System for fusing two or more versions of map data

Through the central computer, different map data versions are received and fused, and map matching registration and maximum likelihood estimation algorithm are used to solve the problem of insufficient efficiency and accuracy of map data acquisition by autonomous vehicles, and map data with higher accuracy is generated.

CN118999518APending Publication Date: 2024-11-22GM GLOBAL TECHNOLOGY OPERATIONS LLC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311396188.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2023-05-16
Filing Date
2023-10-25
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

In the prior art, autonomous vehicles face problems of crowdsourcing map data calibration, GPS data deviation and noise, telemetry data lacks lane line attributes, inconsistent aerial map data and time-consuming update of high-definition map data, resulting in maps Insufficient data acquisition efficiency and accuracy.

Method used

Design a system to receive multiple versions of map data through a central computer, execute map matching and registration algorithms and maximum likelihood estimation algorithms, align and fuse them, generate more accurate fusion map data, and use iterative nearest point (ICP) Algorithm and maximum likelihood estimation algorithm, calculate map fusion offset, evaluate fusion performance, and generate map data with higher accuracy and accuracy.

Benefits of technology

It realizes the integration of different map data versions into map data with higher accuracy and accuracy, reducing the problem of insufficient time and accuracy of map data acquisition, and providing more stable map data for autonomous vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118999518A_ABST
    Figure CN118999518A_ABST
Patent Text Reader

Abstract

A system for fusing two or more versions of map data together includes one or more central computers to receive road network data representing a road network of a predefined geo-fenced area. A central computer calculates a plurality of points, each point being a predetermined distance from each other. The central computer creates a plurality of bounding boxes for the road network based on the plurality of points and creates a closest matching set of map data points for each bounding box as part of the road network by executing a map matching registration algorithm to align the two or more versions of map data with each other. The central computer executes a maximum likelihood estimation algorithm to determine probability distribution parameters for a set of matching map data points that are closest compared to real map data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to a system for fusing two or more versions of map data together to create fused map data. Background Art

[0002] Autonomous vehicles perform a variety of tasks, such as, but not limited to, perception, localization, mapping, path planning, decision making, and motion control. Autonomous vehicles rely on map data for performing many tasks, such as localization, mapping, and path planning. It should be understood that different versions of map data representing the same geographic area may be generated, where each version of the map data is generated from a different data source.

[0003] An example of a version of map data is crowdsourced map data. Crowdsourced map data is generated based on global positioning system (GPS) data and data collected by vehicles on the road. Some examples of data collected by vehicles include image data collected by onboard cameras, vehicle speed and yaw sensor data. Although crowdsourced maps provide real-time information for constructing lane lines and clarifying lane line attributes, crowdsourced maps may face challenges due to calibration issues of onboard cameras. Moreover, GPS data can introduce bias and random noise into crowdsourced maps.

[0004] Map data based on telemetry data, such as high-speed vehicle telemetry (HSVT) sources, provides high confidence and is generally relatively easy to obtain. However, telemetry-based map data does not provide lane line attributes such as color and type. In addition, telemetry-based map data is determined based on the behavior of individual drivers operating the vehicle. Aerial map data based on satellite image data can provide high accuracy. However, aerial map data may not consistently provide lane line attributes. In addition, aerial maps are time-consuming to create, may be outdated depending on when the images were collected, and may contain areas that are obscured due to issues such as being covered by trees. Finally, high-definition map data can be generated based on data collected from survey vehicles and can provide high accuracy as well as lane line attributes. However, updating high-definition map data is time-consuming.

