High-precision map format conversion system and method for intelligent driving simulation testing

Abstract

This invention discloses a high-precision map format conversion system and method for intelligent driving simulation testing. The format conversion system includes a data reading module, a road network processing module, a traffic accessory object processing module, and a CarMaker road model IPGRoad5 file generation module. The data reading module is used to acquire a non-standard format high-precision map to be processed. The road network processing module is used to extract road network information from the high-precision map, reassemble the extracted data into a data structure used by the CarMaker road model, and draw the IPGRoad5 map network using the CarMaker road model API. The traffic accessory object processing module is used to extract traffic accessory object information from the high-precision map and draw the IPGRoad5 traffic accessory objects using the CarMaker road model API. The CarMaker road model IPGRoad5 file generation module is used to write the IPGRoad5 map format generated by the road network processing module and the traffic accessory object processing module into a file, outputting a file with the .rd5 extension.

Description

Technical Field

[0001] This invention belongs to the field of basic technology of intelligent driving, and relates to the front-end technology of intelligent driving perception, positioning and fusion technology and corresponding testing and evaluation system. It relates to intelligent driving software-in-the-loop, hardware-in-the-loop and driver-in-the-loop testing and verification system, specifically a high-precision map format conversion system and conversion method for intelligent driving simulation testing. Background Technology

[0002] In recent years, the fundamental technologies for autonomous driving have been continuously innovating and developing, with high-precision maps being a crucial foundational technology for high-precision positioning. The development of high-precision maps is closely related to autonomous vehicles; since autonomous vehicles began public testing on public roads, the high-precision map industry has emerged and developed rapidly. Compared to traditional navigation maps, high-precision maps are specifically designed for autonomous driving, serving not human drivers but autonomous vehicles. With the industry's development, more and more ADAS products are also applying high-precision maps to improve beyond-line-of-sight perception and planning capabilities.

[0003] For Level 3 and above autonomous vehicles, high-definition maps are essential. On one hand, high-definition maps are a crucial foundation for planning routes for autonomous vehicles, providing them with location, decision-making, and traffic dynamic information. On the other hand, high-definition maps ensure basic driving safety when autonomous vehicle sensors malfunction or in adverse environments. Simultaneously, high-definition maps are equally important in establishing simulation test scenarios. By injecting pre-collected high-definition maps into hardware-in-the-loop, software-in-the-loop, or driver-in-the-loop systems, real-world vehicle environmental information can be simulated, the actual operating environment can be reconstructed, and in-the-loop testing of controllers, control algorithms, and drivers can be completed. The introduction of high-definition maps brings a series of advantages: 1. Enhanced spatiotemporal perception capabilities, adding more prior information; 2. The ability to receive high-definition maps and traffic flow information even in adverse weather conditions, enabling stable vehicle operation; 3. Elimination of hardware errors caused by sensor failure; 4. The application of high-definition maps in simulation testing accelerates testing speed and enhances product safety.

[0004] Currently, major OEMs, technology companies, and map providers (Toyota, Baidu, NavInfo, etc.) are accelerating the acquisition of high-precision maps and the promotion of high-precision map products to improve the performance of their autonomous driving products and drive product iteration and upgrades. High-precision map acquisition requires significant manpower and time costs, and the annotation and processing of static map elements, as well as the updating and maintenance of dynamic map elements, also require substantial costs. Therefore, reducing the cost of using high-precision maps is crucial. However, due to differences in the definitions of high-precision map elements, objects, and format information among different manufacturers, even high-precision maps containing the same elements in the same area are difficult to reuse. This leads to a large number of identical maps being repeatedly collected in various stages of autonomous driving, such as development, verification, and testing, resulting in significant resource waste. Furthermore, the lack of format standardization hinders the further development of the industry. Summary of the Invention

