A method for verifying map data using virtual road information

Abstract

An embodiment of the present invention provides a method for verifying map data using virtual road information, including: establishing virtual road information that simulates a real road environment for a target collection road; using the established virtual road information to verify received vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data; and determining that the map data verification has passed when the vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data all pass the verification.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving, and more particularly to a method for verifying map data using virtual road information. Background Art

[0002] With the continuous iterative development of autonomous driving technology, high-precision, high-quality map data is becoming increasingly important. Due to the widespread use of crowdsourcing, current high-precision map data must undergo multiple rounds of optimization and fusion of vehicle-based mapping data before being generated in the cloud. Due to the complexity and diversity of roads and the sheer volume of data, map data collection is inherently time-consuming. Data must then be transferred back to the local server for analysis and troubleshooting. The process from data generation to final quality results is particularly lengthy, and due to differences in real-time environments, problem analysis is labor-intensive and ineffective. Summary of the Invention

[0003] This solution provides a method for verifying map data using virtual road information. It compares and verifies map data in a real-time road environment, and analyzes map data from multiple dimensions such as road surface features, driving trajectories, and geometric integrity. It solves the pain points of unreliable data verification and difficulty in tracing problems in offline environments. It can detect problems in vehicle-side perception and mapping early, avoiding the need to push problems back to the vehicle after multiple cloud-based mapping fusions. This reduces problem locating time and improves data verification efficiency.

[0004] The technical solution of the present invention is:

[0005] The present invention provides a method for verifying map data using virtual road information, comprising:

[0006] Establish virtual road information that simulates the real road environment for the target collection road;

[0007] Using the established virtual road information, the received vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are verified;

[0008] When the vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are all verified, it is determined that the map data verification has passed.

[0009] Preferably, the step of establishing virtual road information simulating a real road environment for a target acquisition road includes:

[0010] For the roads to be collected, the road shape information, lane number information, slope information of each lane, curvature information of each lane, heading angle information of each lane, single and double line information of the road, virtual and real line information of the road, road surface marking information and road traffic light information, speed limit information of each lane, recommended speed information, linear change point information where the lane dividing line changes, and road interchange scene information are depicted.

[0011] Preferably, the process of verifying the received vehicle-side mapping data using the established virtual road information includes:

[0012] Perform field verification on the received vehicle-side mapping data to determine whether the fields of the vehicle-side mapping data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0013] Perform geometric information verification on the received vehicle-side mapping data. The geometric information verification process includes:

[0014] 1) Verifying the road surface elements of the received vehicle-side mapping data to determine whether the road surface elements of the vehicle-side mapping data meet the corresponding predetermined element design process requirements;

[0015] 2) Compare the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the vehicle-side mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information.

[0016] Preferably, the step of verifying the road surface elements of the received vehicle-side mapping data includes:

[0017] Verify whether the arrow direction in the vehicle-side mapping data is consistent with the driving direction, and whether the intersection is correctly generated with stop lines, zebra crossings, and traffic lights;

[0018] The step of comparing the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information includes:

[0019] Utilize the corresponding road surface elements and aerial elements in the established virtual road information to determine whether key identifications, centerline dividing lines and key signs are missing in the received vehicle-side mapping data.

[0020] Preferably, the process of verifying the received cloud-based mapping data using the established virtual road information includes:

[0021] Perform field verification on the received cloud-based mapping data to determine whether the fields of the cloud-based mapping data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0022] Perform geometric information verification on the received cloud-based mapping data. The geometric information verification process includes:

[0023] 1) Performing road element verification on the received cloud-based mapping data to determine whether the road elements in the cloud-based mapping data meet the corresponding predetermined element design process requirements;

[0024] 2) comparing the road surface elements and aerial elements of the received cloud-based mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the cloud-based mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information;

[0025] The topological relationship between elements in the received cloud-based mapping data is checked to determine whether the topological relationship between elements in the cloud-based mapping data meets the predetermined design topological relationship requirements.

[0026] Preferably, performing road feature verification on the received cloud-based mapping data includes:

[0027] Determine whether the center lines, dividing lines, and road boundaries in the received cloud-based mapping data are repeated or intersecting;

[0028] A buffer zone is defined using the selected target line. If a line of the same type exists within the buffer zone and its length difference from the target line is within a predetermined length range, then it is determined that there is line duplication in the cloud-based mapping data; the target line is a center line, a dividing line, or a road boundary line.