[0005] Therefore, while autonomous vehicles achieve their intended purpose, there is a need in the art for an improved method for acquiring map data that mitigates the above-mentioned challenges. Summary of the invention

[0006] According to several aspects, a system for fusing two or more versions of map data is disclosed. The system includes one or more central computers in wireless communication with one or more communication networks, and is used to receive two or more versions of map data and real map data, wherein each version of the map data represents a predefined geo-fenced area. The one or more central computers execute instructions to receive road network data representing a road network for the predefined geo-fenced area, wherein the road network is a diagram of a road modeled based on a plurality of road segments. The central computer calculates a plurality of points each located at a predetermined distance from each other, wherein the plurality of points divide a single road segment of the plurality of road segments by a predetermined distance. The central computer creates a plurality of bounding boxes of the road network based on the plurality of points. The central computer aligns the two or more versions of the map data with each other by executing a map matching registration algorithm, and creates a closest matching set of map data points for each bounding box that is part of the road network. The central computer executes a maximum likelihood estimation algorithm to determine a probability distribution parameter of the closest matching set of map data points compared to the real map data. The central computer fuses the closest matching set of map data points together based on the probability distribution parameter to create fused map data.

[0007] In another aspect, the one or more central computers execute instructions to calculate a map fusion offset between a single fused map data point of the fused map data and a corresponding map data point of the real map data, wherein the map fusion offset represents an absolute map error.

[0008] In yet another aspect, the probability distribution parameters include a mean of a first offset between a first set of map data points and the real map data, a mean of a second offset between a second set of map data points and the real map data, a variance of the first offset, and a variance of the second offset.

[0009] In one aspect, the one or more central computers execute instructions to evaluate map fusion performance by comparing the mean and standard deviation of the map fusion offset to the mean and standard deviation of the first offset, and by comparing the mean and standard deviation of the map fusion offset to the mean and standard deviation of the second offset.

[0010] In another aspect, the plurality of bounding boxes each represent a single unit for performing the map matching registration algorithm to align the two or more versions of map data with each other.

[0011] In yet another aspect, the predetermined distance defines a width of a single bounding box.

[0012] In one aspect, each of the plurality of bounding boxes includes a lateral dimension and a longitudinal dimension corresponding to a perception range of the vehicle.

[0013] In another aspect, the plurality of nodes each represent an endpoint of one of the road segments.

[0014] In yet another aspect, the road network includes a plurality of line strings, and wherein each line string connects two of the plurality of nodes together.

[0015] In one aspect, the map matching registration algorithm is an iterative closest point (ICP) algorithm.

[0016] In another aspect, the two or more versions of map data are each based on one or more of: Global Positioning System (GPS) data, image data for the vehicle collected by an onboard camera, High Speed ​​Vehicle Telemetry (HSVT) sources, satellite image data, and data collected from a survey vehicle.

[0017] In yet another aspect, the one or more central computers execute instructions to generate an offset distribution histogram based on the real map data and a closest matching set of map data points for each bounding box that is part of the road network.

[0018] In one aspect, the portion of the map data points that are part of the closest matching set of map data points represent one or more landmarks.

[0019] On the other hand, a method for fusing two or more versions of map data together. The method includes receiving, by one or more central computers, road network data representing a road network of a predefined geo-fenced area, wherein the road network is a diagram of a road modeled based on a plurality of road segments. The method includes calculating, by the one or more central computers, a plurality of points each located at a predetermined distance from each other, wherein the plurality of points divide a single road segment of the plurality of road segments by a predetermined distance. The method includes creating, by the one or more central computers, a plurality of bounding boxes for the road network based on the plurality of points. The one or more central computers create a set of closest matching map data points for each bounding box that is part of the road network by executing a map matching registration algorithm to align the two or more versions of map data with each other, wherein each version of the map data represents the same predefined geo-fenced area; the method includes executing a maximum likelihood estimation algorithm to determine probability distribution parameters of the closest matching set of map data points compared to the real map data. Finally, the method includes fusing the closest matching map data points together based on the probability distribution parameters to create fused map data.

[0020] In another aspect, the method includes calculating a map fusion offset between a single fused map data point of the fused map data and a corresponding map data point of the real map data, wherein the map fusion offset represents an absolute map error.