[0005] To address the problems of existing technologies where high-precision maps collected by map-collecting vehicles require manual conversion into high-precision map formats for simulation testing, resulting in low efficiency, high cost, and difficulties in updating and maintenance, this invention provides a high-precision map format conversion system and method for intelligent driving simulation testing. By unifying and systematically processing manufacturer-specific, non-standard high-precision maps, it converts non-standard high-precision maps collected by vehicles equipped with sensors such as cameras and LiDAR into universal high-precision maps readable by the intelligent driving simulation testing software CarMaker. The converted high-precision map maintains the same level of accuracy as the original map, serving subsequent intelligent driving simulation testing. The converted standard format map also assists in the software and hardware development and testing of ADAS and autonomous driving controllers.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] As a first aspect of the present invention, a high-precision map format conversion system for intelligent driving simulation testing is provided, comprising a data reading module, a road network processing module, a traffic-related object processing module, and a CarMaker road model IPGRoad5 file generation module; wherein:

[0008] The data reading module is used to acquire a non-standard format high-precision map to be processed;

[0009] The road network processing module is used to extract road network information from the high-precision map from the data reading module, reassemble the extracted data into the data structure used by the CarMaker road model, and draw the IPGRoad5 map network using the CarMaker road model API.

[0010] The traffic accessory object processing module is used to extract traffic accessory object information from high-precision maps and draw IPGRoad5 traffic accessory objects using the CarMaker road model API.

[0011] The CarMaker road model IPGRoad5 file generation module is used to write the IPGRoad5 map format generated by the road network processing module and the traffic ancillary object processing module into a file, and output a file with the .rd5 extension.

[0012] Furthermore, the road network processing module includes an external road processing unit and an internal road processing unit; wherein:

[0013] The road processing unit outside the intersection is used for:

[0014] Extract high-precision map road data to determine the centerline coordinates of the high-precision map roads and the lane boundary information that makes up each road;

[0015] Except for connecting roads marked as intersections, each road in the high-precision map is connected end to end according to its extension direction to obtain high-precision map road information;

[0016] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0017] Using the CarMaker road model API, draw the road centerline based on the extracted road centerline coordinates and extension direction;

[0018] The width of each lane is calculated based on the extracted lane boundaries, and each lane in the road is drawn using the CarMaker road model API.

[0019] Furthermore, the road processing unit within the intersection is used for:

[0020] Delete the connecting roads marked as being within the intersection, but save their connection correspondence with roads outside the intersection;

[0021] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0022] The intersections are drawn using the CarMaker road model API, and the intersection connections are reconstructed based on the extracted and retained road connection correspondences.

[0023] Furthermore, the traffic-related object processing module is used for:

[0024] Extract high-precision map road data to determine traffic lights and road markings;

[0025] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0026] Using the CarMaker road model API, traffic signals and road markings are drawn based on the extracted traffic light and road marking information.

[0027] As a second aspect of the present invention, the present invention also provides a high-precision map format conversion method for intelligent driving simulation testing, comprising the following steps:

[0028] S1. Obtain the non-standard format high-precision map to be processed;

[0029] S2. Extract road network information from non-standard format high-precision maps, reassemble the extracted data into the data structure used by CarMaker road models, and use the CarMaker road model API to draw the IPGRoad5 map network;

[0030] S3. Extract information on traffic-related objects (such as traffic lights and road markings) from the high-precision map and use the CarMaker road model API to draw IPGRoad5 traffic-related objects;

[0031] S4. The IPGRoad5 format data generated by the road network processing module and the traffic ancillary object processing module is written into a file and output as a file with the .rd5 extension.

[0032] Further, step S2 specifically includes:

[0033] S21. Extract road data from non-standard format high-precision maps, determine the centerline coordinates of the roads in the high-precision map, and the lane boundary information that makes up each road;

[0034] S22. Except for connecting roads marked as intersections, connect each road in the high-precision map end to end in sequence according to its extension direction to obtain the road information of the high-precision map;

[0035] S23. Map the extracted high-precision map road information to the CarMaker road model data structure;

[0036] S24. Using the CarMaker road model API, draw the road centerline based on the extracted high-precision map's road centerline coordinates and extension direction;

[0037] S25. Calculate the width of each lane based on the lane boundaries extracted from the high-precision map, and draw each lane in the road using the CarMaker road model API;

[0038] S26. Delete the connecting roads marked as being inside intersections in the high-precision map, and save their correspondence with roads outside intersections;

[0039] S27. Map the extracted high-precision map road information to the CarMaker road model data structure;

[0040] S28. Use the CarMaker road API to draw intersections, and reconstruct the intersection connections based on the extracted and retained intersection road connection correspondences.

