Driving map and parking map fusion method and system
By calculating and integrating the error matrix and position error in driving maps and parking maps, the optimal positioning is formed, and the problem of map sharing and data mismatch in the existing technology is solved, and the map data sharing of driving function and parking function is realized , improves the accuracy and stability of the map in the parking lot environment.
Patent Information
- Application Number
- CN202411208703.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-13
AI Technical Summary
In the prior art, there are major differences in the elements, protocols, etc. of the map used in driving functions and parking functions, which leads to the inability to share the same map. When switching functions, data mismatch will occur, and in the parking lot. Vehicles change rapidly in the environment, and point cloud characteristics deteriorate significantly.
By extracting the lane lines and signs in the driving map and parking map, the error matrix and position error are calculated to form an optimized position, and the adjusted parking map and driving map are integrated to achieve the sharing of the map.
The map data sharing of driving function and parking function is realized, avoiding the problem of data mismatch during function switching, and improving the map accuracy and stability in the parking lot environment.
Smart Images
Figure CN119124134A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of electronic maps and automatic driving, and in particular relates to a method and system for fusing a driving map with a parking map. Background Art
[0002] Under current technical conditions, the maps used by the driving function and the parking function have significant differences in elements, broadcast protocols, etc., so the two functions require two sets of maps with different standards.
[0003] In some existing technologies, the driving function uses high-precision maps and the parking function is abandoned. The parking function uses the parking lot map. When switching between the driving scene and the parking scene, the corresponding function must be stopped first, and then a new map is retrieved from the system and the function is run again. The two functions cannot share a map, and data mismatches will occur during the function switching process. There will be a sense of disconnection between function switching, and the function switching interval needs to be reinitialized.
[0004] In other existing technologies, driving and parking functions use the point cloud feature map established by laser SLAM. However, the laser point cloud feature map has a large amount of data and cannot store a large range of feature maps. In addition, in a parking environment, vehicles change quickly and the point cloud features degrade significantly. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide a method and system for fusing a driving map with a parking map, so that the maps of the driving function and the parking function can be shared.
[0006] The technical solution adopted by the present invention to solve the above technical problems is: a method for fusing a driving map with a parking map,
[0007] Extract all lane lines and signs before and after the parking lot entrance in the driving map and parking map respectively;
[0008] Taking the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line, calculate the lane line error matrix, and retain the lateral error through the lane line heading;
[0009] Taking the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, calculate the sign position error, and retain the longitudinal error through the lane line heading;
[0010] Integrate the lateral error and the longitudinal error to form an optimal posture, and complete the initial position adjustment of the parking map through the optimal posture;
[0011] The adjusted parking map is integrated with the driving map.
[0012] According to the above scheme, the positional relationship between the driving map and the parking map is obtained by comparing the positional relationship between the driving map and the parking map when the map collection vehicle is located at the same parking lot entrance during the construction of the driving map and the parking map.
[0013] According to the above scheme, the driving map is produced using high-precision map specifications, and the map elements collected by the map collection vehicle include lane lines and signboard information; map elements are collected at the parking lot entrance that needs to be integrated with the parking map, and the parking lot entrance is added to the map data as POI information, and the position of the map collection vehicle when it is located at the parking lot entrance is recorded.
[0014] According to the above scheme, the parking map is constructed by a parking lot map point cloud, and the parking lot map point cloud is completed by a map collection vehicle equipped with laser SLAM; the position of the map collection vehicle when it is located at the entrance of the parking lot is recorded.
[0015] According to the above scheme, lane line points and sign points in the parking lot map point cloud are extracted;
[0016] Through the positional relationship between the driving map and the parking map, the coordinates of the lane line points and the signboard points are unified with the driving map.
[0017] According to the above solution, the lane line type point is also accompanied by lane line type point attributes, and the lane line type point attributes include the color and type corresponding to the lane line type point.
