Laser point cloud target classification method and device and medium thereof

Abstract

The invention relates to the technical field of mobile robots, in particular to a laser point cloud target classification method and device and a medium thereof, and the method comprises the steps: scanning an underground parking lot through a three-dimensional laser radar carried by a mobile robot, and obtaining original point cloud data; processing the original point cloud data according to a set first threshold value to obtain spatial point cloud information; filtering the spatial point cloud information according to a set second threshold value to obtain first type column point cloud information and spatial point cloud information of non-first type column point cloud; performing statistical feature analysis on the point cloud information of the first type of columns to obtain size information of the first type of columns; screening spatial point cloud information of non-first type of column point clouds according to the first type of column size information to obtain second type of column point cloud information and residual spatial point cloud information; and carrying out feature fusion on the residual space point cloud information, and obtaining the point cloud information of each obstacle through point cloud segmentation. According to the invention, stable, real-time and high-precision target identification and classification can be realized.

Description

Technical Field

[0001] This invention relates to the field of mobile robot technology, and in particular to a laser point cloud target classification method, device and medium. Background Technology

[0002] With the widespread application of autonomous driving and intelligent mobile robots in indoor environments (such as underground parking lots and warehouses), accurate and robust object perception has become crucial for achieving safe autonomous navigation. Currently, mainstream object detection solutions have the following limitations: 1. Two-dimensional LiDAR-based solution: This solution can only provide two-dimensional profile information of the environment, which has a serious problem of missing dimensions. It cannot obtain three-dimensional features such as the height and outline of the target, which makes it impossible to effectively distinguish vehicles (with significant height and complex shape), pillars (vertical columnar objects) and ground debris (low obstacles), which is prone to misjudgment and missed detection.

[0003] 2. Vision camera-based solution: This solution is severely affected by lighting conditions, such as dim lighting and drastic changes in light and dark in underground parking lots (such as entrance and exit passages), which can easily lead to overexposure or underexposure, making feature extraction difficult. At the same time, the visual algorithm has high computational complexity and poses a risk of privacy leakage.

[0004] 3. Traditional 3D point cloud processing methods: This approach processes all point clouds of pillars and vehicles together. Due to the limitations of point cloud resolution, these methods are unable to effectively handle the sparsity, noise, and dynamic changes in the scene, leading to model instability and decreased recognition accuracy. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0006] Therefore, this invention provides a laser point cloud target classification method. This invention can classify targets in indoor environments with changing lighting, dynamic interference, and complex structures, avoiding the shortcomings of existing technologies such as missing dimensions, poor environmental adaptability, and low recognition accuracy, and achieving stable, real-time, and fast high-precision target recognition and classification.

[0007] The laser point cloud target classification method according to an embodiment of the present invention includes the following steps: S1 uses a 3D LiDAR mounted on a mobile robot to scan the underground parking lot and obtain raw point cloud data; S2, process the raw point cloud data according to the set first threshold to obtain spatial point cloud information; S3, filter the spatial point cloud information according to the set second threshold to obtain the first type of column point cloud information and the spatial point cloud information of non-first type column point cloud; S4, Perform statistical feature analysis on the point cloud information of the first type of pillars to obtain the size information of the first type of pillars; S5, Based on the first type of column size information, filter the spatial point cloud information of non-first type of column point cloud to obtain the second type of column point cloud information and the remaining spatial point cloud information; S6 performs feature fusion on the remaining spatial point cloud information and obtains the point cloud information of each obstacle through point cloud segmentation.

[0008] The beneficial effects of this invention are that the laser point cloud target classification method can quickly distinguish between ground point clouds and pillar point clouds by setting two thresholds, further filter out pillar point clouds through statistical feature analysis, and finally obtain vehicle point clouds through feature fusion and segmentation. This invention can perform target classification in indoor environments with changing lighting, dynamic interference, and complex structures, avoiding the shortcomings of existing technologies such as missing dimensions, poor environmental adaptability, and low recognition accuracy. In addition, this invention does not rely on high-density point clouds, deep learning models, or high-precision maps, and has the advantages of low computational complexity, strong real-time performance, good interpretability, and strong adaptability to sparse point clouds and occlusion. It can achieve stable, real-time, and fast high-precision target recognition and classification under various complex working conditions, providing reliable environmental perception capabilities for mobile robots.

