Laser radar position recognition method and device

Abstract

The embodiment of the present invention discloses a method and device for position recognition of a laser radar. A method for position recognition of a laser radar includes: creating a virtual laser radar corresponding to a target laser radar; searching for scanning points of the virtual laser radar on a map according to the parameter information of the virtual laser radar, and constructing a virtual frame according to the scanning points, wherein the map is constructed according to the data collected by the designated laser radar, and the designated laser radar and the target laser radar are heterogeneous devices; obtaining the query frame of the target laser radar; selecting the target frame from the virtual frame according to the query frame; and identifying the position information of the target frame as the position information of the query frame. The technical solution of the embodiment of the present invention solves the technical problem that the existing high-precision maps from other heterogeneous laser radars cannot be used for positioning, and achieves the effect of making the existing laser radar map applicable to the positioning of other heterogeneous laser radars.

Description

Technical Field

[0001] The embodiments of the present invention relate to position recognition technology, and in particular to a position recognition method and device for a laser radar. Background Art

[0002] With the rapid development of intelligent transportation, the requirements for the accuracy and stability of location recognition technology are increasing. Multi-line LiDAR can quickly and accurately obtain three-dimensional depth information around the radar, and is not affected by changes in ambient lighting, making it widely used in location recognition.

[0003] Currently, location recognition based on multi-line lidar often relies on estimating the similarity between a query frame and candidate frames in a high-precision map to determine the multi-line lidar's location. However, this method requires that the query frame and candidate frames come from lidars with the same structure, which makes some existing high-precision maps from heterogeneous lidars incapable of localization. Summary of the Invention

[0004] The present invention provides a laser radar position recognition method and device, so as to achieve the effect that the existing laser radar map can be applied to the positioning of other heterogeneous laser radars.

[0005] In a first aspect, an embodiment of the present invention provides a position recognition method of a laser radar, comprising:

[0006] Create a virtual lidar corresponding to the target lidar;

[0007] Find the virtual LiDAR's scanning points on the map based on its parameter information, and build a virtual frame based on the scanning points. The map is built based on the data collected by the specified LiDAR. The specified LiDAR and the target LiDAR are heterogeneous devices.

[0008] Get the target lidar query frame;

[0009] Selecting a target frame from the virtual frames according to the query frame;

[0010] The position information of the target frame is identified as the position information of the query frame.

[0011] In a second aspect, an embodiment of the present invention further provides a position recognition device for a laser radar, comprising:

[0012] Virtual LiDAR creation module, used to create a virtual LiDAR corresponding to the target LiDAR;

[0013] The virtual frame construction module is used to find the scanning points of the virtual lidar on the map based on the parameter information of the virtual lidar and construct a virtual frame based on the scanning points. The map is constructed based on the data collected by the specified lidar. The specified lidar and the target lidar are heterogeneous devices.

[0014] A query frame acquisition module is used to obtain the query frame of the target laser radar;

[0015] A target frame selection module is used to select a target frame from a virtual frame according to a query frame;

[0016] The position information recognition module is used to recognize the position information of the target frame as the position information of the query frame.

[0017] In a third aspect, an embodiment of the present invention provides an electronic device, the electronic device including:

[0018] one or more processors;

[0019] a memory for storing one or more programs,

[0020] When one or more programs are executed by one or more processors, the one or more processors implement the laser radar position recognition method as described in any of the embodiments.

[0021] In a fourth aspect, an embodiment of the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to execute a laser radar position recognition method as described in any one of the embodiments.