[0018] According to the above scheme, the identification card type point is also accompanied by identification card type point attributes, and the identification card type point attributes include the type and enumeration value corresponding to the identification card type point.
[0019] According to the above scheme, the lane line error matrix and the sign position error are calculated using the ICP matching algorithm respectively.
[0020] According to the above scheme, the fused map is broadcast externally through the ADASIS v3 protocol.
[0021] A driving map and parking map fusion system, comprising:
[0022] The lane line and signboard extraction module is used to extract all lane lines and signboards before and after the parking lot entrance in the driving map and parking map respectively;
[0023] The lateral error calculation module is used to take the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line, calculate the lane line error matrix, and retain the lateral error through the lane line heading;
[0024] The longitudinal error calculation module is used to take the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, calculate the sign position error, and retain the longitudinal error through the lane line heading;
[0025] A parking map initial position adjustment module is used to integrate the lateral error and the longitudinal error to form an optimized posture, and complete the initial position adjustment of the parking map through the optimized posture;
[0026] The map fusion module is used to integrate the adjusted parking map with the driving map.
[0027] The beneficial effects of the present invention are as follows: by taking the parking lot entrance as a virtual topological relationship, the lane lines and signboards of the driving map and the parking map located at the same parking lot entrance are error matched, and at the same time, the posture relationship between the driving map and the parking map is combined for adjustment and fusion, ultimately achieving the sharing of map data for driving functions and parking functions. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 It is a schematic diagram of the principle of an embodiment of the present invention. DETAILED DESCRIPTION
[0029] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0030] Glossary:
[0031] Driving map: map data that driving functions in intelligent driving systems (such as NOA, ICA, etc.) rely on
[0032] Parking map: map data that parking functions in intelligent driving systems (such as LAPA, FAPA, etc.) rely on
[0033] POI information: short for Point of Interest, is a term in geographic information, referring to all geographic objects that can be abstracted as points.
[0034] ADASIS v3 protocol: ADASIS (Advanced Driver Assistance Systems Interface Specification) v3 protocol is an advanced driver assistance system interface specification used to achieve vehicle-to-infrastructure communication in intelligent transportation systems. This protocol can transmit data such as the vehicle's surrounding environment, road conditions, and road topology maps to the vehicle through the vehicle's built-in sensors, so as to achieve more intelligent and efficient data communication.
[0035] ICP matching algorithm: ICP (Iterative Closest Point) algorithm is an iterative calculation method, mainly used for point cloud matching, which achieves accurate stitching by continuously iterating and minimizing the corresponding points between source data and target data.
[0036] The present invention provides a method and system for fusing a driving map with a parking map, so as to realize the expansion of driving functions in parking scenarios and achieve data reuse.
[0037] Since the data formats of the driving map and the parking map are different, the driving map and the parking map need to be preprocessed separately before fusion.
[0038] When designing the driving map, the driving map is produced using high-precision map specifications, and a map collection vehicle is used to collect map elements, including lane lines and signboards, to form map data. Map elements are collected for the parking lot entrance that needs to switch between driving and parking functions (i.e., the parking lot entrance that needs to be integrated with the parking map), and the parking lot entrance is added to the map data as POI information. At this time, the end point of the map collection vehicle is the parking lot entrance, and the position T1 at this location is recorded.
[0039] When designing the parking map, a map collection vehicle equipped with laser SLAM starts from the entrance of the parking lot, records the initial position T2 there, and then uses laser SLAM to complete the construction of the parking lot map point cloud.
[0040] Calculate the position relationship T between the driving map and the parking map r :
[0041]
[0042] Extract the lane line point, sign point and parking space point data in the parking lot map point cloud, and use the position relationship T between the driving map and the parking map r , respectively convert the lane line point, sign point and parking space point data into the global coordinate system, so as to unify them with the coordinates in the driving map, which is convenient for subsequent calculations.
