A method and device for updating all elements of a high-precision map based on newly added road scenes

Abstract

The present invention relates to a method and device for updating all elements of a high-precision map based on a newly added road scene. The method comprises: obtaining differential 2D pixel coordinates of the newly added road scene, collecting camera intrinsic and extrinsic parameters, collecting trajectory and position information, and point cloud data, and determining the interval position information of the newly added road on the trajectory line based on the obtained information; cropping and rotating the point cloud based on element classification, a first preset resolution, and the interval position information, and projecting the rotated point cloud, and combining it with a deep learning model to ensure that the projected image covers the entire width direction; segmenting the widened projected image to obtain the two-dimensional outline and pixel coordinates of each element, and then performing inverse calculation on the image and vectorizing each element based on the inverse calculation results. The present invention automatically extracts elements of the newly added road based on the a priori differential 2D information, greatly improving update efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-precision map production, and specifically relates to a method and device for updating all elements of a high-precision map based on a newly added road scene. Background Art

[0002] Against the backdrop of the rapid development of autonomous driving technology, high-precision maps for autonomous driving have emerged. They play an indispensable role in ADAS safety, playing a crucial role in vehicle-side positioning, decision-making, and real-world simulation verification. However, for high-precision maps to maximize their value, they must be sufficiently up-to-date, which places high demands on the frequency of map updates. In fact, for any map vendor to achieve competitive map products, in addition to ensuring that their products meet customer needs, another key factor in commercialization is to maintain quality and quantity while reducing costs and increasing efficiency. This requires the maximum possible automation of map production and updates. Compared to the production of full-feature HD map basemaps, local single-element map updates are characterized by fragmentation, relatively simple elements, and small data volumes. To address these characteristics, in order to meet the high timeliness requirements of map updates, an automated map update solution, distinct from full-feature map production, is necessary. Summary of the Invention

[0003] In order to improve the timeliness and efficiency of updating a single element of a high-precision map, a first aspect of the present invention provides a method for updating all elements of a high-precision map based on a newly added road scene, comprising: obtaining the difference 2D pixel coordinates of the newly added road scene, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and point cloud data; determining the interval position information of the newly added road on the trajectory line based on the difference 2D pixel coordinates, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information and the point cloud data; cropping and rotating the point cloud based on the element classification, the first preset resolution and the interval position information, and projecting the rotated point cloud to obtain a The method comprises the following steps: projecting images of multiple perspectives of one or more segments of point cloud data; identifying and extracting one or more newly added road images from the projected images of multiple perspectives based on a deep learning model, and calculating the width change information of each newly added road; widening the projected images of the multiple perspectives according to the width change information and trajectory points of each newly added road, so that the projected images of the multiple perspectives cover the road area in all width directions; segmenting the widened projected images of the multiple perspectives to obtain the two-dimensional outline and pixel coordinates of each element; inversely calculating the two-dimensional outline of each element and its pixel coordinates into the world coordinate system, and vectorizing each element according to the inverse calculation result.

[0004] In some embodiments of the present invention, determining the interval position information of the newly added roads on the trajectory line based on the difference 2D pixel coordinates, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information and point cloud data includes: based on the internal and external parameters of the acquisition camera, through the PNP perspective transformation principle, the prior difference 2D pixel coordinates of the newly added road scene are reversed into the world coordinate system to obtain the coordinates of each newly added road in the world coordinate system; according to the coordinates of each newly added road in the world coordinate system, determining the interval range of each newly added road on the acquisition trajectory line corresponding to the acquisition camera.

[0005] In some embodiments of the present invention, the point cloud is cropped and rotated based on the feature classification, the first preset resolution and the interval position information, and the rotated point cloud is projected to obtain projection images of multiple perspectives of one or more segments of point cloud data, including: according to the interval position information, cropping one or more segments of point cloud containing newly added roads from the point cloud; according to the trajectory and posture information corresponding to each segment of point cloud, the point cloud is rotated to a preset direction; based on the first preset resolution and feature classification, each segment of the rotated point cloud is projected to obtain projection images of multiple perspectives of one or more segments of point cloud data.

