each observation edge point are calculated. Based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point, and the first true confidence and first parameter confidence of each observation edge point calculated, the lane model is updated and the observation confidence of each lane boundary point and the true confidence of each lane edge point in the lane model are determined.

3. The method as described in claim 2, characterized in that, For each lane model, for each observation cross-section of each observation data point, based on the observation cross-section data and the initial configuration confidence of each lane boundary point, the first maximum observation confidence, the first configuration confidence, and the first true confidence are calculated for each observation boundary point. Furthermore, the first true confidence and the first parameter confidence are calculated for each observation edge point, including: Based on the lane cross-section data of each lane in the lane model and the observation cross-section data at the corresponding location, the observation confidence level of each observation boundary point in the observation cross-section data corresponding to each lane boundary point in the lane cross-section data is calculated. The maximum value among the multiple observation confidence scores corresponding to each observation boundary point is taken as the first maximum observation confidence score of the observation boundary point; For each observation boundary point, the multiple observation confidence scores corresponding to the observation boundary point are multiplied by the initial allocation confidence scores of the lane boundary points corresponding to the observation boundary point to obtain multiple probability factors corresponding to the observation boundary point. The maximum value of the normalized multiple probability factors corresponding to the observation boundary point is taken as the first sub-configuration confidence of the observation boundary point. Based on the observation cross-sectional data and the first maximum observation confidence of the observation boundary point, the first true confidence of the observation boundary point is calculated. Based on the observation cross-sectional data, the first true confidence score and the first parameter confidence score of each observation edge point in the observation cross-sectional data are calculated.

4. The method as described in claim 3, characterized in that, The observation cross-sectional data also includes trajectory points and the initial true confidence level of each observation edge point; The calculation of the first true confidence score and the first parameter confidence score for each observation edge point in the observation cross-section data based on the observation cross-section data includes: For each observation edge point, calculate the confidence factor of the observation edge point; The maximum value between the confidence factor of the observed edge point and the initial true confidence is taken as the first true confidence of the observed edge point; For each observation edge point, calculate the sum of the distances from every two adjacent observation boundary points in the observation cross-sectional data where the observation edge point is located to the trajectory point, and obtain the coordinates of the two adjacent observation boundary points with the smallest sum of distances; Calculate the mean of the coordinates of the two adjacent observation boundary points with the smallest sum of distances to obtain the matching lane center point corresponding to the observation edge point; The first parameter confidence level of the observed edge point is calculated based on the matching lane center point corresponding to the observed edge point.

5. The method as described in claim 4, characterized in that, The lane model also includes the lane width; the observation edge points include the observation left edge point and the observation right edge point; The initial true confidence level of each observed edge point is obtained in the following manner: The initial true confidence level of each observed left edge point is calculated based on the coordinates of each observed left edge point, the lane width, and the number of lanes. The initial true confidence level of each observed right edge point is calculated based on the coordinates of each observed right edge point, the lane width, and the number of lanes.

6. The method as described in claim 2, characterized in that, The step of updating the lane model and determining the observation confidence of each lane boundary point and the true confidence of each lane edge point based on the first maximum observation confidence, first sub-configuration confidence, and first true confidence of each observation boundary point, as well as the first true confidence and first parameter confidence of each observation edge point, includes: Based on the first maximum observation confidence, first subconfiguration confidence, and first true confidence of each observation boundary point of all the aforementioned observation cross-sectional data, determine the maximum observation confidence, subconfiguration confidence, and true confidence of each observation boundary point respectively; Based on the first true confidence and the first parameter confidence of each observation edge point of all the observation cross-sectional data, the true confidence and parameter confidence of each observation edge point are determined respectively; The lane model is updated based on the configuration confidence and true confidence of each observation boundary point, and the true confidence and parameter confidence of each observation edge point, and the true confidence of each lane edge point in the lane model is obtained. The observation confidence level of each lane boundary point in the lane model is determined based on the first maximum observation confidence level of each observation boundary point of all the observation cross-sectional data.