[0029] Use Shapely, a geometric set operation module in Python software, to perform pairwise intersection detection on all lines in the cloud mapping data to detect whether there are line intersections in the cloud mapping data.

[0030] Preferably, the process of verifying the received cloud-based high-precision map data using the established virtual road information includes:

[0031] Perform field verification on the received cloud-based high-precision map data to determine whether the fields of the cloud-based high-precision map data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0032] Perform logic verification on the received cloud-based high-precision map data to determine whether the logic of the cloud-based high-precision map data meets the predetermined design logic requirements;

[0033] Perform accuracy verification on the received high-precision map data in the cloud to determine whether the accuracy of the high-precision map data in the cloud meets the predetermined design accuracy requirements.

[0034] Preferably, the step of performing a logical check on the received cloud-based high-precision map data includes:

[0035] Verify the lane connection relationship of the center line in the cloud-based high-precision map data;

[0036] Verify whether the centerline lanes and boundaries in the cloud-based high-precision map data are missing;

[0037] Enter any valid starting point and valid end point in the cloud-based high-precision map data to verify whether a valid driving path can be generated in the cloud-based high-precision map data;

[0038] The steps of performing accuracy verification on the received cloud-based high-precision map data include: combining the established virtual road information, verifying whether the road curvature, slope, road length and number of segments in the cloud-based high-precision map data meet the preset process standards.

[0039] The beneficial effects of the present invention are:

[0040] Map data is compared and verified in a real-time road environment, and map data is analyzed from multiple dimensions such as road surface features, driving trajectories, and geometric integrity. This solves the pain points of unreliable data verification and difficulty in tracing problems in offline environments. It can detect problems in vehicle-side perception and mapping early, avoid multiple cloud-based map fusions and then push problems back to the vehicle side, reduce problem locating time, and improve data verification efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of crowdsourcing vehicle-side data reception and verification in the present invention;

[0042] Figure 2 This is a flow chart of crowdsourcing cloud data verification in the present invention;

[0043] Figure 3 This is a flowchart of receiving and verifying high-precision map data in the present invention;

[0044] Figure 4 Flowchart of a method for verifying map data using virtual road information in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The technical solutions involved are described in detail below with reference to the drawings in the embodiments of the present invention.

[0046] like Figure 4 , an embodiment of the present invention provides a method for verifying map data using virtual road information, comprising:

[0047] Establish virtual road information that simulates the real road environment for the target collection road;

[0048] Using the established virtual road information, the received vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are verified;

[0049] When the vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are all verified, it is determined that the map data verification has passed.

[0050] The steps of establishing virtual road information simulating a real road environment for a target acquisition road include:

[0051] For the roads to be collected, the road shape information, lane number information, slope information of each lane, curvature information of each lane, heading angle information of each lane, single and double line information of the road, virtual and real line information of the road, road surface marking information and road traffic light information, speed limit information of each lane, recommended speed information, linear change point information where the lane dividing line changes, and road interchange scene information are depicted.

[0052] The process of verifying the received vehicle-side mapping data using the established virtual road information includes:

[0053] Perform field verification on the received vehicle-side mapping data to determine whether the fields of the vehicle-side mapping data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0054] Perform geometric information verification on the received vehicle-side mapping data. The geometric information verification process includes:

[0055] 1) Verifying the road surface elements of the received vehicle-side mapping data to determine whether the road surface elements of the vehicle-side mapping data meet the corresponding predetermined element design process requirements;

[0056] 2) Compare the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the vehicle-side mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information.

[0057] The steps for verifying the road features of the received vehicle-side mapping data include:

[0058] Verify whether the arrow direction in the vehicle-side mapping data is consistent with the driving direction, and whether the intersection is correctly generated with stop lines, zebra crossings, and traffic lights;

[0059] The step of comparing the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information includes:

[0060] Utilize the corresponding road surface elements and aerial elements in the established virtual road information to determine whether key identifications, centerline dividing lines and key signs are missing in the received vehicle-side mapping data.