[0021] In yet another aspect, the probability distribution parameters include a mean of a first offset between a first set of map data points and the real map data, a mean of a second offset between a second set of map data points and the real map data, a variance of the first offset, and a variance of the second offset.

[0022] In one aspect, the method includes evaluating map fusion performance by comparing the mean and standard deviation of the map fusion offset to the mean and standard deviation of the first offset, and by comparing the mean and standard deviation of the map fusion offset to the mean and standard deviation of the second offset.

[0023] On the other hand, a system for fusing two or more versions of map data is disclosed. The system includes one or more central computers, in wireless communication with one or more communication networks, for receiving two or more versions of map data and real map data, wherein each version of the map data represents a predefined geo-fenced area. The one or more central computers execute instructions to receive road network data representing a road network for the predefined geo-fenced area, wherein the road network is a diagram of a road modeled based on a plurality of road segments. The one or more central computers calculate a plurality of points each located at a predetermined distance from each other, wherein the plurality of points divide a single road segment of the plurality of road segments by a predetermined distance. The central computer creates a plurality of bounding boxes for the road network based on the plurality of points, wherein the plurality of bounding boxes each represent a single unit for performing an iterative closest point (ICP) algorithm to align the two or more versions of map data with each other. The central computer creates a closest matching set of map data points for each bounding box that is part of the road network by performing the ICP algorithm to align the two or more versions of map data with each other, and wherein the predetermined distance defines the width of a single bounding box. The one or more central computers execute a maximum likelihood estimation algorithm to determine probability distribution parameters of a set of map data points that are closest to the real map data. The one or more central computers fuse the sets of map data points that are closest to the real map data together based on the probability distribution parameters to create fused map data.

[0024] In yet another aspect, the one or more central computers execute instructions to calculate a map fusion offset between a single fused map data point of the fused map data and a corresponding map data point of the real map data, wherein the map fusion offset represents an absolute map error.

[0025] In one aspect, the probability distribution parameters include a mean of a first offset between a first set of map data points and the real map data, a mean of a second offset between a second set of map data points and the real map data, a variance of the first offset, and a variance of the second offset.

[0026] Further areas of applicability will become apparent from the description provided herein.It should be understood that the description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings described herein are for illustration purposes only and are not intended to limit the scope of the present disclosure in any way.

[0028] Figure 1 is a schematic diagram of a disclosed system for fusing two or more versions of map data together according to an exemplary embodiment, wherein the system includes one or more central computers that receive map data from different sources over a network;

[0029] Figure 2 FIG. 1 is a diagram showing a method according to an exemplary embodiment. Figure 1 A block diagram of the software architecture of one or more central computers shown;

[0030] Figure 3 is an exemplary illustration of a portion of a road network according to an exemplary embodiment, wherein a plurality of points are arranged along a line string of the road network;

[0031] Figure 4A According to an exemplary embodiment Figure 3 An exemplary illustration of a road network is shown, wherein a bounding box is created based on a plurality of points;

[0032] Figure 4B According to an exemplary embodiment Figure 4A An exemplary illustration of one of the bounding boxes shown in ;

[0033] Figure 5A shows an exemplary bounding box including map data points prior to executing a map matching registration algorithm according to an exemplary embodiment;

[0034] Figure 5B FIG. 1 shows a diagram of an exemplary embodiment of a map matching registration algorithm after executing the map matching registration algorithm. Figure 5A The bounding box shown;

[0035] Fig. 6A An exemplary sensor noise model according to an exemplary embodiment is shown;

[0036] Figure 6B showing vertical distances measured between exemplary map data points and corresponding actual map data points; and

[0037] Figure 7 is a diagram showing a method of performing a Figure 1The system shown is a flow chart of a method for fusing two or more versions of map data together. DETAILED DESCRIPTION

[0038] The following description is merely exemplary in nature and is not intended to limit the present disclosure, application, or uses.