[0041] Preferably, in step S21, the direction of road extension is obtained based on the centerline coordinates of each road segment in the non-standard format high-precision map, and then the width and number of lanes of the road are restored by combining the lane boundary information of each road.

[0042] Preferably, in step S23, the high-precision map uses the WGS84 coordinate system to be converted into the CarMaker road model coordinate system.

[0043] Furthermore, step S3 specifically includes:

[0044] S31. Extract road information from the high-precision map and determine the traffic lights and road markings in the lanes of the high-precision map;

[0045] S32. Map the extracted high-precision map road information to the CarMaker road model data structure;

[0046] S33. Using the CarMaker road model API, draw traffic signals and road markings based on the extracted traffic light and road marking information.

[0047] Preferably, in step S31, the reconstruction of traffic lights and road markings requires obtaining their location and attribute information: first, the lane IDs bound to the traffic lights and road markings are obtained, and then their accurate location is obtained based on their lane-based position offset information; finally, the attribute information of the traffic lights and road markings, including the type of traffic lights and the type of road markings, is obtained.

[0048] The present invention has the following beneficial effects:

[0049] This invention provides a high-precision map format conversion system and method for intelligent driving simulation testing. By unifying and systematically processing manufacturer-specific, non-standard high-precision maps, it converts non-standard high-precision maps collected by vehicles equipped with sensors such as cameras and lidar into universal high-precision maps that can be read by the intelligent driving simulation testing software CarMaker, while ensuring that the converted high-precision map has the same level of accuracy as the original map.

[0050] This invention differs from high-precision map format conversion via satellite maps, which is costly and requires manual addition and correction of traffic lights, traffic signs, etc. This invention can perform high-precision map format conversion in a fully automated manner, which is efficient, low-cost, and does not require manual correction of the converted map. This provides convenience for subsequent intelligent driving simulation testing and improves testing efficiency.

[0051] In terms of social benefits, this invention has been successfully applied in actual intelligent driving simulation testing, maximizing the use of map acquisition resources, meeting the high-precision map requirements of intelligent driving simulation testing, and improving testing efficiency. In terms of economic benefits, to some extent, the results of this invention can replace open road testing for intelligent driving, significantly saving manpower and time costs associated with open road testing. Attached Figure Description

[0052] Figure 1 This is a flowchart of a high-precision map format conversion method for intelligent driving simulation testing as described in Embodiment 2 of the present invention;

[0053] Figure 2 This is a schematic diagram illustrating the application of the high-precision map format conversion method described in Embodiment 2 of the present invention;

[0054] Figure 3 This is a schematic diagram illustrating the application effect of the high-precision map format conversion method described in Embodiment 2 of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0056] Example 1

[0057] A high-precision map format conversion system for intelligent driving simulation testing includes a data reading module, a road network processing module, a traffic-related object processing module, and a CarMaker road model IPGRoad5 file generation module; wherein:

[0058] The data reading module is used to acquire a non-standard format high-precision map to be processed;

[0059] The road network processing module is used to extract road network information from the high-precision map from the data reading module, reassemble the extracted data into the data structure used by the CarMaker road model, and draw the IPGRoad5 map network using the CarMaker road model API.

[0060] The traffic accessory object processing module is used to extract traffic light and road marking information from high-precision maps and draw IPGRoad5 traffic accessory objects using the CarMaker road model API.

[0061] The CarMaker road model IPGRoad5 file generation module is used to write the RD5 map format generated by the data processing module into a file, outputting a file with the .rd5 extension.

[0062] Furthermore, the road network processing module includes an external road processing unit and an internal road processing unit, wherein:

[0063] The road processing unit outside the intersection is used for:

[0064] Extract high-precision map road data to determine the centerline coordinates of the high-precision map roads and the lane boundary information that makes up each road;

[0065] Except for connecting roads marked as intersections, each road in the high-precision map is connected end to end according to its extension direction to obtain high-precision map road information;

[0066] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0067] Using the CarMaker road model API, draw the road centerline based on the extracted road centerline coordinates and extension direction;

[0068] The width of each lane is calculated based on the extracted lane boundaries, and each lane in the road is drawn using the CarMaker road model API.