[0061] The process of verifying the received cloud-based mapping data using the established virtual road information includes:

[0062] Perform field verification on the received cloud-based mapping data to determine whether the fields of the cloud-based mapping data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0063] Perform geometric information verification on the received cloud-based mapping data. The geometric information verification process includes:

[0064] 1) Performing road element verification on the received cloud-based mapping data to determine whether the road elements in the cloud-based mapping data meet the corresponding predetermined element design process requirements;

[0065] 2) comparing the road surface elements and aerial elements of the received cloud-based mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the cloud-based mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information;

[0066] The topological relationship between elements in the received cloud-based mapping data is checked to determine whether the topological relationship between elements in the cloud-based mapping data meets the predetermined design topological relationship requirements.

[0067] Verification of road features on the received cloud-based mapping data includes:

[0068] Determine whether the center lines, dividing lines, and road boundaries in the received cloud-based mapping data are repeated or intersecting;

[0069] A buffer zone is defined using the selected target line. If a line of the same type exists within the buffer zone and its length difference from the target line is within a predetermined length range, then it is determined that there is line duplication in the cloud-based mapping data; the target line is a center line, a dividing line, or a road boundary line.

[0070] Use Shapely, a geometric set operation module in Python software, to perform pairwise intersection detection on all lines in the cloud mapping data to detect whether there are line intersections in the cloud mapping data.

[0071] The process of verifying the received high-precision map data from the cloud using the established virtual road information includes:

[0072] Perform field verification on the received cloud-based high-precision map data to determine whether the fields of the cloud-based high-precision map data are complete and valid. Field verification includes: field integrity verification and field validity verification;

[0073] Perform logic verification on the received cloud-based high-precision map data to determine whether the logic of the cloud-based high-precision map data meets the predetermined design logic requirements;

[0074] Perform accuracy verification on the received high-precision map data in the cloud to determine whether the accuracy of the high-precision map data in the cloud meets the predetermined design accuracy requirements.

[0075] The steps for logically verifying the received high-precision map data from the cloud include:

[0076] Verify the lane connection relationship of the center line in the cloud-based high-precision map data;

[0077] Verify whether the centerline lanes and boundaries in the cloud-based high-precision map data are missing;

[0078] Enter any valid starting point and valid end point in the cloud-based high-precision map data to verify whether a valid driving path can be generated in the cloud-based high-precision map data;

[0079] The steps of performing accuracy verification on the received cloud-based high-precision map data include: combining the established virtual road information, verifying whether the road curvature, slope, road length and number of segments in the cloud-based high-precision map data meet the preset process standards.

[0080] like Figures 1 to 3 Specifically, the above method of this embodiment is executed according to the following process when actually executed:

[0081] 1. Establish reliable virtual road geometry information for target collection roads

[0082] Based on the road to be collected, the accurate road shape, number of lanes, and data such as the slope, curvature, and heading of each lane must be depicted. Furthermore, detailed information on geometric elements such as single and double lane lines, dashed and solid lines, road markings, and traffic lights must be included. For autonomous driving considerations, details such as lane speed limits, recommended speeds, linear change points where dividing lines change, and interchange road scenes should also be described as much as possible. This creates virtual road information that is as close to the real road environment as possible, which can be used as a reference for verification in subsequent steps.

[0083] 2. Regularly receive crowdsourced real-time vehicle-side mapping data and automatically verify it

[0084] As the test vehicle collects road data for mapping, it generates object data and vehicle trajectory protobuf data at approximately 1-second intervals, including latitude and longitude coordinate groups, IDs, type, color, and other information. First, the real-time data is received through a specific port, and the driving trajectory data is automatically checked for accuracy. If trajectory deviation or data is missing, a warning message is immediately issued, avoiding the need to wait until the entire road section is collected and the data is copied back to the local analysis before discovering any problems. The road surface feature object data is then analyzed for field integrity and geometric correctness, including verifying that the direction of arrows aligns with the direction of travel and that stop lines, zebra crossings, and traffic lights are correctly generated at intersections. The data is then compared with the standard virtual road geometry established in Step 1 to check for missing key markings, centerline dividing lines, and key signage.

[0085] This step can help identify problems such as poor perception or incorrect data collection caused by serious program errors, sensor or vehicle positioning errors, etc. during the early stage of collecting road data for vehicle-side mapping, so that timely on-site debugging can be carried out to avoid wasting a lot of time copying the data back to the development environment for data analysis.