[0006] Furthermore, projecting each segment of the rotated point cloud to obtain projection images of multiple perspectives of one or more segments of point cloud data includes: projecting each segment of the rotated point cloud into a front view, a top view, and a side view according to the reflection intensity of the point.

[0007] Furthermore, the projected images of the multiple perspectives are widened according to the width change information and trajectory points of each newly added road so that the projected images of the multiple perspectives cover the road area in the entire width direction, including: dividing each point cloud segment into blocks with the trajectory point as the center, and projecting each blocked point cloud based on the second preset resolution and feature classification to obtain the projected image of each point cloud block; based on the width change information of each newly added road, judging whether the projected image of each point cloud block completely covers the road area in the entire width direction.

[0008] Furthermore, the projected images of the multiple perspectives are widened according to the width change information and trajectory points of each newly added road so that the projected images of the multiple perspectives cover the road area in the entire width direction, including: dividing each point cloud segment into blocks with the trajectory point as the center, and projecting each blocked point cloud based on the second preset resolution and feature classification to obtain the projected image of each point cloud block; based on the width change information of each newly added road, judging whether the projected image of each point cloud block completely covers the road area in the entire width direction.

[0009] In the above embodiment, the back-calculation of the two-dimensional contour of each element and its pixel coordinates into the world coordinate system includes: based on the depth information and trajectory posture information corresponding to the two-dimensional contour of each element, using the pnp perspective transformation principle, the two-dimensional contour of each element and its pixel coordinates are back-calculated into the world coordinate system.

[0010] The second aspect of the present invention provides a high-precision map full-element update device based on a newly added road scene, including: an acquisition module for acquiring the difference 2D pixel coordinates of the newly added road scene, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and point cloud data; determining the interval position information of the newly added road on the trajectory line based on the difference 2D pixel coordinates, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information and the point cloud data; a projection module for cropping and rotating the point cloud based on element classification, a first preset resolution and the interval position information, and projecting the rotated point cloud to obtain multiple perspectives of one or more segments of point cloud data Projection image; a widening module, which is used to identify and extract one or more new road images from the projection images of multiple perspectives based on a deep learning model, and calculate the change information of the width of each new road; according to the change information and trajectory points of the width of each new road, the projection images of the multiple perspectives are widened so that the projection images of the multiple perspectives cover the road area in all width directions; a segmentation module, which is used to segment the widened projection images of the multiple perspectives to obtain the two-dimensional outline and pixel coordinates of each element; an inverse calculation module, which is used to inversely calculate the two-dimensional outline and pixel coordinates of each element into the world coordinate system, and vectorize each element according to the inverse calculation result.

[0011] The third aspect of the present invention provides an electronic device comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for updating all elements of a high-precision map of a newly added road scene provided in the first aspect of the present invention.

[0012] A fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for updating all elements of a high-precision map of a newly added road scene provided in the first aspect of the present invention is implemented.

[0013] The beneficial effects of the present invention are:

[0014] This invention builds on the foundation of an automated, full-element basemap production solution with customized optimizations. Instead of indiscriminately processing the entire collection, this approach incorporates 2D difference information and 3D reconstruction technology to pinpoint the location to be updated, meeting the requirements for efficient single-element updates. Compared to purely manual map updates, this significantly improves efficiency. Compared to full-element production solutions, this solution precisely locates the elements to be updated, significantly reducing redundant data. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of the basic flow of a method for updating all elements of a high-precision map for a newly added road scene in some embodiments of the present invention;

[0016] Figure 2 A schematic diagram of a specific process of a method for updating all elements of a high-precision map for a newly added road scene in some embodiments of the present invention;

[0017] Figure 3 This is a schematic diagram of the structure of a device for updating all elements of a high-precision map with a newly added road scene in some embodiments of the present invention;

[0018] Figure 4 Schematic diagram of the structure of an electronic device in some embodiments of the present invention. DETAILED DESCRIPTION