[0022] The technical solution of this embodiment is to create a virtual laser radar corresponding to the target laser radar; search for the scanning points of the virtual laser radar on the map according to the parameter information of the virtual laser radar, and construct a virtual frame according to the scanning points. The map is constructed according to the data collected by the specified laser radar, and the specified laser radar and the target laser radar are heterogeneous devices; obtain the query frame of the target laser radar; select the target frame from the virtual frame according to the query frame; identify the position information of the target frame as the position information of the query frame. That is to say, by creating a virtual laser radar and constructing a virtual frame corresponding to the virtual laser radar, the technical problem that the laser radar position recognition requires the query frame and the candidate frame to come from the laser radar of the same structure, which makes the existing high-precision maps from other heterogeneous laser radars unable to be used for positioning is solved, and the existing laser radar map can be used for positioning of other heterogeneous laser radars. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flowchart of a laser radar position recognition method provided in Example 1 of the present invention;

[0024] Figure 2 This is a flowchart of a laser radar position recognition method provided in the second embodiment of the present invention;

[0025] Figure 3 This is a structural diagram of a laser radar position recognition device provided in Example 3 of the present invention;

[0026] Figure 4 This is a structural diagram of an electronic device provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0027] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0028] LiDAR, also known as a laser sensor, is a new type of measuring instrument that utilizes laser technology for measurement, enabling contactless, long-distance measurement. LiDAR operates by emitting a laser pulse from a laser diode aimed at a target. After reflection from the target, the laser light scatters in all directions, with some of the scattered light returning to the sensor receiver, forming a cloud of points. LiDAR can be categorized as either single-line or multi-line. Single-line LiDAR emits a single laser beam for measurement, while multi-line LiDAR uses multiple laser beams simultaneously. Therefore, multi-line LiDAR offers greater accuracy than single-line LiDAR.

[0029] LiDAR positioning technology is an important supplement to satellite positioning technology. In short, the core idea of LiDAR positioning technology is: assuming positioning within area A, LiDAR B, as the designated LiDAR, first moves within area A and captures a point cloud image at each position. In other words, each position corresponds to a unique point cloud image, and all the point cloud images captured by LiDAR B constitute the map of area A. The current goal is to obtain the position of LiDAR C, which is also moving in area A. The method used is that LiDAR C captures a point cloud image at its current position, compares this point cloud image with the map of area A, and finds the point cloud image in the map of area A that is identical to the point cloud image. The current position of LiDAR C is the same as the position where LiDAR B captured the same point cloud image. Since each point cloud image and position in the map of area A correspond one-to-one, the current position of LiDAR C is obtained. Because the point cloud image generated by LiDAR B is compared with the point cloud image generated by LiDAR C, LiDAR B and LiDAR C must be identical. This ensures that the point cloud images captured by LiDAR B and LiDAR C at the same location are identical. Each point cloud image in the map of area A is also called a candidate frame, and the point cloud image captured by LiDAR C at the current location is called a query frame.

[0030] Example 1

[0031] Figure 1 This is a flowchart of a laser radar position recognition method provided in Example 1 of the present invention. This embodiment is applicable to situations where the query frame and the candidate frame come from laser radars with different structures. The method can be executed by a laser radar position recognition device through software or hardware, and specifically includes the following steps:

[0032] S110: Create a virtual laser radar corresponding to the target laser radar.

[0033] The target LiDAR refers to the LiDAR at the location to be tested, and the designated LiDAR corresponds to the LiDAR used to generate the point cloud map. Because the target LiDAR and the designated LiDAR in this invention have different structures, the point cloud map captured by the target LiDAR cannot be directly compared with the point cloud map captured by the designated LiDAR. Instead, the point cloud map captured by the designated LiDAR must be processed into a virtual point cloud map that matches the point cloud map captured by the target LiDAR.

[0034] The purpose of creating a virtual lidar for the target lidar is to use the virtual lidar to reprocess the cloud point map, so as to obtain a virtual cloud point map that can match the cloud point map taken by the target lidar.

[0035] Optionally, creating a virtual laser radar corresponding to the target laser radar includes: determining a passage area of the target laser radar on a map; and creating the virtual laser radar in the passage area.

[0036] Virtual LiDAR is used to simulate a target LiDAR. When a specified LiDAR generates a cloud point map, it usually generates a corresponding cloud point map at every location in the area. That is, the number of candidate frames is very large, and the target LiDAR's range of motion may be limited. Therefore, there is no need to compare the query frame with all candidate frames. For example, the target LiDAR can be a LiDAR installed on a vehicle for vehicle positioning. Its passage area is limited to the road. Therefore, there is no need to compare the query frame generated by the target LiDAR with the candidate frames at non-road locations. That is, at this time, the passage area of the target LiDAR on the map is the road area. Only the virtual LiDAR needs to be created in the road area, and thus only the candidate frames in the road area are needed.