[0086] 3. Receive cloud data and automatically perform geometric logic verification

[0087] Although cloud data requires multiple rounds of cleaning and fusion with vehicle-side data, and there's no guarantee that cloud data will be generated in real time while the vehicle is driving, verifying the generated cloud data by referencing the virtual road information in step 1 is more convincing in a real-world environment than offline data verification, and can expose problems more directly. Specific steps include:

[0088] Poll and scan the designated port. When the latest cloud data is received, perform the following series of verification steps, comparing it with the virtual road geometry standard information in step 1.

[0089] Basic field-level validation, including whether the data ID value is unique, whether the enumeration value is valid, and whether the length and other thresholds are within the range;

[0090] Based on the geometric information of the road, check whether the centerline, dividing line, and road boundary are repeated or intersecting. A buffer area is defined based on the current centerline or dividing line range. The same type of line is searched and a length threshold is set. When the same line is matched in the buffer area and the length of the intersecting area is greater than the set threshold, the two lines are considered duplicates. Use Shapely to generate LineString objects for each pair of lines and perform intersection detection.

[0091] A center line will be generated based on the lane dividing lines on both sides. Where the dividing lines change (dashed line to solid line, single line to double line, etc.), a linear change point should be made on the corresponding center line: in the dividing line list, the connected dividing lines are range-matched according to the starting and ending points. If the two matching dividing lines are of different line shapes or additional line shapes, the vertical projection point of the intersection of the two dividing lines is made on the corresponding center line as the line shape change record point. A range match is performed around this point to see if there is a corresponding point in the generated data. In addition, on the center line, geometric points will be made for marking based on the surrounding traffic lights, signs, and special road scenes. For these points, it is necessary to verify whether the associated information in their geometric information is correct.

[0092] 4. Receive high-precision map data and perform automated verification

[0093] Poll through a specific port to obtain the latest serialized data generated by the high-precision map module after processing crowdsourced map data and store it in a specific location:

[0094] Poll the SDK to obtain the latest serialized data, deserialize and parse it into the geojson format commonly used for geography, so as to facilitate subsequent steps;

[0095] Extract all field names from the converted data (1) and search the field names in the proto file in the extracted information to determine whether all signals are output normally. Then perform field integrity verification, including checking whether there are missing fields, whether existing fields are legal, whether the values ​​are within the specified threshold range, and checking the uniqueness of the centerline boundary ID.

[0096] ① Perform a logical check on the road information, such as whether the lane connection relationship of the centerline is correct, whether there are any missing lanes and boundaries, and whether any valid starting and ending points can find a valid driving path in the data. Match a unique lane from the given starting point, and match the associated subsequent lanes from the lane in the driving direction. Traverse each connection to see if a driving path between the starting and ending points can be found. If this road does not intersect the boundary line, it is considered that at least one valid driving path has been found.

[0097] ② Verify the accuracy of high-precision map data: For road curvature, slope, road length and number of segments, automatically verify the data processed in step ① based on the standard value or error threshold of the process standard to check whether the map data accuracy and process value meet the applicable standards.

[0098] The above-mentioned method of the embodiment of the present invention compares and verifies map data in a real-time road environment, and analyzes map data from multiple dimensions such as road surface elements, driving trajectories, and geometric integrity, thereby solving the pain points of unreliable data verification and difficulty in tracing problems in an offline environment. It can detect problems in vehicle-side perception and mapping early, avoid multiple fusion maps in the cloud, and then push problems back to the vehicle side, thereby reducing problem locating time and improving data verification efficiency.

Claims

1. A method for verifying map data using virtual road information, characterized in that: include: Establish virtual road information that simulates the real road environment for the target collection road; Using the established virtual road information, the received vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are verified; When the vehicle-side mapping data, cloud-based mapping data, and cloud-based high-precision map data are all verified, it is determined that the map data verification has passed; The process of verifying the received vehicle-side mapping data using the established virtual road information includes: Perform field verification on the received vehicle-side mapping data to determine whether the fields of the vehicle-side mapping data are complete and valid; Verify the geometric information of the received vehicle-side mapping data; The process of verifying the received cloud-based mapping data using the established virtual road information includes: Perform field verification on the received cloud-based mapping data to determine whether the fields of the cloud-based mapping data are complete and valid; Verify the geometric information of the received cloud-based mapping data; Among them, before verifying the vehicle-side mapping data, it includes: generating object data and vehicle trajectory protobuf data at predetermined time intervals during the test vehicle's road data mapping process; receiving real-time generated data through a specific port, and automatically checking whether the driving trajectory data is correct. If the trajectory is found to be offset or data is missing, a warning message will be immediately issued.