[0009] According to an embodiment of the present invention, step S4 specifically includes the following steps: S41. Based on the angle parameters in the vehicle coordinate system, sort the original point cloud data after filtering in step S3. During the sorting process, if the distance between the current point and the next adjacent point is within a preset threshold range, then it is determined that the two points belong to the point cloud of the same object cluster. S42, based on the point cloud information in the same object cluster, calculate the axis alignment boundary of the bounding box under xy projection, wherein the axis alignment boundary of the bounding box is the first type of column size information.

[0010] According to an embodiment of the present invention, the feature fusion of the raw point cloud data after statistical analysis in step S4 in step S5 specifically includes the following steps: S61, sort the remaining spatial point cloud information according to the angle sequence; S62, divide the grid into equal parts according to the angle range, and calculate the local features of the point cloud in each angle grid. The local features include the average height, average distance, average intensity and surface normal vector of the point cloud in each angle grid. S63, Process the local features of the point cloud in the current angle grid and the local features of the point cloud in the adjacent angle grid to obtain feature data; S64, calculate the smoothness of each point cloud based on the coordinate information of the point cloud; S65, normalize the feature data and the smoothness of each point cloud to obtain the nearest neighbor feature information between each point cloud and its neighboring point clouds. S66, the remaining spatial point cloud information is segmented based on the nearest neighbor feature information to obtain the point cloud information of each obstacle.

[0011] According to one embodiment of the present invention, S66 specifically includes: The support of the same object is calculated based on the nearest neighbor feature information. The calculation formula is as follows:

[0012] in, This indicates the total number of points contained in the current and next selected segments. Points belonging to the same object are represented by the following formula:

[0013] This represents the nearest neighbor feature information of the k-th point on angle grid i. Indicates the judgment threshold. Indicates the index function; like Greater than the preset boundary Then, point cloud segmentation is performed to obtain the point cloud information of each obstacle.

[0014] According to one embodiment of the present invention, the formula for calculating the average height is:

[0015] in, This indicates the number of point clouds within the current angle grid. It is the height value of the current point.

[0016] According to one embodiment of the present invention, the formula for calculating the average distance is:

[0017] in, This is the value of the current point in the xy coordinate system.

[0018] According to one embodiment of the present invention, the formula for calculating the average strength is:

[0019] in, It is the intensity value at the current point.

[0020] According to one embodiment of the present invention, the formula for calculating the smoothness is:

[0021] in, This represents the curvature change of a point sequence in the x-direction. It represents the curvature change in the y-direction and is sensitive to the vehicle edges.

[0022] A computer device according to an embodiment of the present invention includes: processor; Memory, used to store executable instructions; The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the laser point cloud target classification method as described above.

[0023] According to an embodiment of the present invention, a computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the laser point cloud target classification method as described above.

[0024] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0026] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0027] Figure 1 This is a flowchart of the method of the present invention.

[0028] Figure 2 This is a schematic diagram of the computer device structure of the present invention.

[0029] In the diagram, 10 is a computer device; 1002 is a processor; 1004 is a memory; and 1006 is a transmission device. Detailed Implementation

[0030] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0031] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, features defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0032] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0033] Example 1 This application provides a laser point cloud target classification method, such as... Figure 1 As shown, the method includes the following steps: S1 uses a mobile robot equipped with a 3D LiDAR to scan the underground parking lot and acquire raw point cloud data, which consists of several 3D coordinate points. The set that constitutes the composition.

[0034] S2, process the raw point cloud data according to the set first threshold to obtain spatial point cloud information; for example, the first threshold is set to... If three-dimensional coordinate points height Within the first threshold range, it is spatial point cloud information.

[0035] S3, filter the spatial point cloud information according to the set second threshold to obtain the first type of column point cloud information and the spatial point cloud information of non-first type column point cloud; Generally, the height of a pillar is always higher than other objects. A second threshold is set to filter out the first type of pillar point cloud. In this embodiment, the second threshold is set according to the installation position of the 3D LiDAR. For example, if the 3D LiDAR is 0.2m above the ground, the second threshold can be set to 2.0m; if the 3D LiDAR is 1.5m above the ground, the second threshold can be set to 0.5m. If the 3D coordinate points filtered in step S2... height Within the second threshold range, it is a first-class column point cloud, which is a partial point cloud of the column.