[0037] The advantage of this setting is that by creating a virtual lidar in the passable area, the processing of candidate frames in other non-passable areas is omitted, which saves computing resources and improves computing efficiency.

[0038] S120. Search for scanning points of the virtual laser radar on the map according to the parameter information of the virtual laser radar, and construct a virtual frame according to the scanning points. The map is constructed according to the data collected by the specified laser radar. The specified laser radar and the target laser radar are heterogeneous devices.

[0039] The scanning points mentioned here refer to the points present in each candidate frame, that is, each cloud point map, and not the location of the virtual LiDAR. The parameter information of the virtual LiDAR is the same as that of the target LiDAR. For example, due to the variety of target LiDARs, different target LiDARs may have different firmware characteristics, which constitute the parameter information of the target LiDAR. For example, some target LiDARs have a wider field of view, some have more lines, or some have a higher resolution. Heterogeneous refers to different parameter information, specifically indicating that certain specific parameters of the designated LiDAR and the target LiDAR are different.

[0040] Optionally, the parameter information includes radar angular range and radar resolution.

[0041] Radar angular range refers to the scanning angle range of the target lidar or virtual lidar. The target lidar cannot scan areas outside the radar angular range, so there are no points corresponding to areas outside the radar angular range in the query frame's point cloud. Radar resolution refers to the radar's recognition accuracy, for example, how far away the target lidar can distinguish objects.

[0042] The advantage of this setting is that it provides parameter information of two commonly used laser radars for reference. Of course, the types of laser radar parameter information are not limited to this.

[0043] Optionally, the scanning point of the virtual laser radar is found on the map according to the parameter information of the virtual laser radar, including: calculating the angle between each point on the map and the virtual laser radar; and determining the point whose angle matches the parameter information as the scanning point of the virtual laser radar.

[0044] When both the target and designated LiDARs are multi-line LiDARs, the designated LiDAR generating the point cloud map typically has higher precision, resulting in a denser distribution of points within each candidate frame. In contrast, the target LiDAR has lower precision, and the corresponding virtual LiDAR also has lower precision. Therefore, even if the target and designated LiDARs capture point clouds at the same location, the designated LiDAR's point cloud will be denser than the target LiDAR's.

[0045] Therefore, calculating the angle between each point on the map and the virtual LiDAR can be understood as calculating the emission angle of the laser beam for each point on the generated map, finding the points among all the emission angles that match the emission angle of the virtual LiDAR, and then retaining these matching points in the point cloud map as the scanning points of the virtual LiDAR. In other words, based on the parameters of the virtual LiDAR, each point in the point cloud map is determined to be a scanning point of the virtual LiDAR. If the scanning point is on the virtual LiDAR's scan line, it is retained; otherwise, it is deleted. The remaining points ultimately constitute the virtual frame.

[0046] The advantage of this setting is that it removes scanning points that are useless or even interfere with recognition, making the final position recognition result more accurate.

[0047] The identified scanning points are processed to construct a virtual frame. It can be understood that the virtual frame is a point cloud image that is simpler than the candidate frame after some useless points in the candidate frame are removed.

[0048] S130: Obtain a query frame of the target laser radar.

[0049] The query frame of the target lidar is a point cloud image captured by the target lidar, which is used to determine the current position of the target lidar by matching it with the virtual frame.

[0050] S140 : Select a target frame from the virtual frames according to the query frame.

[0051] The query frame is compared with the virtual frame. If they are consistent, it means that the position of the target lidar is the same as the position of the specified lidar when the candidate frame corresponding to the virtual frame is shot. The virtual frame that is consistent with or closest to the query frame is selected as the target frame.

[0052] S150: Identify the location information of the target frame as the location information of the query frame.