[0039] refer to Figure 1 , an exemplary system 10 for fusing two or more versions of map data together is shown. The system 10 includes one or more central computers 20 located in a back-end office 22, wherein the one or more central computers 20 communicate wirelessly with one or more communication networks 24. The one or more central computers 20 obtain two or more versions of map data and real map data via one or more communication networks 24. Each version of the map data represents the same predefined geo-fenced area, wherein the predefined geo-fenced area represents a real-world geographic area defined by a virtual perimeter. It should be understood that each version of the map data is based on a unique data source, such as global positioning system (GPS) data, image data collected by a vehicle's onboard camera, a high-speed vehicle telemetry (HSVT) source, satellite image data, or data collected from a survey vehicle. Some examples of versions of map data include, but are not limited to, crowd-sourced map data, telemetry-based map data, and aerial map data. It should also be understood that high-definition map data represents real map data.

[0040] In one non-limiting embodiment, the map data is created specifically for an autonomous driving system (e.g., an automated driving system (ADS) or an advanced driver assistance system (ADAS)). In another embodiment, the map data is created for an inventory of road features, control devices, etc. In an embodiment, other dynamic map layers may be introduced on top of the map data, such as a map layer indicating the road surface or traffic conditions. As described below, the system 10 fuses two or more versions of map data together to create fused map data 30 ( Figure 2 ). Because the fused map data 30 is based on multiple unique data sources, the fused map data 30 includes greater precision and accuracy than the original versions of the map data alone. Figure 1 In the non-limiting embodiment shown, one or more central computers 20 communicate wirelessly with one or more autonomous vehicles 26. After determining fused map data 30, one or more central computers 20 may share fused map data 30 with one or more autonomous vehicles 26.

[0041] Figure 2 It shows Figure 1 A block diagram of the software architecture of one or more central computers 20 is shown in FIG. Figure 2In the example shown in FIG. 1 , one or more central computers 20 include a road segment module 32, a bounding box module 34, a closest pairing module 36, a sensor noise model module 38, a fusion module 40, and a statistical analysis module 42. The road segment module 32 of the one or more central computers 20 receives information representing the information received from one or more communication networks 24 ( Figure 1 ) input road network for predefined geo-fenced areas 50 ( Figure 3 )'s road network data 44. One example of the road network data 44 is an open street map (OSM), however, it should be understood that other types of road network data may also be used.

[0042] Figure 3 is an exemplary illustration of a portion of a road network 50. Figure 2 and Figure 3 , the road network 50 is a diagram for modeling a road based on a plurality of road segments 52, a plurality of nodes 54, a plurality of line strings 56, and opposing road edges 58. The plurality of nodes 54 each represent an endpoint of one of the road segments 52, wherein each line string 56 connects two nodes of the plurality of nodes 54 together. It should be understood that the opposing road edges 58 represent theoretical lanes or road edges, rather than opposing topological graphic edges. Figure 2 and Figure 3 , the road segment module 32 of one or more central computers 20 calculates a plurality of points 60 arranged along the plurality of line strings 56. Each of the plurality of points 60 is located at a predetermined distance from each other. The plurality of points 60 divides each road segment 52 into segments 62 at a predetermined distance. The predetermined distance defines a single bounding box 70 ( Figure 4A ) is shown in FIG. The predetermined distance represents the sensing range of a vehicle traveling along a road represented by the road network 50. In one non-limiting embodiment, the predetermined distance is approximately fifty meters, however, it should be understood that other dimensions may also be used.

[0043] Reference Figure 2 and Figure 3 In an embodiment, the segment module 32 of the one or more central computers 20 determines the plurality of points 60 by starting at a first node N1 of the initial segment 52 and placing a first predetermined point P1 at the first node N1. The segment module 32 then determines a subsequent point P2 at a predetermined distance along the line string 56. The segment module 32 continues to determine the predetermined points [P3, P4 ... Pi] for the entire length of the segment 52. In the event that the last point P1 is disposed at a distance less than the predetermined distance from the end node N2 where the segment 52 terminates, the length of the last segment 62 is less than the predetermined distance.