[0069] The road processing unit within the intersection is used for:

[0070] Delete the connecting roads marked as being within the intersection, but save their connection correspondence with roads outside the intersection;

[0071] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0072] The intersections are drawn using the CarMaker road model API, and the intersection connections are reconstructed based on the extracted and retained road connection correspondences.

[0073] Furthermore, the traffic-related object processing module is specifically used for:

[0074] Extract high-precision map road data to determine traffic lights and road markings;

[0075] The extracted high-precision map road information is mapped to the CarMaker road model data structure;

[0076] Using the CarMaker road model API, traffic signals and road markings are drawn based on the extracted traffic light and road marking information.

[0077] Example 2

[0078] like Figure 1 As shown, a high-precision map format conversion method for intelligent driving simulation testing includes the following steps:

[0079] S1. Obtain the non-standard format high-precision map to be processed;

[0080] S2. Extract road network information from non-standard format high-precision maps, reassemble the extracted data into the data structure used by CarMaker road models, and use the CarMaker road model API to draw the IPGRoad5 map network;

[0081] S3. Extract information on traffic-related objects (such as traffic lights and road markings) from the high-precision map and use the CarMaker road model API to draw IPGRoad5 traffic-related objects;

[0082] S4. The IPGRoad5 format data generated by the road network processing module and the traffic ancillary object processing module is written into a file and output as a file with the .rd5 extension.

[0083] Further, step S2 specifically includes:

[0084] S21. Extract road data from the non-standard format high-precision map, and determine the centerline coordinates of the roads in the high-precision map and the lane boundary information that makes up each road. It should be noted that in this embodiment, a matching data structure is created according to the file format definition of the non-standard format high-precision map to be processed to store the useful information contained in the high-precision map.

[0085] By using the centerline coordinates of each road segment in a non-standard format high-precision map, the direction of road extension can be determined. Combined with the lane boundary information of each road, the width and number of lanes of the road can be reconstructed.

[0086] S22. Except for connecting roads marked as intersections, connect each road in the high-precision map end-to-end according to its extension direction to obtain the high-precision map road information. It should be noted that the high-precision maps to be processed store road network information in a segmented manner. Therefore, during the reconstruction process, it is necessary to first connect the segmented road information end-to-end according to its extension direction to restore the original road network. It should be pointed out here that using the CarMaker road model API only requires determining the connecting road between two intersections as a single road segment for drawing, without needing to redraw the road network using the segmented storage method of the high-precision map. Therefore, connecting roads end-to-end according to their extension direction also facilitates later reconstruction in CarMaker.

[0087] S23. Map the extracted high-precision map road information to the CarMaker road model data structure. Before drawing a map using the CarMaker road model API, you first need to assign values ​​to the road data structure used by the API. This involves mapping the extracted high-precision map information to the CarMaker road model data structure one by one. It's important to note that the high-precision map uses the WGS84 coordinate system, meaning that all coordinates in the map's roads are represented by longitude, latitude, and altitude. Converting the high-precision map to a CarMaker-readable map format requires converting the WGS84 coordinates to the CarMaker road model coordinate system.

[0088] S24. Using the CarMaker road model API, draw the road centerline based on the extracted high-precision map's road centerline coordinates and extension direction.

[0089] S25. Calculate the width of each lane based on the lane boundaries extracted from the high-precision map, and draw each lane in the road using the CarMaker road model API.