[0053] The target frame's location information is the location information of the candidate frame corresponding to the target frame, that is, the location of the designated LiDAR when the candidate frame was captured. This location is identified as the query frame's location, that is, the location of the target LiDAR.

[0054] The technical solution of this embodiment is to create a virtual laser radar corresponding to the target laser radar; search for the scanning points of the virtual laser radar on the map according to the parameter information of the virtual laser radar, and construct a virtual frame according to the scanning points. The map is constructed according to the data collected by the specified laser radar, and the specified laser radar and the target laser radar are heterogeneous devices; obtain the query frame of the target laser radar; select the target frame from the virtual frame according to the query frame; identify the position information of the target frame as the position information of the query frame. That is to say, by creating a virtual laser radar and constructing a virtual frame corresponding to the virtual laser radar, the technical problem that the laser radar position recognition requires the query frame and the candidate frame to come from the laser radar of the same structure, which makes the existing high-precision maps from other heterogeneous laser radars unable to be used for positioning is solved, and the existing laser radar map can be used for positioning of other heterogeneous laser radars.

[0055] Optionally, calculating the angle between each point on the map and the virtual laser radar includes: calculating the vertical angle and horizontal angle between each point and the virtual laser radar based on the position information of each point on the map and the position information of the virtual laser radar on the map.

[0056] In other words, the angle between the laser and the virtual LiDAR is calculated for each point in the cloud point map. Based on a certain direction, any direction can be represented by two quantities: vertical angle and horizontal angle. The specific formula for angle calculation is as follows:

[0057]

[0058]

[0059] Among them, i represents point i on the map, V i Represents the vertical angle between point i and the virtual laser radar, H i represents the horizontal angle between point i and the virtual laser radar, P i Indicates the position of point i in the high-precision map, vL krepresents the position of the virtual lidar, x, y, and z represent the horizontal, vertical, and longitudinal directions, respectively.

[0060] The advantage of this setting is that it provides a simple method to calculate the angle between each point on the map and the virtual lidar.

[0061] Example 2

[0062] Figure 2 This is a flowchart of a laser radar position recognition method provided in Example 2 of the present invention. This embodiment is applicable to situations where the query frame and the candidate frame come from laser radars of different structures. The method can be executed by the laser radar position recognition device through software or hardware. It is a further refinement of Example 1, and the steps that are the same or similar to Example 1 will not be repeated. Specifically, the method includes the following steps:

[0063] S210: Create a virtual laser radar corresponding to the target laser radar.

[0064] S220. Search for scanning points of the virtual laser radar on the map according to the parameter information of the virtual laser radar, and construct a virtual frame according to the scanning points. The map is constructed according to the data collected by the specified laser radar. The specified laser radar and the target laser radar are heterogeneous devices.

[0065] S230: Obtain a query frame of the target laser radar.

[0066] S240 : Select candidate virtual frames from the virtual frames according to the query frame.

[0067] The core of this embodiment is to decompose the step of selecting a target frame from virtual frames into first selecting candidate virtual frames from the virtual frames, and then selecting the target frame from the candidate virtual frames. S240 is equivalent to an initial screening process, which finds several candidate virtual frames that are relatively similar to the query frame from a large number of virtual frames. Then, a detailed comparison and analysis is performed to obtain the target frame that is most similar to or identical to the query frame.

[0068] Optionally, selecting a candidate virtual frame from the virtual frame according to the query frame includes: extracting fast global features of the virtual frame and extracting fast global features of the query frame; and selecting the candidate virtual frame from the virtual frame according to the fast global features of the virtual frame and the fast global features of the query frame.

[0069] Fast global features correspond to the fine global features mentioned below. Compared to fine global features, fast global features are coarser, have a simpler comparison method, and are faster. They are suitable for initial screening of virtual frames to obtain candidate virtual frames. The method for extracting fast global features for virtual frames should be the same as that for query frames, so that the virtual frame and query frame can be compared using fast global features.

[0070] Optionally, extracting fast global features of the virtual frame includes: dividing the virtual frame into grids of a preset size and counting the number of points in each grid to obtain the fast global features of the virtual frame; extracting fast global features of the query frame includes: dividing the query frame into grids of a preset size and counting the number of points in each grid to obtain the fast global features of the query frame.