[0044] Figure 4A yes Figure 3FIG. 1 is an exemplary diagram of a road network 50 including a plurality of bounding boxes 70, wherein the bounding box module 34 ( Figure 2 ) creates a plurality of bounding boxes 70 based on the plurality of points 60. The plurality of bounding boxes 70 each represent a single unit for performing a point cloud registration algorithm (referred to as a map matching registration algorithm) to align two or more versions of map data points 72, 74 with each other. Figure 2 and 4A , each bounding box 70 includes a lateral dimension (i.e., width W) perpendicular to the road segment 52 and a longitudinal dimension (i.e., height H) aligned with the road segment 52, which dimensions correspond to the perception range of a vehicle traveling along the roads represented by the road network 50. Figure 4A In the example shown, a width W of the bounding box 70 is defined between points P1 and P2 , which is a predetermined distance between the plurality of points 60 .

[0045] Figure 4B Shows Figure 4A One of the bounding boxes 70 shown. Figure 2 and Figure 4B , an exemplary method for determining the width W and height H of the bounding box 70 will now be described. Figure 4B A vehicle coordinate system x'y' is shown, where the x' axis is parallel to the road edge 58 ( Figure 4A ), and the y' axis is perpendicular to the road edge 58. The heading angle θ represents the direction along the road segment 52 ( Figure 4A ), the coordinates x1 and y1 represent the coordinates of the first point P1, the coordinates x2 and y2 represent the coordinates of the second point P2, the distance d represents half of the height H of the bounding box 70, and Q1, Q2, Q3, and Q4 represent the corner points of the bounding box 70. In the example shown, the coordinates of the upper left corner point Q1 are represented by Q1=(x1, y1+d), the coordinates of the upper right corner point Q2 are represented by Q2=(x2, y2+d), the coordinates of the lower left corner point Q3 are represented by Q3=(x1, y1-d), and the coordinates of the lower right corner point Q4 are represented by Q4=(x2, y2-d).

[0046] Reference Figure 2 Once the bounding box module 34 creates the bounding box 70, the closest pairing module 36 of the one or more central computers 20 creates a closest matching set of map data points for each bounding box 70 that is part of the road network 50 by executing a map matching registration algorithm to align the two or more versions of the map data with each other. Figure 5A and Figure 5B An exemplary bounding box 70 is illustrated, which includes a first set of map data points 72 corresponding to a first version of the map data and a second set of map data points 74 corresponding to a second version of the map data. Figure 5AThe bounding box 70 shown shows the map data points 72, 74 before the map matching registration algorithm is performed, and Figure 5B The bounding box 70 shown shows the map data points 72, 74 after performing a map matching registration algorithm. In an embodiment, the map matching registration algorithm is an iterative closest point (ICP) algorithm, however, other types of algorithms may also be used. In an embodiment, the map data points 72, 74 that are part of the closest matching set of map data points represent lane lines set along a road. However, it should be understood that in addition to lane lines, a portion of the map data points 72, 74 may also represent one or more landmarks. Some examples of landmarks include, but are not limited to, traffic lights, road signs, or transverse road surface markings (such as stop lines, yield lines, and crosswalks).

[0047] The map matching registration algorithm aligns the first set of map data points 72 and the second set of map data points 74 with each other by translating or rotating one of the sets of map data points 72, 74 to be positioned closer to the other set of map data points 72, 74. A k-nearest neighbor algorithm is then performed to identify the closest matching set of map data points between the first set of map data points 72 and the second set of map data points 74. Figure 5B In the example shown, p1 from the first set of map data points 72 and p1' from the second set of map data points 74 represent sets of matched map data points. It should be appreciated that some map data points may not have corresponding matching map data points belonging to another set of map data points. It should also be appreciated that the crowdsourced map data and the telemetry-based map data do not change position when the map matching registration algorithm is executed.