[0043] Save the color, type and other attributes corresponding to the lane line point as the additional attributes of the lane line point; save the sign type, enumeration value and other attributes as the additional attributes of the sign point. Save the parking space as POI information, number it according to the parking space number, as the additional parking space attribute of the parking space point data. After saving, the lane line point, sign point and parking space point data have additional attributes in addition to the coordinates in the global coordinate system, making the expression of each point richer.
[0044] After the above preprocessing is completed, the driving map and the parking map are fused, such as Figure 1 As shown, the specific steps are as follows:
[0045] S1. Extract all lane lines and signs before and after the parking lot entrance in the driving map and parking map respectively. Specifically, the mapping vehicle can be used to obtain the driving map and parking map at the entrance of the parking lot at the same time, and then extract all lane lines and signs before and after the entrance of the parking lot respectively. Since the preprocessing has been done, the coordinates of the extracted lane lines and signs are all in the global coordinate system. Since the lane lines and signs at the entrance of the parking lot are the same for the driving map and the parking map, the comparison of all lane lines and signs is used for verification, so as to splice and fuse the two maps. The mapping vehicle needs to do this at the entrance of the parking lot during the first fusion. After the map is fused, it can be directly loaded into the vehicle for direct use.
[0046] In this embodiment, the lane line set in the driving map is recorded as L1, and each lane line point set is recorded as l 1x , x is the number corresponding to the lane line.
[0047] L1=l 11 +l 12 +l 13 +……
[0048] The set of signboards in the driving map is denoted as S1, and each signboard is denoted as s 1y , y is the number corresponding to the sign, and the coordinates of the center point of the sign are recorded.
[0049] S1=s 11 +s 12 +s 13 +……
[0050] The lane line set in the parking map is denoted as L2, and each lane line point set is denoted as l 2x , x is the number corresponding to the lane line.
[0051] L2=l 21 +l 22 +l 23 +……
[0052] The set of signboards in the parking map is denoted as S2, and each signboard is denoted as s 2y , y is the number corresponding to the sign, and the coordinates of the center point of the sign are recorded.
[0053] S2=S21+S22+S23+…
[0054] S2, taking the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line and calculate the lane line error matrix R through the ICP matching algorithm L ,t L , and retain the lateral error R* through the lane heading L , t* L :
[0055]
[0056] Among them, the ICP matching algorithm is a known mathematical algorithm that calculates the errors of points corresponding to the source point set and the target point set, and continuously iterates to minimize the sum of the errors. The present invention applies it to the field to calculate the lane line error matrix, and the details are not repeated here.
[0057] S3, similarly, take the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, and calculate the sign position error R through the ICP matching algorithm. S ,t S , and retain the longitudinal error R* through the lane heading S , t* S .
[0058] S4. Integration of the lateral error R* L , t* L and longitudinal error R* S , t* S , forming the optimal posture R, t, and completing the initial position adjustment of the parking map through the optimal posture.
[0059]
[0060]
[0061] S5. Integrate the adjusted parking map with the driving map to form unified map data, and broadcast it externally according to the ADASIS v3 protocol.
[0062] The main body of a set of map data after fusion by the present invention is stored as public road elements, designed with the ADASIS v3 protocol, and the parking map elements are set as POIs through the parking lot entrances and exits as virtual topological relationships. Certain preprocessing is performed when designing the driving map and the parking map, so that the same lane lines and signs can be used for verification at the joint of the two maps, and finally the map data sharing of the driving function and the parking function is realized.
[0063] As a second aspect of the present invention, based on the same inventive principle and concept, the present invention further provides a driving map and parking map fusion system, comprising:
[0064] The lane line and signboard extraction module is used to extract all lane lines and signboards before and after the parking lot entrance of the driving map and the parking map respectively;
[0065] The lateral error calculation module is used to take the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line, calculate the lane line error matrix, and retain the lateral error through the lane line heading;
[0066] The longitudinal error calculation module is used to take the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, calculate the sign position error, and retain the longitudinal error through the lane line heading;
[0067] A parking map initial position adjustment module is used to integrate the lateral error and the longitudinal error to form an optimized posture, and complete the initial position adjustment of the parking map through the optimized posture;
[0068] The map fusion module is used to integrate the adjusted parking map with the driving map.