[0036] S4, Perform statistical feature analysis on the point cloud information of the first type of pillars to obtain the size information of the first type of pillars; S5, Based on the first type of column size information, filter the spatial point cloud information of non-first type of column point cloud to obtain the second type of column point cloud information and the remaining spatial point cloud information; S6 performs feature fusion on the remaining spatial point cloud information and obtains the point cloud information of each obstacle through point cloud segmentation.

[0037] In this embodiment, since the height of the pillars extends throughout the entire space, there are still pillar point clouds below the second threshold in the spatial point cloud information of non-first type pillar point clouds. Step S4 specifically includes the following steps: S41, based on the angle parameters in the vehicle coordinate system, sort the spatial point cloud information of the first type of column point cloud. During the sorting process, if the distance between the current point and the next adjacent point is within a preset threshold range, then it is determined that the two points belong to the point cloud of the same object cluster.

[0038] The spatial point cloud information of the first type of column point cloud is preferably sorted from smallest to largest to facilitate point cloud extraction.

[0039] S42, based on the point cloud information in the same object cluster, calculate the axis-aligned boundary of the bounding box under the xy projection. For example, the axis-aligned boundary of the bounding box under the x projection is... The axis-aligned boundary of the bounding box under y-projection is The axis-aligned boundary of the bounding box is the first type of column size information.

[0040] In step S5, the spatial point cloud information of non-first-type column point clouds is filtered according to the axis-aligned boundary of the first-type column bounding box to obtain the second-type column point cloud information. Specifically, the three-dimensional coordinate points of the spatial point cloud information of non-first-type column point clouds in step S3 are... The x-coordinate of the bounding box is aligned with the x-axis under the x-projection. And the y-coordinate of the bounding box is aligned with the boundary axis under the y-projection. If the value is found to be in the second category of column point cloud information, then the point cloud information of all columns can be obtained.

[0041] In this embodiment, step S6 specifically includes the following steps: S61, Sort the raw point cloud data after statistical analysis in step S4 according to the angle sequence; specifically, based on three-dimensional coordinate points. The formula for calculating the point cloud angle is as follows: .

[0042] S62, divide the grid into equal parts according to the angle range, and calculate the local features of the point cloud in each angle grid. The local features include the average height, average distance, average intensity and surface normal vector of the point cloud in each angle grid. Furthermore, the formula for calculating the average height is:

[0043] in, This indicates the number of point clouds within the current angle grid. It is the height value of the current point.

[0044] The formula for calculating the average distance is:

[0045] in, This is the value of the current point in the xy coordinate system.

[0046] The formula for calculating average strength is:

[0047] in, It is the intensity value at the current point.

[0048] The surface normal vector n[i] is obtained by calculating PCA at local points.

[0049] S63, based on the local features of the point cloud in the current angle grid and the local features of the point cloud in the adjacent angle grid, process to obtain feature data, which includes: height difference data, distance difference data, intensity difference data and normal vector consistency data.

[0050] A first-order difference is performed based on the average height of the point cloud within the current angular grid and the average height of the point cloud within adjacent angular grids to reflect the vertical abrupt change of the obstacle. The calculation formula is as follows: .

[0051] A first-order difference is performed based on the average distance of the point cloud within the current angle grid and the average distance of the point cloud within the adjacent angle grid to distinguish the large values ​​that may occur when obstacles are mixed in at the same angle. The calculation formula is as follows: .

[0052] A first-order difference is performed based on the average intensity of the point cloud within the current angle grid and the average intensity of the point cloud within the adjacent angle grid to distinguish the material or object type of the obstacle. The calculation formula is as follows: .

[0053] Consistency processing is performed based on the surface normal vectors of the point cloud within the current angle grid and the surface normal vectors of the point cloud within the adjacent angle grid. The calculation formula is as follows: .

[0054] In this embodiment, the point cloud automatic segmentation mechanism based on the first-order difference trend of the angle sequence can directly depict the changing trend between adjacent angles without reconstructing the three-dimensional adjacency relationship, thus avoiding complex spatial searches and exhibiting better scale invariance. S64. Based on the coordinate information of the point cloud, calculate the smoothness of each point cloud to evaluate the continuity of the point cloud. The calculation formula is as follows:

[0055] in, This represents the curvature change of a point sequence in the x-direction. It represents the curvature change in the y-direction and is sensitive to the vehicle edges.