[0019] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0020] refer to Figure 1 and Figure 2In a first aspect of the present invention, a method for updating all elements of a high-precision map based on a newly added road scene is provided, comprising: S100. Obtaining the difference 2D pixel coordinates of the newly added road scene, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and the point cloud data; determining the interval position information of the newly added road on the trajectory line according to the difference 2D pixel coordinates, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and the point cloud data; S200. Based on the element classification, the first preset resolution, and the interval position information, the point cloud is cropped and rotated, and the rotated point cloud is projected to obtain multiple views of one or more segments of point cloud data. Based on the deep learning model, one or more newly added road images are identified and extracted from the projection images of multiple perspectives, and the width change information of each newly added road is calculated; according to the width change information and trajectory points of each newly added road, the projection images of the multiple perspectives are widened so that the projection images of the multiple perspectives cover the road area in all width directions; S400. The widened projection images of the multiple perspectives are segmented to obtain the two-dimensional outline and pixel coordinates of each element; S500. The two-dimensional outline of each element and its pixel coordinates are inversely calculated into the world coordinate system, and each element is vectorized according to the inverse calculation result.

[0021] In step S100 of some embodiments of the present invention, the difference 2D pixel coordinates of the newly added road scene, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its posture information, and the point cloud data are obtained; based on the difference 2D pixel coordinates, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its posture information and the point cloud data, the interval position information of the newly added road on the trajectory line is determined.

[0022] It is understood that the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory, and its pose information in the present disclosure are acquired based on the acquisition vehicle or the acquisition device of the acquisition vehicle. Without loss of generality, in order to ensure the synchronization and correspondence of the differential 2D pixel coordinates of the newly added road scene, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its pose information, and the point cloud data, the acquisition camera, trajectory acquisition device, and lidar are deployed or installed on the same acquisition vehicle. The pose information includes but is not limited to longitude and latitude, heading angle, roll angle, pitch angle, etc. A newly added road should be understood as a road that has changed from the same road or the same section of road captured by the previous acquisition vehicle; for example, new roads, forks, lane changes, direction changes, destruction, and other road changes on the original road. Point cloud data includes the spatial position coordinates of the point cloud and its corresponding image information (pixel values ​​and pixel coordinates). The differential 2D pixel coordinates refer to the pixel coordinates of the starting point of the newly added road section in image 1 (the starting point image of the road) and the pixel coordinates of the end point in image 2 (the end point image of the road).

[0023] In some embodiments of the present invention, step S200, determining the interval position information of the newly added road on the trajectory line based on the difference 2D pixel coordinates, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its pose information, and the point cloud data, includes:

[0024] S201. Based on the intrinsic and extrinsic parameters of the acquisition camera, the priori difference 2D pixel coordinates of the newly added road scene are back-calculated into the world coordinate system through the Pnp (Perspective-n-Point) perspective transformation principle to obtain the coordinates of each newly added road in the world coordinate system. Specifically, based on the Pnp (Perspective-n-Point) perspective transformation principle, the camera intrinsic parameters and the extrinsic parameters between the camera and the lidar are used to back-calculate the 2D pixel coordinates in step S100 into WGS84 world coordinates. S202. Based on the coordinates of each newly added road in the world coordinate system, the range of each newly added road on the acquisition trajectory corresponding to the acquisition camera is determined.

[0025] In step S200 of some embodiments of the present invention, the cropping and rotating the point cloud based on the feature classification, the first preset resolution, and the interval position information, and projecting the rotated point cloud to obtain projection images of multiple perspectives of one or more segments of point cloud data includes:

[0026] S201. Based on the interval position information, crop one or more point cloud segments containing the newly added road from the point cloud;

[0027] S202. Rotate the point cloud to a preset direction based on the trajectory and posture information corresponding to each point cloud segment; specifically, based on the point cloud data containing several point cloud segments obtained in step S201 and the trajectory POS information corresponding to each point cloud segment, rotate the point cloud segment data so that the road direction is due north; optionally, the preset direction can be adjusted according to actual needs.