[0069] The present invention can be set up on the vehicle side or in the cloud. In the vehicle side environment: the driving map and the parking map can be collected by a collection vehicle, and according to the method of the present invention, they are spliced in the vehicle controller environment to finally form a fused map, and directly deployed on the vehicle. In the cloud environment: the original data (including driving map data, parking map data, vehicle calibration parameters, etc.) collected by the collection vehicle can be uploaded to the cloud platform, and the map is spliced through the cloud server to form a fused map, and then the demand side (i.e., the vehicle that needs to use the map) makes a request, and the established map is sent to the vehicle side through the network for use.
[0070] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.
Claims
1. A method for fusing a driving map with a parking map, characterized in that: Extract all lane lines and signs before and after the parking lot entrance in the driving map and parking map respectively; Taking the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line, calculate the lane line error matrix, and retain the lateral error through the lane line heading; Taking the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, calculate the sign position error, and retain the longitudinal error through the lane line heading; Integrate the lateral error and the longitudinal error to form an optimal posture, and complete the initial position adjustment of the parking map through the optimal posture; The adjusted parking map is integrated with the driving map.
2. The method for fusing a driving map and a parking map according to claim 1, characterized in that: The positional relationship between the driving map and the parking map is obtained by comparing the positional relationship between the driving map and the parking map when the map collection vehicle is located at the same parking lot entrance during the construction of the driving map and the parking map.
3. The method for fusing a driving map and a parking map according to claim 2, characterized in that: The driving map is produced using high-precision map specifications, and the map elements collected by the map collection vehicle include lane lines and signboard information; map elements are collected at the parking lot entrance that needs to be integrated with the parking map, and the parking lot entrance is added to the map data as POI information, and the position of the map collection vehicle when it is located at the parking lot entrance is recorded.
4. The method for fusing a driving map and a parking map according to claim 2, characterized in that: The parking map is constructed by a parking lot map point cloud, and the parking lot map point cloud is completed by a map collection vehicle equipped with laser SLAM; the position of the map collection vehicle when it is located at the entrance of the parking lot is recorded.
5. The method for fusing a driving map and a parking map according to claim 4, characterized in that: Extract lane line points and sign points from the parking lot map point cloud; Through the positional relationship between the driving map and the parking map, the coordinates of the lane line points and the signboard points are unified with the driving map.
6. The method for fusing a driving map and a parking map according to claim 5, characterized in that: The lane line point also has additional attributes, including the color and type corresponding to the lane line point.
7. The method for fusing a driving map and a parking map according to claim 5, characterized in that: The identification card type point also has additional attributes, which include the type and enumeration value corresponding to the identification card type point.
8. The method for fusing a driving map and a parking map according to claim 1, characterized in that: The lane line error matrix and sign position error are calculated using the ICP matching algorithm.
9. The method for fusing a driving map and a parking map according to claim 1, characterized in that: The fused map is broadcasted via the ADASIS v3 protocol.
10. A driving map and parking map fusion system, characterized in that: include: The lane line and signboard extraction module is used to extract all lane lines and signboards before and after the parking lot entrance in the driving map and parking map respectively; The lateral error calculation module is used to take the lane line of the driving map as the target point and the lane line of the parking map as the point to be optimized, traverse each lane line, calculate the lane line error matrix, and retain the lateral error through the lane line heading; The longitudinal error calculation module is used to take the driving map sign as the target point and the parking map sign as the point to be optimized, traverse each sign, calculate the sign position error, and retain the longitudinal error through the lane line heading; A parking map initial position adjustment module is used to integrate the lateral error and the longitudinal error to form an optimized posture, and complete the initial position adjustment of the parking map through the optimized posture; The map fusion module is used to integrate the adjusted parking map with the driving map.