[0071] For example, each virtual frame is divided into a 5×5 grid, the number of points in each of the 25 grids is counted, and the 25 points are arranged into a 5×5 matrix. The mean or standard deviation of the points in the same row or column of the 5×5 matrix is further calculated, resulting in a 5×1 or 1×5 matrix as the fast global feature of the virtual frame. Corresponding to the virtual frame, the query frame is also divided into a 5×5 grid, the number of points in each of the 25 grids is counted, and the 25 points are arranged into a 5×5 matrix. The mean or standard deviation of the points in the same row or column of the 5×5 matrix is further calculated, resulting in a 5×1 or 1×5 matrix as the fast global feature of the query frame.

[0072] Assuming that the fast global features of both the virtual frame and the query frame are 5×1 matrices, the virtual frame matrix is compared with the query frame matrix. From a large number of virtual frames, those whose matrices are identical to the query frame matrix are identified as candidate virtual frames for further screening. Optionally, a KD-tree search can be performed using the fast global features to quickly obtain candidate virtual frames.

[0073] The advantage of this setting is that it provides a method for easily obtaining fast global features, and uses the fast global features to improve the efficiency of position recognition.

[0074] S250: Calculate the similarity between the query frame and the candidate virtual frame.

[0075] The query frame is compared and analyzed in detail with the candidate virtual frames to obtain the similarity between each candidate virtual frame and the query frame, thereby finding the candidate virtual frame that is closest to the query frame.

[0076] Optionally, calculating the similarity between the query frame and the candidate virtual frame includes: extracting fine global features of the virtual frame and extracting fine global features of the query frame; and calculating the similarity between the query frame and the candidate virtual frame based on the fine global features of the virtual frame and the fine global features of the query frame.

[0077] Compared to fast global features, fine global features are more complex to compare, but more accurate, and are used to more accurately characterize the features of virtual frames or query frames. It should be noted that fine global features are extracted for all virtual frames here, rather than just for candidate virtual frames. The reason for this is that after obtaining the fine global features of all virtual frames in advance, these features can be directly called when the target lidar remains unchanged, without the need to repeatedly extract fine global features.

[0078] Optionally, extracting fine global features of the virtual frame includes analyzing spatial attribute information of the virtual frame to obtain fine global features of the virtual frame; and extracting fine global features of the query frame includes analyzing spatial attribute information of the query frame to obtain fine global features of the query frame. The spatial attribute information may refer to spatial distribution information of points in the frame, and the spatial distribution information may include position information and quantity information.

[0079] Taking a virtual frame as an example, the virtual frame is divided into grids of a preset size. For example, each virtual frame is divided into a 5×5 grid. Each grid is further divided into several sub-grids from top to bottom. Assuming that each grid is divided into 8 sub-grids, the overall value of the grid is determined based on the point distribution in these 8 sub-grids. For example, if there is 1 point in the top sub-grid, it is recorded as 1, and if there is 1 point in the second sub-grid, it is recorded as 0.8, and so on. That is, the point distribution in the sub-grid is abstracted with numbers, and the values of each sub-grid in the grid are added to obtain the value of the grid. The values of each grid of these 25 grids are counted and the 25 values are listed as a 5×5 matrix. This matrix can represent the spatial properties of the points in the virtual frame, that is, the matrix is the fine global feature of each virtual frame. Similarly, the fine global feature of the query frame can be obtained.

[0080] Continuing with the previous example, the fine global features of the query frame and the fine global features of the virtual frame are both represented as 5×5 matrices. Therefore, it is necessary to calculate the similarity between the two matrices to serve as the similarity between the query frame and the candidate virtual frame. The formula for calculating similarity is as follows:

[0081]

[0082] Among them, Dis represents the similarity between two frames, represents the jth column of fine global features of the query frame, represents the jth column of fine global features for the candidate frame, j represents the jth column of fine global features, Nd represents the total number of columns of fine global features, and the double vertical bar operator represents the 2-norm of the solution vector. The similarity range is [0, 1]. The smaller the similarity calculated between two frames, the more similar they are.