[0048] refer to Figure 2 , Figure 5B and Fig. 6A The sensor noise model module 38 of the one or more central computers 20 then creates a deviation distribution histogram (which is referred to as a sensor noise model 80 and is displayed in Figure 1 ) based on the closest matching set of map data points between the first set of map data points 72 and the second set of map data points 74 for each bounding box 70 that is part of the road network 50 and the real map data. Fig. 6A As shown in Fig. 6A As seen in FIG. 8 , the sensor noise model 80 includes an x-axis indicating a map point offset representing the distance between each of the matched map data points from the set of closest matching map data points and the corresponding real map data point (the distance being measured in meters), and a y-axis indicating the number of points corresponding to each distance from the real map data point.

[0049] Specific reference Figure 5B , each bounding box 70 includes a data set D, where D = {(lh,i ,l m,i ,GT i )|i=1,…N}, and l h,i represents a point from the first map data point set 72, l m,i represents a point from the second map data point set 74, and GT i Represents points from real map data. The sensor noise model module 38 of one or more central computers 20 executes a maximum likelihood estimation algorithm to determine probability distribution parameters of a set of map data points that most closely matches the real map data. In one non-limiting embodiment, the probability distribution parameters include: an average value representing a first deviation between the first map data point set 72 and the real map data; an average value representing a second offset between the second map data point set 74 and the real map data; represents the variance of the first offset; represents the variance of the second offset; and represents the correlation coefficient between the first set 72 of map data points and the second set 74 of map data points. Figure 6B is to explain how to measure Fig. 6A 8 is an exemplary illustration of a map point offset, a first offset, a second offset, and a map fusion offset (which are described below) described in the sensor noise model 80 shown in FIG. Specifically, the map point offset, the first offset, the second offset, and the map fusion offset represent the vertical distance D measured between the corresponding map data point M and the imaginary line segment S connecting the two real map data points GT together.

[0050] In one embodiment, the probability distribution parameters are determined based on Equations 1-5, which are as follows:

[0051]

[0052]

[0053]

[0054]

[0055]

[0056] Where N represents the sample size, and GT i Represents real map data.

[0057] refer to Figure 2The fusion module 40 of the one or more central computers 20 then combines the first set of map data points 72 and the second set of map data points 74 ( Figure 5B The closest matching map data point sets between the two are fused together to determine the fused map data 30. Specifically, the fusion module 40 determines the fusion regression amount l based on the probability distribution parameter F It should be understood that based on the fusion regression F The fused map data points of the fused map data 30 are calculated. As described above, in an embodiment, a portion of the map data points 72, 74 represent landmarks. In the case where the map data points 72, 74 represent landmarks, the fusion module 40 of the one or more central computers 20 may first fuse the map data points 72, 74 representing landmarks classified as including the highest priority. After fusing together the map data points 72, 74 representing landmarks classified as having the highest priority, the fusion module 40 then fuses the map data points 72, 74 representing landmarks classified as having a lower priority, wherein the relative distances between the map data points 72, 74 are adjusted.

[0058] The priority of a landmark can be determined based on the uniqueness of the landmark in a particular environment. For example, in some rural or sparsely populated areas, traffic lights may be unique and difficult to find. As a result, in rural areas, traffic lights will be assigned the highest priority. By first determining the average lateral and longitudinal distances between the higher priority landmarks of the first map data point set 72 and the second map data point set 74, and then determining the average lateral and longitudinal distances between the lower priority landmarks of the first map data point set 72 and the second map data point set 74, the relative distances between the landmarks are adjusted. If the average lateral and longitudinal distances between the higher priority landmarks and the lower priority landmarks are approximately equal, then no relative distance adjustment is required. However, if the average lateral and longitudinal distances between the lower priority landmarks are greater than the average lateral and longitudinal distances between the higher priority landmarks, the lateral and longitudinal distances between the lower priority landmarks are gradually reduced until the average lateral and longitudinal distances between the higher priority landmarks and the lower priority landmarks are approximately equal.