[0028] S203. Based on the first preset resolution and feature classification, each segment of the rotated point cloud is projected to obtain projection images of one or more segments of point cloud data from multiple perspectives. Furthermore, in step S203, projecting each segment of the rotated point cloud to obtain projection images of one or more segments of point cloud data from multiple perspectives includes projecting each segment of the rotated point cloud into a front view, a top view, and a side view based on the reflection intensity of the points. Specifically, the point cloud segment data rotated in step S202 is projected into a front view, a top view, and a side view based on the reflection intensity values ​​of the points according to the resolution R1 and feature classification. These images can serve as input or training samples for a deep learning-based road recognition model.

[0029] Furthermore, in step S300 of some embodiments of the present invention, widening the projection images of the multiple perspectives based on the width change information and trajectory points of each newly added road so that the projection images of the multiple perspectives cover the entire road area in the width direction includes:

[0030] S301. Divide each point cloud segment into blocks with the trajectory point as the center, and project each divided point cloud based on the second preset resolution and element classification to obtain a projected image of each point cloud block;

[0031] S302. Based on the change information of the width of each newly added road, determine whether the projected image of each point cloud block completely covers the road area in the entire width direction.

[0032] In step S302 of some embodiments of the present invention, further, widening the projection images of the multiple perspectives based on the width change information of each newly added road and the trajectory point so that the projection images of the multiple perspectives cover the entire road area in the width direction includes: S3021. dividing each point cloud segment into blocks centered on the trajectory point, and projecting each divided point cloud based on a second preset resolution and feature classification to obtain a projection image of each point cloud block;

[0033] S3022. Based on the width change information for each newly added road, determine whether the projected image of each point cloud block completely covers the entire road area along the entire width. Specifically, the point cloud segment is divided into blocks centered on the trajectory point, and projection is performed similarly to S200 at resolution R2. Then, referring to the road width information, determine whether the projected image of the point cloud block centered on the trajectory point can cover the critical area. If not, the trajectory center is shifted left or right to increase the projected area.

[0034] In step S400 of some embodiments of the present invention, the widened projection images of multiple perspectives are segmented to obtain the two-dimensional contour and pixel coordinates of each element. Specifically, a deep learning network model is trained for each element, and the projection image generated in step S300 is subjected to image segmentation processing to obtain a 2D target result for each element; the 2D target result includes the two-dimensional contour and pixel coordinates of each element (the pixel coordinates of each key point). It can be understood that the two-dimensional contour is usually outlined by geometric lines connecting one or more vertices, corner points, and key points. Therefore, the pixel coordinates of the two-dimensional contour correspond to the pixel coordinates of the one or more vertices, corner points, and key points mentioned above.

[0035] It can be understood that elements generally include one or more elements of a lane model, a road model, and a road marking model. For example, a lane model includes lane boundaries, lane width, lane type, etc.; a road model includes road centerlines, road topology, tunnels, toll booths, curvature, and slope. A road marking model includes road signs such as text, arrows, diversion areas, and road limit signs.

[0036] In step S500 of the above-described embodiment, backcalculating the 2D outline of each feature and its pixel coordinates into the world coordinate system includes: utilizing the pnp perspective transformation principle based on the depth information and trajectory pose information corresponding to each feature's 2D outline, backcalculating the 2D outline of each feature and its pixel coordinates into the world coordinate system. Specifically, based on the depth information and trajectory pos information saved during projection, the 2D result from step S400 is backcalculated back into the WGS84 world coordinate system using the pnp perspective transformation principle. Then, according to the feature's production standard, the current state or key points corresponding to each feature are vectorized.

[0037] Finally, the vectorized data of all elements obtained in step S500 is verified and abnormal values ​​are corrected; the correct vectorized results are written into the database to complete the output.