[0083] The advantage of this setting is that it provides a method for calculating fine global features and similarity, and uses fine global features to make position judgments, thereby enhancing the accuracy of position recognition.

[0084] S260: Determine a target frame from the candidate virtual frames according to the similarity.

[0085] The candidate virtual frame with the smallest similarity to the query frame among all candidate virtual frames is selected as the target frame. A similarity threshold can be set in advance. When the minimum similarity is less than the similarity threshold, the target frame is determined, and when the minimum similarity is greater than or equal to the similarity threshold, an error is reported.

[0086] S270: Identify the location information of the target frame as the location information of the query frame.

[0087] Since the point distribution in the target frame point cloud image is very similar to that in the query frame point cloud image, it is considered that the target frame and the query frame are shot at the same position, and the position information of the target frame is identified as the position information of the query frame.

[0088] The technical solution of this embodiment is to select candidate virtual frames from virtual frames based on the query frame; calculate the similarity between the query frame and the candidate virtual frames; and determine the target frame from the candidate virtual frames based on the similarity. In other words, the technical solution is to decompose the step of selecting the target frame from the virtual frame into first selecting the candidate virtual frame from the virtual frame and then selecting the target frame from the candidate virtual frames. This solves the problem of slow laser radar position recognition speed by performing preliminary screening and then fine screening, thereby achieving the effect of improving position recognition efficiency.

[0089] Example 3

[0090] Figure 3 This is a structural diagram of a laser radar position recognition device provided in Example 3 of the present invention. The laser radar position recognition device provided in this embodiment of the present invention can execute the laser radar position recognition method provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0091] A laser radar position recognition device, comprising:

[0092] A virtual laser radar creation module 310 is used to create a virtual laser radar corresponding to the target laser radar;

[0093] A virtual frame construction module 320 is used to find scanning points of the virtual laser radar on the map based on the parameter information of the virtual laser radar and construct a virtual frame based on the scanning points. The map is constructed based on data collected by the specified laser radar. The specified laser radar and the target laser radar are heterogeneous devices.

[0094] A query frame acquisition module 330 is used to acquire a query frame of a target laser radar;

[0095] a target frame selection module 340 for selecting a target frame from the virtual frame according to the query frame;

[0096] The position information identification module 350 is configured to identify the position information of the target frame as the position information of the query frame.

[0097] Optionally, the virtual laser radar creation module 310 includes:

[0098] The passage area determination submodule is used to determine the passage area of the target lidar on the map;

[0099] The virtual lidar creation submodule is used to create a virtual lidar in the traffic area.

[0100] Optionally, the virtual frame construction module 320 includes:

[0101] Angle calculation submodule, used to calculate the angle between each point on the map and the virtual lidar;

[0102] The scanning point determination submodule is used to determine the point whose angle matches the parameter information as the scanning point of the virtual laser radar.

[0103] Optional, angle calculation submodule, used to:

[0104] The vertical angle and horizontal angle between each point and the virtual lidar are calculated based on the position information of each point on the map and the position information of the virtual lidar on the map.

[0105] Optionally, the target frame selection module 340 includes:

[0106] a candidate virtual frame selection submodule, configured to select a candidate virtual frame from the virtual frames according to the query frame;

[0107] A similarity calculation submodule is used to calculate the similarity between the query frame and the candidate virtual frame;

[0108] The target frame determination submodule is used to determine the target frame from the candidate virtual frames based on similarity.

[0109] Optionally, the candidate virtual frame selection submodule includes:

[0110] A fast global feature extraction unit, configured to extract fast global features of the virtual frame and fast global features of the query frame;

[0111] The candidate virtual frame selection unit is used to select a candidate virtual frame from the virtual frames according to the fast global features of the virtual frame and the fast global features of the query frame.

[0112] Optionally, a fast global feature extraction unit is used to: divide the virtual frame into grids of a preset size and count the number of points in each grid, thereby obtaining a fast global feature of the virtual frame;

[0113] The fast global feature extraction unit is further used to divide the query frame into grids of a preset size and count the number of points in each grid, thereby obtaining the fast global features of the query frame.