[0090] After steps S21 to S25, the drawing of all road networks in the high-precision map, except for intersections, is completed. It should be noted that the definition of intersections in the high-precision map differs significantly from the definition of intersections in the CarMaker road model. The main difference lies in the fact that the high-precision map defines intersections as representing the connection between the intersection's entrance and exit. For example, in a crossroads, the first lane of the north entrance can pass through the intersection and exit via the first and second lanes of the south exit, and so on. A series of connecting roads are generated within the intersection to connect the entrance and exit lanes. It should be pointed out that when representing the connection relationship of intersection turns, the connecting roads also have curvature. Therefore, the high-precision map does not have a single "physical" intersection; instead, it defines the connection relationships between the intersection's entrance and exit. In contrast, the CarMaker road model defines an intersection by adding a "physical" intersection. After all the entrance and exit roads are connected to the intersection, all connections within the intersection are automatically calculated by the road model, eliminating the need to redefine the connection relationships between the entrance and exit lanes. After adding intersections to the CarMaker road model, all intersection connections in the high-precision map can be reproduced. Therefore, this embodiment proposes to use "solid" intersections from the CarMaker road model to replace the connections within intersections in the high-precision map.

[0091] S26. Delete the connecting roads marked as being within intersections in the high-precision map, but preserve their correspondence with roads outside the intersections. Since the connecting road data within intersections in the high-precision map will no longer be used, it should be deleted before conversion. It should be noted that the entrance and exit road connections for all intersections must be retained, as these roads will connect to the newly created intersections in the CarMaker road model.

[0092] S27. Map the extracted high-precision map road information to the CarMaker road model data structure for use in the next step of building the CarMaker road model.

[0093] S28. Use the CarMaker road API to draw intersections, and reconstruct the intersection connections based on the extracted and retained intersection road connection correspondences.

[0094] Furthermore, step S3 specifically includes:

[0095] S31. Extract road information from the high-precision map and determine the traffic lights and road markings within the lanes of the high-precision map. It should be noted that reconstructing the traffic lights and road markings requires obtaining their location and attribute information. First, their location is based on lane positioning, so it is necessary to first obtain their bound lane ID, and then obtain their accurate location using their lane-based position offset information. Second, obtain the attribute information of the traffic lights and road markings (such as traffic light type and road marking type).

[0096] S32. Map the extracted high-precision map road information to the CarMaker road model data structure for use in the next step of building the CarMaker road model.

[0097] S33. Using the CarMaker road model API, draw traffic signals and road markings based on the extracted traffic light and road marking information.

[0098] Based on the above steps, the high-precision map information has been correctly read and the IPGRoad5 format road network has been generated using the CarMaker road model API.

[0099] Application examples of this invention:

[0100] First, high-precision map data was collected from a section of urban road in Yizhuang, Beijing. Then, using the system and method provided in this invention, the collected high-precision map was automatically converted into a high-precision map format used by the simulation software Carmaker, such as... Figure 2 As shown; based on the format-converted high-precision simulation map, random traffic vehicles were added, and the simulation was applied to the CarMaker simulation software for urban autonomous driving function testing. The specific simulation test results are as follows. Figure 3 As shown.

[0101] In terms of social benefits, this invention has been successfully applied in actual intelligent driving simulation testing, maximizing the use of map acquisition resources, meeting the high-precision map requirements of intelligent driving simulation testing, and improving testing efficiency. In terms of economic benefits, to some extent, the results of this invention can replace open road testing for intelligent driving, significantly saving manpower and time costs associated with open road testing.

[0102] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision map format conversion system for intelligent driving simulation testing, characterized in that, It includes a data reading module, a road network processing module, a traffic-related object processing module, and a CarMaker road model IPGRoad5 file generation module; among which: The data reading module is used to acquire a non-standard format high-precision map to be processed; The road network processing module is used to extract road network information from the high-precision map from the data reading module, reassemble the extracted data into the data structure used by the CarMaker road model, and draw the IPGRoad5 map network using the CarMaker road model API. The traffic-related object processing module is used to extract traffic-related object information from high-precision maps and to draw IPGRoad5 traffic-related objects using the CarMaker road model API; specifically, the traffic-related object processing module is used for: Extract high-precision map road data to determine traffic lights and road markings; The extracted high-precision map road information is mapped to the CarMaker road model data structure; Using the CarMaker road model API, traffic signals and road markings are drawn based on the extracted traffic light and road marking information; The CarMaker road model IPGRoad5 file generation module is used to write the IPGRoad5 map format generated by the road network processing module and the traffic ancillary object processing module into a file, and output a file with the .rd5 extension.