[0038] Example 2

[0039] refer to Figure 3 In a second aspect of the present invention, a device 1 for updating all elements of a high-precision map based on a newly added road scene is provided, comprising: an acquisition module 11 for acquiring the difference 2D pixel coordinates of the newly added road scene, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and point cloud data; determining the interval position information of the newly added road on the trajectory line according to the difference 2D pixel coordinates, the internal and external parameters of the acquisition camera, the acquisition trajectory and its posture information, and the point cloud data; a projection module 12 for cropping and rotating the point cloud based on element classification, a first preset resolution, and the interval position information, and projecting the rotated point cloud to obtain multiple perspectives of one or more segments of point cloud data. Projection image; a widening module 13, which is used to identify and extract one or more newly added road images from the projection images of multiple perspectives based on a deep learning model, and calculate the change information of the width of each newly added road; according to the change information and trajectory points of the width of each newly added road, the projection images of the multiple perspectives are widened so that the projection images of the multiple perspectives cover the road area in all width directions; a segmentation module 14, which is used to segment the widened projection images of the multiple perspectives to obtain the two-dimensional contour and pixel coordinates of each element; an inverse calculation module 15, which is used to inversely calculate the two-dimensional contour and pixel coordinates of each element into the world coordinate system, and vectorize each element according to the inverse calculation result.

[0040] Furthermore, the projection module 12 includes: a cropping unit for cropping one or more segments of point cloud containing the newly added road from the point cloud according to the interval position information; a rotation unit for rotating the point cloud to a preset direction according to the trajectory and posture information corresponding to each segment of the point cloud; a projection unit for projecting each rotated segment of the point cloud based on a first preset resolution and feature classification to obtain projection images of multiple perspectives of one or more segments of point cloud data.

[0041] Example 3

[0042] refer to Figure 4 According to a third aspect of the present invention, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the high-precision map full-element update method based on the newly added road scene according to the first aspect of the present invention.

[0043] The electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the electronic device 500 are also stored in the RAM 503. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0044] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Figure 4 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead. Figure 4 Each block shown in the figure may represent one device, or may represent multiple devices as needed.

[0045] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed. It should be noted that the computer-readable medium described in the embodiment of the present disclosure can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, an electromagnetic signal, an optical signal, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wire, optical cable, RF (radio frequency), etc., or any suitable combination thereof.

[0046] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device. The computer-readable medium carries one or more computer programs, which, when executed by the electronic device, cause the electronic device to:

[0047] Computer program code for performing operations of embodiments of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, Python, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be written entirely on the user's computer.

[0048] The program may be executed on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server.

[0049] In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, through the Internet using an Internet service provider).

[0050] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure.

[0051] Each box in the block diagram may represent a module, a program segment, or a portion of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also be

[0052] For example, two blocks shown in succession may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved.

[0053] It should be noted that each block in the block diagrams and / or flow charts, and combinations of blocks in the block diagrams and / or flow charts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0054] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for updating all elements of a high-precision map based on a newly added road scene, characterized in that: include: Obtain the difference 2D pixel coordinates of the newly added road scene, collect the camera's intrinsic and extrinsic parameters, collect the trajectory and its pose information, and point cloud data; Determine the interval position information of the newly added roads on the trajectory line based on the difference 2D pixel coordinates, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its pose information, and the point cloud data: Based on the intrinsic and extrinsic parameters of the acquisition camera, use the PNP perspective transformation principle to reversely calculate the a priori difference 2D pixel coordinates of the newly added road scene into the world coordinate system to obtain the coordinates of each newly added road in the world coordinate system; determine the interval range of each newly added road on the acquisition trajectory line corresponding to the acquisition camera based on the coordinates of each newly added road in the world coordinate system; Based on the feature classification, the first preset resolution, and the interval position information, the point cloud is cropped and rotated, and the rotated point cloud is projected to obtain projection images of multiple perspectives of one or more segments of point cloud data; Based on a deep learning model, the system identifies and extracts one or more newly added road images from the projected images from multiple perspectives, and calculates the width change information of each newly added road. Based on the width change information and trajectory points of each newly added road, the projected images from the multiple perspectives are widened so that the projected images from the multiple perspectives cover the entire road area in the width direction. Segment the widened projection images of multiple perspectives to obtain the two-dimensional outline and pixel coordinates of each element; The two-dimensional outline of each feature and its pixel coordinates are inversely calculated into the world coordinate system, and each feature is vectorized based on the inverse calculation results.