[0114] Optional similarity calculation submodule, including:

[0115] A fine global feature extraction unit, configured to extract fine global features of the virtual frame and fine global features of the query frame;

[0116] The similarity calculation unit is used to calculate the similarity between the query frame and the candidate virtual frame according to the fine global features of the virtual frame and the fine global features of the query frame.

[0117] Optionally, a fine global feature extraction unit is used to analyze the spatial attribute information of the virtual frame to obtain fine global features of the virtual frame;

[0118] The fine global feature extraction unit is further used to analyze the spatial attribute information of the query frame to obtain the fine global features of the query frame.

[0119] The technical solution of this embodiment is to create a virtual laser radar corresponding to the target laser radar; search for the scanning points of the virtual laser radar on the map according to the parameter information of the virtual laser radar, and construct a virtual frame according to the scanning points. The map is constructed according to the data collected by the specified laser radar, and the specified laser radar and the target laser radar are heterogeneous devices; obtain the query frame of the target laser radar; select the target frame from the virtual frame according to the query frame; identify the position information of the target frame as the position information of the query frame. That is to say, by creating a virtual laser radar and constructing a virtual frame corresponding to the virtual laser radar, the technical problem that the laser radar position recognition requires the query frame and the candidate frame to come from the laser radar of the same structure, which makes the existing high-precision maps from other heterogeneous laser radars unable to be used for positioning is solved, and the existing laser radar map can be used for positioning of other heterogeneous laser radars.

[0120] Example 4

[0121] Figure 4This is a structural diagram of an electronic device provided in the fourth embodiment of the present invention, such as Figure 4 As shown, the device includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of processors 410 in the device can be one or more. Figure 4 In the embodiment, a processor 410 is used as an example; the processor 410, memory 420, input device 430 and output device 440 in the device can be connected via a bus or other means. Figure 4 The bus connection is taken as an example.

[0122] Memory 420, as a computer-readable storage medium, can be used to store software programs, computer executable programs, and modules, such as the program instructions / modules corresponding to the laser radar position recognition method in the embodiments of the present invention (for example, the virtual laser radar creation module 310, virtual frame construction module 320, query frame acquisition module 330, target frame selection module 340, and position information recognition module 350 in the laser radar position recognition device). Processor 410 executes the software programs, instructions, and modules stored in memory 420 to execute various functional applications and data processing of the device, thereby implementing the aforementioned laser radar position recognition method.

[0123] The memory 420 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal, etc. In addition, the memory 420 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 420 may further include a memory remotely located relative to the processor 410, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0124] The input device 430 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the device. The output device 440 may include a display device such as a display screen.

[0125] Example 5

[0126] Embodiment 5 of the present invention further provides a storage medium containing computer-executable instructions. When the computer-executable instructions are executed by a computer processor, the computer-executable instructions are used to perform a laser radar position recognition method, the method comprising:

[0127] Create a virtual lidar corresponding to the target lidar;

[0128] Find the virtual LiDAR's scanning points on the map based on its parameter information, and build a virtual frame based on the scanning points. The map is built based on the data collected by the specified LiDAR. The specified LiDAR and the target LiDAR are heterogeneous devices.

[0129] Get the target lidar query frame;

[0130] Selecting a target frame from the virtual frames according to the query frame;

[0131] The position information of the target frame is identified as the position information of the query frame.

[0132] Of course, the storage medium containing computer-executable instructions provided in an embodiment of the present invention is not limited to the method operations described above, and can also execute related operations in the laser radar position recognition method provided in any embodiment of the present invention.