2. The method according to claim 1, characterized in that The steps of establishing virtual road information simulating a real road environment for a target acquisition road include: For the roads to be collected, the road shape information, lane number information, slope information of each lane, curvature information of each lane, heading angle information of each lane, single and double line information of the road, virtual and real line information of the road, road surface marking information and road traffic light information, speed limit information of each lane, recommended speed information, linear change point information where the lane dividing line changes, and road interchange scene information are depicted.

3. The method according to claim 1, characterized in that The process of field verification of the received vehicle-side mapping data includes: field integrity verification and field validity verification; The process of geometric information verification of the received vehicle-side mapping data includes: 1) Verify the road surface elements of the received vehicle-side mapping data to determine whether the road surface elements of the vehicle-side mapping data meet the corresponding predetermined element design process requirements; 2) Compare the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the vehicle-side mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information.

4. The method according to claim 3, characterized in that The steps for verifying the road features of the received vehicle-side mapping data include: Verify whether the arrow direction in the vehicle-side mapping data is consistent with the driving direction, and whether the intersection is correctly generated with stop lines, zebra crossings, and traffic lights; The step of comparing the road surface elements and aerial elements of the received vehicle-side mapping data with the corresponding road surface elements and aerial elements in the established virtual road information includes: Utilize the corresponding road surface elements and aerial elements in the established virtual road information to determine whether key identifications, centerline dividing lines and key signs are missing in the received vehicle-side mapping data.

5. The method according to claim 1, wherein The process of field verification for received cloud-based mapping data includes: field integrity verification and field validity verification; The process of geometric verification of the received cloud-based mapping data includes: 1) Performing road element verification on the received cloud-based mapping data to determine whether the road elements in the cloud-based mapping data meet the corresponding predetermined element design process requirements; 2) Comparing the road surface elements and aerial elements of the received cloud-based mapping data with the corresponding road surface elements and aerial elements in the established virtual road information to determine whether the road surface elements and aerial elements of the cloud-based mapping data are consistent with the corresponding road surface elements and aerial elements in the established virtual road information; The topological relationship between elements in the received cloud-based mapping data is checked to determine whether the topological relationship between elements in the cloud-based mapping data meets the predetermined design topological relationship requirements.

6. The method according to claim 5, characterized in that Verification of road features on the received cloud-based mapping data includes: Determine whether the center lines, dividing lines, and road boundaries in the received cloud-based mapping data are repeated or intersecting; A buffer zone is defined using the selected target line. If a line of the same type exists within the buffer zone and its length difference from the target line is within a predetermined length range, then it is determined that there is line duplication in the cloud-based mapping data; the target line is a center line, a dividing line, or a road boundary line. Use Shapely, a geometric set operation module in Python software, to perform pairwise intersection detection on all lines in the cloud mapping data to detect whether there are line intersections in the cloud mapping data.

7. The method according to claim 1, characterized in that The process of verifying the received high-precision map data from the cloud using the established virtual road information includes: Perform field verification on the received cloud-based high-precision map data to determine whether the fields of the cloud-based high-precision map data are complete and valid. Field verification includes: field integrity verification and field validity verification; Perform logic verification on the received cloud-based high-precision map data to determine whether the logic of the cloud-based high-precision map data meets the predetermined design logic requirements; Perform accuracy verification on the received high-precision map data in the cloud to determine whether the accuracy of the high-precision map data in the cloud meets the predetermined design accuracy requirements.

8. The method according to claim 7, characterized in that The steps for logically verifying the received high-precision map data from the cloud include: Verify the lane connection relationship of the center line in the cloud-based high-precision map data; Verify whether the centerline lanes and boundaries in the cloud-based high-precision map data are missing; Enter any valid starting point and valid end point in the cloud-based high-precision map data to verify whether a valid driving path can be generated in the cloud-based high-precision map data; The steps of performing accuracy verification on the received cloud-based high-precision map data include: combining the established virtual road information to verify whether the road curvature, slope, road length and number of segments in the cloud-based high-precision map data meet the preset process standards.

Images