2. The method for updating all elements of a high-precision map based on a newly added road scene according to claim 1 is characterized in that: The step of cropping and rotating the point cloud based on the feature classification, the first preset resolution, and the interval position information, and projecting the rotated point cloud to obtain projection images of multiple perspectives of one or more segments of point cloud data includes: According to the interval position information, cutting out one or more segments of point cloud containing the newly added road from the point cloud; According to the trajectory and pose information corresponding to each point cloud segment, the point cloud is rotated to a preset direction; Based on the first preset resolution and feature classification, each segment of the rotated point cloud is projected to obtain projection images of multiple perspectives of one or more segments of point cloud data.

3. The method for updating all elements of a high-precision map based on a newly added road scene according to claim 2 is characterized in that: The step of projecting each segment of the rotated point cloud to obtain projection images of one or more segments of point cloud data from multiple perspectives includes: According to the reflection intensity of the point, each segment of the rotated point cloud is projected into front view, top view and side view.

4. The method for updating all elements of a high-precision map based on a newly added road scene according to claim 2 is characterized in that: The step of widening the projection images of the multiple perspectives according to the width change information and trajectory points of each newly added road, so that the projection images of the multiple perspectives cover the entire road area in the width direction, includes: Divide each point cloud segment into blocks centered on the trajectory point, and project each divided point cloud based on the second preset resolution and feature classification to obtain a projected image of each point cloud block; Based on the change information of the width of each newly added road, it is determined whether the projected image of each point cloud block completely covers the road area in all width directions.

5. The method for updating all elements of a high-precision map based on a newly added road scene according to any one of claims 1 to 4, characterized in that: The back-calculation of the two-dimensional outline of each element and its pixel coordinates into the world coordinate system includes: According to the depth information and trajectory pose information corresponding to the two-dimensional outline of each element, the two-dimensional outline of each element and its pixel coordinates are back-calculated to the world coordinate system using the PNP perspective transformation principle.

6. A device for updating all elements of a high-precision map based on a newly added road scene, characterized in that: include: An acquisition module is configured to obtain the difference 2D pixel coordinates of the newly added road scene, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its position and posture information, and point cloud data; determine the interval position information of the newly added roads on the trajectory line based on the difference 2D pixel coordinates, the intrinsic and extrinsic parameters of the acquisition camera, the acquisition trajectory and its position and posture information, and the point cloud data: based on the intrinsic and extrinsic parameters of the acquisition camera, and through the PNP perspective transformation principle, the prior difference 2D pixel coordinates of the newly added road scene are reversely calculated into the world coordinate system to obtain the coordinates of each newly added road in the world coordinate system; based on the coordinates of each newly added road in the world coordinate system, determine the interval range of each newly added road on the acquisition trajectory line corresponding to the acquisition camera; a projection module, configured to crop and rotate the point cloud based on feature classification, the first preset resolution, and the interval position information, and project the rotated point cloud to obtain projection images of multiple perspectives of one or more segments of point cloud data; A widening module is configured to identify and extract one or more newly added road images from the projected images from multiple perspectives based on a deep learning model, and calculate the width change information of each newly added road; and widen the projected images from the multiple perspectives based on the width change information and trajectory points of each newly added road so that the projected images from the multiple perspectives cover the entire road area in the width direction; A segmentation module is used to segment the widened projection images of multiple perspectives to obtain the two-dimensional contour and pixel coordinates of each element; The inverse calculation module is used to inversely calculate the two-dimensional outline of each feature and its pixel coordinates into the world coordinate system, and vectorize each feature based on the inverse calculation results.

7. The high-precision map single element updating device based on a newly added road scene according to claim 6 is characterized in that: The projection module includes: a clipping unit, configured to clip one or more segments of point cloud containing the newly added road from the point cloud according to the interval position information; A rotation unit, configured to rotate the point cloud to a preset direction according to the trajectory and posture information corresponding to each segment of the point cloud; The projection unit is used to project each segment of the rotated point cloud based on the first preset resolution and feature classification to obtain projection images of multiple perspectives of one or more segments of point cloud data.

8. An electronic device comprising: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the high-precision map full-element update method based on the newly added road scene as described in any one of claims 1 to 5.

9. A computer-readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method for updating all elements of a high-precision map based on a newly added road scene as described in any one of claims 1 to 5 is implemented.

Images