[0056] S65, normalize the feature data and the smoothness of each point cloud to obtain the nearest neighbor feature information between each point cloud and its neighboring point clouds. Furthermore, the average height of the first-order difference is normalized to the range [0,1], expressed as follows:

[0057] The mean distance normalization of the first-order difference is mapped to the range [0,1], expressed as:

[0058] The average intensity normalization mapping of the first-order difference to the range [0,1] is expressed as follows:

[0059] The expression for normalizing the surface normal vector to the range [0,1] is:

[0060] To normalize the smoothness to the range [0,1], the expression is:

[0061] Where: k = 10,

[0062] S66. The remaining spatial point cloud information is segmented based on the nearest neighbor feature information to obtain the point cloud information of each obstacle. Obstacles include, but are not limited to, vehicles, trash cans, electric vehicles, or others.

[0063] Furthermore, step S66 specifically includes: The support of the same object is calculated based on the nearest neighbor feature information. The calculation formula is as follows:

[0064] in, This indicates how many point clouds at that boundary belong to the same object. This represents the total number of points contained in the current and subsequent segments. In other words, during the advancement of the angle sequence, based on the first-order difference trend, it temporarily assumes that the points belong to the same object and have not yet been finally confirmed as a continuous set of angle point clouds. Points belonging to the same object are represented by the following formula:

[0065] This represents the nearest neighbor feature information of the k-th point on angle grid i, such as: height difference, distance difference, normal vector difference, smoothness, and intensity difference. (1 indicates strong support for "this is a breakpoint / boundary", 0 indicates no support). The threshold values ​​are: height difference 0.6, normal vector difference 0.5, intensity difference 0.5, smoothness difference 0.6, and distance difference 0.6. This represents the index function; if the expression within the parentheses is true, it takes the value 1, otherwise it takes the value 0.

[0066] like Greater than the preset boundary Then, point cloud segmentation is performed to obtain point cloud data belonging to the same object. S[i]∈[0,1] represents the proportion of evidence supporting the "segmentation" at the boundary. The higher the S, the more credible the breakpoint. By jointly enhancing the boundary credibility, supporting noisy environments and occlusion conditions, the accuracy of point cloud segmentation is improved. It should be noted that in indoor settings The recommended value is 0.45~0.55. In this embodiment, the point cloud is divided into obstacle object clusters according to the segmentation points. For each obstacle object cluster, the length L, width W, and height H of the obstacle object cluster are calculated. Furthermore, the calculation formula is as follows:

[0067]

[0068]

[0069] Where i represents the i-th object cluster, Let represent the maximum value in the x-direction of the point cloud of the i-th object. Let represent the minimum value in the x-direction of the point cloud of the i-th object. This represents the maximum value in the y-direction of the point cloud of the i-th object. Let represent the minimum value in the y-direction of the point cloud of the i-th object. This represents the maximum value in the z-direction of the point cloud of the i-th object. Let represent the minimum value in the z-direction of the point cloud of the i-th object.

[0070] The laser point cloud target classification method in this application can quickly distinguish between ground point clouds and pillar point clouds by setting two thresholds. Statistical feature analysis is then used to further filter out pillar point clouds, and finally, vehicle point clouds are obtained through feature fusion and segmentation. This invention can classify targets in indoor environments with varying lighting, dynamic interference, and complex structures, avoiding the shortcomings of existing technologies such as missing dimensions, poor environmental adaptability, and low recognition accuracy. Furthermore, this invention does not rely on high-density point clouds, deep learning models, or high-precision maps, and has advantages such as low computational complexity, strong real-time performance, good interpretability, and strong adaptability to sparse point clouds and occlusion. It can achieve stable, real-time, and fast high-precision target recognition and classification under various complex working conditions, providing reliable environmental perception capabilities for mobile robots.

[0071] Example 2 This application provides a computer device including a processor and a memory. The memory stores at least one instruction or at least one program, which is loaded and executed by the processor to implement a laser point cloud target classification method as provided in the above method embodiments.

[0072] Figure 2 This diagram illustrates a hardware structure of a device for implementing a laser point cloud target classification method provided in the embodiments of this application. The device may constitute or include the apparatus or system provided in the embodiments of this application. Figure 2As shown, the computer device 10 may include one or more processors 1002 (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 1004 for storing data, and a transmission device 1006 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 2 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer device 10 may also include... Figure 2 The more or fewer components shown, or having the same Figure 2 The different configurations shown.