[0059] Fusion regressor l F and the ensemble variance It can be determined based on Equation 6-16:

[0060] l h =L(p)+e h Equation 6

[0061] l m =L(p)+e m Equation 7

[0062] e h ~N(μ h ,σ h )Equation 8

[0063] e m ~N(μ m ,σ m )Equation 9

[0064] ρ=E{(e h -μ h )(e h -μ h )} / (σ h σ m ) Equation 10

[0065]

[0066]

[0067]

[0068]

[0069] l F =(A T ∑ -1 A) -1 A T ∑ -1 Equation 15

[0070]

[0071] Among them l h represents the lane line estimation of the first map data point set 72, l m represents the lane line estimation of the second map data point set 72, L(p) represents the true value at the position p in the lane line, and e h is a Gaussian random variable of the first map data point set 72, e m is a Gaussian random variable of the second map data point set 74, A is a matrix with all elements equal to 1, E represents the expected value, l represents a point in the map data, μ represents the mean value, and Σ represents the covariance matrix.

[0072] Once the fused map data 30 is determined, in an embodiment, the statistical analysis module 42 of one or more central computers 20 can then evaluate the map fusion performance through statistical analysis. In one embodiment, the statistical analysis module 42 calculates the map fusion offset between each fused map data point of the fused map data 30 and the corresponding map data point of the real map data, wherein the map fusion offset represents the absolute map error. As described above and as Figure 6B As shown, it should be understood that the map fusion offset represents the vertical distance D measured between the corresponding map data point M and the imaginary line segment S connecting the two real map data points GT together. The statistical analysis module 42 then calculates the mean and standard deviation of the map fusion offset. The statistical analysis module 42 then evaluates the map fusion performance by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the first offset, and by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the second offset.

[0073] Figure 7 is a process flow diagram illustrating a method 200 for fusing two or more versions of map data together. Figure 1-7 , the method 200 may begin at decision block 202. At decision block 202, the road segment module 32 of the one or more central computers 20 continues to monitor the one or more communication networks 24 until a road network 50 ( Figure 3 ) is a road network data 44. As described above, the road network 50 is a diagram that models roads based on a plurality of road segments 52. The method 200 may then proceed to block 204.

[0074] In block 204, the segment modules 32 of the one or more central computers 20 calculate a plurality of points 60 ( Figure 3 ), wherein the plurality of points 60 divides a single road segment 52 of the plurality of road segments 52 by a predetermined distance. The method 200 may then proceed to block 206.

[0075] In block 206, the bounding box module 34 of the one or more central computers 20 creates a plurality of bounding boxes 70 ( Figure 4A ). Specifically, Figure 4A As shown, the predetermined distance defines a width W of the single bounding box 70 . The method 200 may then proceed to block 208 .

[0076] In block 208 , the closest pair module 36 of the one or more central computers 20 creates a closest matching set of map data points for each bounding box 70 that is part of the road network 50 by executing a map matching registration algorithm to align the two or more versions of map data with each other. The method 200 may then proceed to block 210 .

[0077] In block 210, the sensor noise model module 38 of the one or more central computers 20 creates a sensor noise model 80 ( Fig. 6A). The sensor noise model module 38 also performs a maximum likelihood estimation algorithm to determine the probability distribution parameters of the set of map data points that most closely matches the true map data. The method 200 may then proceed to block 212.

[0078] In block 212 , the fusion module 40 of the one or more central computers 20 fuses the closest matching map data points together based on the probability distribution parameters to create the fused map data 30 . The method 200 may then proceed to block 214 .