7. The method as described in claim 6, characterized in that, After dividing each lane model and each observation data into multiple lane cross-section data and multiple observation cross-section data respectively, the method further includes setting the initial true confidence of each lane edge point and the initial true confidence of each lane boundary point in the multiple lane cross-section data. The process of updating the lane model based on the configuration confidence and true confidence of each observation boundary point, and the true confidence and parameter confidence of each observation edge point, and obtaining the true confidence of each lane edge point in the lane model, includes: The updated confidence level of each observation boundary point is obtained by multiplying the distributed confidence level of each observation boundary point with the true confidence level. The updated confidence of each observed edge point is obtained by multiplying the true confidence of each observed edge point by the parameter confidence. If the updated confidence of each observation boundary point or observation edge point is greater than the initial true confidence of the observation boundary point or observation edge point, then the lane boundary point or lane edge point in the lane model is updated to the observation boundary point or observation edge point, and the true confidence of the lane boundary point or lane edge point is updated to the updated confidence of the observation boundary point or observation edge point.

8. The method as described in claim 7, characterized in that, The process involves obtaining multiple lane models based on acquired observation data, the initial lane model, the number of lanes, and the edge line information, including: The initial lane model is used as a lane model; The coordinates of the leftmost and rightmost observation boundaries are calculated based on the multiple observation data. The coordinates of the left and right observation edges are calculated based on the multiple observation data and the coordinates of the leftmost and rightmost observation boundaries. The total road width, the number of estimated lanes, and the probability of each estimated lane number are calculated based on the coordinates of the observed left edge line and the observed right edge line. When the probability that the estimated number of lanes equals the number of lanes plus one is greater than a preset threshold, determine whether the left edge line exists in the edge line information: If not, add a new dividing line and a new edge line to the left of the initial lane model to obtain a lane model; When the probability that the estimated number of lanes equals the number of lanes plus one is greater than the preset threshold, determine whether the right edge line exists in the edge line information: If not, add a new dividing line and a new edge line to the right of the initial lane model to obtain a new lane model.

9. The method as described in claim 8, characterized in that, The lane information also includes the total number of lane dividers; The initial true confidence level of each lane edge point, and the initial true confidence level and initial partial confidence level of each lane boundary point in the multiple lane cross-section data are set in the following manner: The initial true confidence of lane boundary points and lane edge points on newly added boundary lines and newly added edge lines in all the lane cross-section data is set to the preset new true confidence value, and the initial true confidence of other lane boundary points and lane edge points in the lane cross-section data is set to the preset initial true confidence value. For each lane cross-section model, determine whether there is a lane dividing point on the newly added dividing line in the lane cross-section model: If so, the initial configuration confidence of all lane dividing points in the lane cross-section model is set to the reciprocal of the total number of dividing lines plus one. If not, then set the initial configuration confidence of all lane dividing points in the lane cross-section model to the reciprocal of the total number of dividing lines.

10. A lane map data fusion device, characterized in that, include: The first acquisition module is used to acquire the initial lane model and its lane information; The lane information of the initial lane model includes the number of lanes and edge line information; The second acquisition module is used to obtain multiple lane models based on the acquired multiple observation data, the initial lane model, the number of lanes, and the edge line information; The first calculation module is used to update the lane model and determine the observation confidence level of each lane boundary point and the true confidence level of each lane edge point in the lane model based on multiple lane cross-section data and multiple observation cross-section data obtained by cutting each lane model and observation data. The second calculation module is used to calculate the overall confidence of each lane model based on the observed confidence of each lane boundary point and the true confidence of each lane edge point in each lane model, and to obtain the lane model with the highest overall confidence.

11. A computer-readable storage medium storing instructions that, when executed on a terminal, cause the terminal to perform the lane map data fusion method as described in any one of claims 1-9.

12. A computer device, characterized in that, The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the lane map data fusion method as described in any one of claims 1-9.

13. A computer program product containing instructions that, when run on a computer device, causes the computer device to perform the lane map data fusion method as described in any one of claims 1-9.

Images