[0073] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuit may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer device 10 (or mobile device). As involved in the embodiments of this application, the data processing circuit serves as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0074] The memory 1004 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to a laser point cloud target classification method in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 1004, thereby implementing the aforementioned method. The memory 1004 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1004 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer device 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0075] The transmission device 1006 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer device 10. In one example, the transmission device 1006 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 1006 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0076] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer device 10 (or mobile device).

[0077] Example 3 This application embodiment also provides a computer-readable storage medium, which can be disposed in a server to store at least one instruction or at least one program related to implementing a laser point cloud target classification method in the method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the laser point cloud target classification method provided in the above method embodiment.

[0078] Optionally, in this embodiment, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0079] Example 4 This invention also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform a laser point cloud target classification method provided in the various optional embodiments described above.

[0080] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are also possible or may be advantageous.

[0081] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device, equipment, and storage medium embodiments are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0082] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0083] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A laser point cloud target classification method, characterized in that, The method includes the following steps: S1 uses a 3D LiDAR mounted on a mobile robot to scan the underground parking lot and obtain raw point cloud data; S2, process the raw point cloud data according to the set first threshold to obtain spatial point cloud information; S3, filter the spatial point cloud information according to the set second threshold to obtain the first type of column point cloud information and the spatial point cloud information of non-first type column point cloud; S4, Perform statistical feature analysis on the point cloud information of the first type of pillars to obtain the size information of the first type of pillars; S5, Based on the first type of column size information, filter the spatial point cloud information of non-first type of column point cloud to obtain the second type of column point cloud information and the remaining spatial point cloud information; S6 performs feature fusion on the remaining spatial point cloud information and obtains the point cloud information of each obstacle through point cloud segmentation.

2. The laser point cloud target classification method as described in claim 1, characterized in that, Step S4 specifically includes the following steps: S41, based on the angle parameters in the vehicle coordinate system, sort the spatial point cloud information of the first type of column point cloud. During the sorting process, if the distance between the current point and the next adjacent point is within a preset threshold range, then it is determined that the two points belong to the point cloud of the same object cluster. S42, based on the point cloud information in the same object cluster, calculate the axis alignment boundary of the bounding box under xy projection, wherein the axis alignment boundary of the bounding box is the first type of column size information.

3. The laser point cloud target classification method as described in claim 1, characterized in that, Step S6 specifically includes the following steps: S61, sort the remaining spatial point cloud information according to the angle sequence; S62, divide the grid into equal parts according to the angle range, and calculate the local features of the point cloud in each angle grid. The local features include the average height, average distance, average intensity and surface normal vector of the point cloud in each angle grid. S63, Process the local features of the point cloud in the current angle grid and the local features of the point cloud in the adjacent angle grid to obtain feature data; S64, calculate the smoothness of each point cloud based on the coordinate information of the point cloud; S65, normalize the feature data and the smoothness of each point cloud to obtain the nearest neighbor feature information between each point cloud and its neighboring point clouds. S66, segment the remaining spatial point cloud information based on the nearest neighbor feature information to obtain the point cloud information of each obstacle.

4. The laser point cloud target classification method as described in claim 3, characterized in that, S66 specifically includes: The support of the same object is calculated based on the nearest neighbor feature information. The calculation formula is as follows: in, This indicates the total number of points contained in the current and next selected segments. Points belonging to the same object are represented by the following formula: This represents the nearest neighbor feature information of the k-th point on angle grid i. Indicates the judgment threshold. Indicates the index function; like Greater than the preset boundary Then, point cloud segmentation is performed to obtain the point cloud information of each obstacle.

5. The laser point cloud target classification method as described in claim 3, characterized in that, The formula for calculating the average height is: in, This indicates the number of point clouds within the current angle grid. It is the height value of the current point.

6. The laser point cloud target classification method as described in claim 5, characterized in that, The formula for calculating the average distance is: in, This is the value of the current point in the xy coordinate system.

7. The laser point cloud target classification method as described in claim 5, characterized in that, The formula for calculating the average strength is: in, It is the intensity value at the current point.

8. The laser point cloud target classification method as described in claim 5, characterized in that, The formula for calculating the smoothness is: in, This represents the curvature change of a point sequence in the x-direction. It represents the curvature change in the y-direction and is sensitive to the vehicle edges.

9. A computer device, characterized in that, include: processor; Memory, used to store executable instructions; The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the laser point cloud target classification method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the laser point cloud target classification method as described in any one of claims 1 to 8.

Images