2. The high-precision map format conversion system for intelligent driving simulation testing as described in claim 1, characterized in that, The road network processing module includes an external road processing unit and an internal road processing unit; wherein: The road processing unit outside the intersection is used for: Extract high-precision map road data to determine the centerline coordinates of the high-precision map roads and the lane boundary information that makes up each road; Except for connecting roads marked as intersections, each road in the high-precision map is connected end to end according to its extension direction to obtain high-precision map road information; The extracted high-precision map road information is mapped to the CarMaker road model data structure; Using the CarMaker road model API, draw the road centerline based on the extracted road centerline coordinates and extension direction; The width of each lane is calculated based on the extracted lane boundaries, and each lane in the road is drawn using the CarMaker road model API.

3. The high-precision map format conversion system for intelligent driving simulation testing as described in claim 2, characterized in that, The road processing unit within the intersection is used for: Delete the connecting roads marked as being within the intersection, but save their connection correspondence with roads outside the intersection; The extracted high-precision map road information is mapped to the CarMaker road model data structure; The intersections are drawn using the CarMaker road model API, and the intersection connections are reconstructed based on the extracted and retained road connection correspondences.

4. A method for converting high-precision map formats for intelligent driving simulation testing, characterized in that, Includes the following steps: S1. Obtain the non-standard format high-precision map to be processed; S2. Extract road network information from non-standard format high-precision maps, reassemble the extracted data into the data structure used by CarMaker road models, and use the CarMaker road model API to draw the IPGRoad5 map network; S3. Extract traffic-related information from the high-precision map and use the CarMaker road model API to draw IPGRoad5 traffic-related objects; Step S3 specifically includes: S31. Extract road information from the high-precision map and determine the traffic lights and road markings in the lanes of the high-precision map; S32. Map the extracted high-precision map road information to the CarMaker road model data structure; S33. Using the CarMaker road model API, draw traffic signals and road markings based on the extracted traffic light and road marking information; S4. The IPGRoad5 format data generated by the road network processing module and the traffic ancillary object processing module is written into a file and output as a file with the .rd5 extension.

5. A high-precision map format conversion method for intelligent driving simulation testing as described in claim 4, characterized in that, Step S2 specifically includes: S21. Extract road data from non-standard format high-precision maps, determine the centerline coordinates of the roads in the high-precision map, and the lane boundary information that makes up each road; S22. Except for connecting roads marked as intersections, connect each road in the high-precision map end to end in sequence according to its extension direction to obtain the road information of the high-precision map; S23. Map the extracted high-precision map road information to the CarMaker road model data structure; S24. Using the CarMaker road model API, draw the road centerline based on the extracted high-precision map's road centerline coordinates and extension direction; S25. Calculate the width of each lane based on the lane boundaries extracted from the high-precision map, and draw each lane in the road using the CarMaker road model API; S26. Delete the connecting roads marked as being inside intersections in the high-precision map, and save their correspondence with roads outside intersections; S27. Map the extracted high-precision map road information to the CarMaker road model data structure; S28. Use the CarMaker road API to draw intersections, and reconstruct the intersection connections based on the extracted and retained intersection road connection correspondences.

6. The high-precision map format conversion method for intelligent driving simulation testing as described in claim 5, characterized in that, In step S21, the direction of road extension is determined by the centerline coordinates of each road segment in the non-standard format high-precision map, and then the width and number of lanes of the road are restored by combining the lane boundary information of each road.

7. A high-precision map format conversion method for intelligent driving simulation testing as described in claim 5, characterized in that, In step S23, the high-precision map needs to be converted from the WGS84 coordinate system to the CarMaker road model coordinate system.

8. A high-precision map format conversion method for intelligent driving simulation testing as described in claim 4, characterized in that, In step S31, the reconstruction of traffic lights and road markings requires obtaining their location and attribute information: first, the lane IDs bound to the traffic lights and road markings are obtained, and then their accurate location is obtained based on their lane-based position offset information. Finally, obtain the attribute information of traffic lights and road markings, including the type of traffic lights and the type of road markings.

Images