[0079] In block 214, the statistical analysis module 42 of the one or more central computers 20 may then evaluate the map fusion performance based on the statistical analysis. Specifically, the statistical analysis module 42 evaluates the map fusion performance by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the first offset, and by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the second offset. The method 200 may then terminate.

[0080] In general, with reference to the accompanying drawings, the disclosed system for fusing two different versions of map data provides various technical effects and benefits. Specifically, the fused map data provides higher precision and accuracy compared to the original version of the map data. In addition, the disclosed system also provides a method for constructing a sensor noise model based on the fused map data and the real map data. It should be understood that the sensor noise model can be applied to roads including similar geographical features in the future when the real map data is not available.

[0081] The central computer may refer to an electronic circuit, a combinational logic circuit, a field programmable gate array (FPGA), a processor (shared, dedicated or grouped) that executes code, or a combination of some or all of the above, or as part of them (such as in a system on a chip). In addition, the controller may be microprocessor-based, such as a computer having at least one processor, memory (RAM and / or ROM), and associated input and output buses. The processor may operate under the control of an operating system resident in the memory. The operating system may manage computer resources so that computer program code embodied as one or more computer software applications (such as applications resident in the memory) may have instructions executed by the processor. In an alternative embodiment, the processor may directly execute the application, in which case the operating system may be omitted.

[0082] The description of the present disclosure is merely exemplary in nature, and variations that do not depart from the gist of the present disclosure are intended to be within the scope of the present disclosure. Such variations should not be regarded as departing from the spirit and scope of the present disclosure.

Claims

1. A system for fusing two or more versions of map data, the system comprising: One or more central computers in wireless communication with one or more communication networks for receiving two or more versions of map data and real map data, wherein each version of the map data represents a predefined geo-fenced area, and the one or more central computers execute instructions to: receiving road network data representing a road network for the predefined geo-fenced area, wherein the road network is a graph modeling roads based on a plurality of road segments; calculating a plurality of points each located at a predetermined distance from each other, wherein the plurality of points divide a single road segment of the plurality of road segments by the predetermined distance; creating a plurality of bounding boxes for the road network based on the plurality of points; aligning the two or more versions of map data with each other by executing a map matching registration algorithm to create a closest matching set of map data points for each bounding box that is part of the road network; executing a maximum likelihood estimation algorithm to determine probability distribution parameters of a set of map data points that most closely matches the true map data; and Based on the probability distribution parameters, the sets of closest matching map data points are fused together to create fused map data.

2. The system according to claim 1, wherein: The one or more central computers execute instructions to: A map fusion offset is calculated between each fused map data point of the fused map data and a corresponding map data point of the real map data, wherein the map fusion offset represents an absolute map error.

3. The system according to claim 2, wherein: The probability distribution parameters include a mean of a first offset between a first set of map data points and the real map data, a mean of a second offset between a second set of map data points and the real map data, a variance of the first offset, and a variance of the second offset.

4. The system according to claim 3, wherein: The one or more central computers execute instructions to: Map fusion performance is evaluated by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the first offset, and by comparing the mean and standard deviation of the map fusion offset with the mean and standard deviation of the second offset.

5. The system according to claim 1, wherein: The plurality of bounding boxes each represent a single unit for performing the map matching registration algorithm to align the two or more versions of map data with each other.

6. The system according to claim 1, wherein: The predetermined distance defines the width of a single bounding box.

7. The system according to claim 1, wherein: Each of the plurality of bounding boxes includes a lateral size and a longitudinal size corresponding to a perception range of the vehicle.

8. The system according to claim 1, wherein: The plurality of nodes each represent an endpoint of one of the road segments.

9. The system according to claim 8, wherein: The road network includes a plurality of line strings, and wherein each line string connects two of the plurality of nodes together.

10. The system according to claim 1, wherein: The map matching registration algorithm is an iterative closest point (ICP) algorithm.