[0133] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0134] It is worth noting that in the embodiment of the above-mentioned laser radar position recognition device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0135] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A laser radar position recognition method, characterized in that: include: Create a virtual lidar corresponding to the target lidar; Searching for scanning points of the virtual LiDAR on a map based on parameter information of the virtual LiDAR, and constructing a virtual frame based on the scanning points, wherein the map is constructed based on data collected by a designated LiDAR, and the designated LiDAR and the target LiDAR are heterogeneous devices; wherein the scanning points refer to points present in each candidate frame, i.e., each point cloud map; Obtaining a query frame of the target laser radar; selecting a target frame from the virtual frames according to the query frame; identifying the position information of the target frame as the position information of the query frame; The step of creating a virtual laser radar corresponding to the target laser radar includes: Determining a passage area of the target laser radar on the map; Creating the virtual laser radar in the traffic area; The step of searching for a scanning point of the virtual laser radar on a map according to parameter information of the virtual laser radar includes: Calculating the angle between each point on the map and the virtual laser radar; Determine a point whose angle matches the parameter information as a scanning point of the virtual laser radar; Calculating the angle between each point on the map and the virtual laser radar includes: The vertical angle and the horizontal angle between each point and the virtual laser radar are calculated according to the position information of each point on the map and the position information of the virtual laser radar on the map.

2. The laser radar position recognition method according to claim 1, characterized in that: The selecting a target frame from the virtual frame according to the query frame includes: selecting a candidate virtual frame from the virtual frames according to the query frame; Calculating the similarity between the query frame and the candidate virtual frame; The target frame is determined from the candidate virtual frames according to the similarity.

3. The laser radar position recognition method according to claim 2, characterized in that: The selecting a candidate virtual frame from the virtual frames according to the query frame includes: Extracting fast global features of the virtual frame and extracting fast global features of the query frame; The candidate virtual frame is selected from the virtual frames according to the fast global feature of the virtual frame and the fast global feature of the query frame.

4. The laser radar position recognition method according to claim 3, characterized in that: The extracting of the fast global features of the virtual frame includes: dividing the virtual frame into grids of a preset size, and counting the number of points in each grid, thereby obtaining the fast global features of the virtual frame; The extracting of the fast global features of the query frame includes: dividing the query frame into grids of the preset size, and counting the number of points in each grid, thereby obtaining the fast global features of the query frame.

5. The laser radar position recognition method according to claim 2, characterized in that: Calculating the similarity between the query frame and the candidate virtual frame includes: extracting fine global features of the virtual frame and extracting fine global features of the query frame; The similarity between the query frame and the candidate virtual frame is calculated according to the fine global features of the virtual frame and the fine global features of the query frame.

6. The laser radar position recognition method according to claim 5, characterized in that: The extracting of the fine global features of the virtual frame includes: analyzing the spatial attribute information of the virtual frame to obtain the fine global features of the virtual frame; The extracting of the fine global features of the query frame includes: analyzing the spatial attribute information of the query frame to obtain the fine global features of the query frame.

7. A laser radar position recognition device, characterized in that: include: Virtual LiDAR creation module, used to create a virtual LiDAR corresponding to the target LiDAR; A virtual frame construction module is configured to search for scanning points of the virtual LiDAR on a map based on parameter information of the virtual LiDAR and to construct a virtual frame based on the scanning points. The map is constructed based on data collected by a designated LiDAR, and the designated LiDAR and the target LiDAR are heterogeneous devices. The scanning points are points present in each candidate frame, i.e., each point cloud map. A query frame acquisition module, used to acquire the query frame of the target laser radar; a target frame selection module, configured to select a target frame from the virtual frame according to the query frame; a position information identification module, configured to identify the position information of the target frame as the position information of the query frame; The virtual laser radar creation module includes a passage area determination submodule and a virtual laser radar creation submodule; The passage area determination submodule is used to determine the passage area of the target laser radar on the map; The virtual laser radar creation submodule is used to create a virtual laser radar in the traffic area; The virtual frame construction module includes an angle calculation submodule and a scanning point determination submodule; The angle calculation submodule is used to calculate the angle between each point on the map and the virtual laser radar; The scanning point determination submodule is used to determine the point whose angle matches the parameter information as the scanning point of the virtual laser radar; The angle calculation submodule is specifically used to: The vertical angle and horizontal angle between each point and the virtual lidar are calculated based on the position information of each point on the map and the position information of the virtual lidar on the map.

Images