Method, apparatus, controller and movable device for identifying an obstacle

Abstract

This application belongs to the field of road emergency rescue and discloses a method, device, controller, and mobile device for obstacle identification. This application acquires target point cloud data of a detection area, detects the target point cloud data using a point cloud segmentation algorithm, and determines candidate areas and the initial obstacle types of obstacles within the candidate areas, wherein the detection area includes the candidate areas. Furthermore, the candidate areas are divided into grids, and the grid point cloud information is determined. Based on the grid point cloud information, the target detection result of the obstacles within the candidate areas is determined. When the target detection result indicates that the obstacle type is the initial obstacle type, the pose information of the obstacle is acquired. Through the processing of the acquired data, this application can accurately detect the type of obstacle in the driving path, which helps to coordinate with rescue equipment and other mobile devices to complete driving tasks and meets the autonomous operation requirements of mobile devices.

Description

Technical Field

[0001] This application relates to the field of mechanical equipment, and more specifically to a method, apparatus, controller, and movable equipment for identifying obstacles. Background Technology

[0002] Using rescue equipment for disaster relief can quickly, safely, and efficiently meet rescue needs and greatly alleviate the pressure on rescue personnel. Autonomous perception of the disaster site environment plays a crucial role in achieving autonomous disaster relief, even decisively influencing the development of corresponding autonomous rescue strategies by rescue equipment. This is a prerequisite for ensuring the safe and smooth execution of rescue operations and the safety of people and property. Currently, the sensing and detection devices of rescue equipment are relatively simple, with limited product functions. In complex road conditions, incomplete information collection can easily occur, making it difficult to assist in completing rescue missions. Summary of the Invention

[0003] The purpose of this application is to provide a method, device, controller, and mobile device for identifying obstacles, in order to solve the problem that in the prior art, information collection is often incomplete when road conditions are complex.

[0004] To achieve the above objectives, a first aspect of this application provides a method for identifying obstacles, applied to a mobile device, the method comprising:

[0005] Acquire target point cloud data of the detection area;

[0006] The target point cloud data is detected by a point cloud segmentation algorithm to determine the candidate region and the initial obstacle type of the obstacle in the candidate region, wherein the detected region contains the candidate region;

[0007] The candidate region is divided into grids and the grid point cloud information is determined;

[0008] The target detection results of obstacles in the candidate area are determined based on the grid point cloud information;

[0009] If the target detection result indicates that the obstacle type is the initial obstacle type, obtain the obstacle pose information.

[0010] In embodiments of this application, the mobile device includes a lidar and an inertial measurement unit, and acquiring target point cloud data of the detection area includes:

[0011] Initial point cloud data of the detection area is acquired using lidar;

[0012] Attitude data of mobile devices are acquired through inertial measurement units;

[0013] The initial point cloud data is corrected based on the attitude data to determine the target point cloud data.

[0014] In the embodiments of this application, the detection of target point cloud data using a point cloud segmentation algorithm to determine candidate regions and the initial obstacle types of obstacles within those candidate regions includes:

[0015] Based on concavity / convexity, the detection region is segmented using a point cloud segmentation algorithm to obtain multiple faces. The target point cloud data is used to determine if three consecutive faces have heights lower than a first preset height. If three consecutive faces are found to be lower than the first preset height, it is determined whether any two adjacent faces within these three faces are concave. If all adjacent faces are determined to be concave, the region enclosed by these three faces and the ground plane is identified as a candidate region, and the initial obstacle type is determined to be a ditch.

[0016] Obtain the normal vector of each point in the target point cloud data of the detection area and the normal vector of the ground plane; determine the area enclosed by the target point cloud data and the ground plane where the angle between the normal vector and the ground plane normal vector is less than 90 degrees as the candidate area, and determine the initial obstacle type as a slope; or

[0017] Based on concavity and convexity, the detection area is segmented using a point cloud segmentation algorithm to obtain multiple faces. Based on the target point cloud data, it is determined whether there are three consecutive faces whose heights are all higher than a second set height. If it is determined that there are three consecutive faces whose heights are higher than the second set height, it is determined whether the relationship between two adjacent faces in the three consecutive faces is one concave and one convex. If it is determined that the relationship between two adjacent faces in the three consecutive faces is one concave and one convex, the area enclosed by the three consecutive faces and the ground plane is determined as a candidate area, and the initial obstacle type of the obstacle is determined to be a vertical staircase.

[0018] In the embodiments of this application, the grid point cloud information includes at least one of the following:

[0019] Maximum elevation, minimum elevation, standard deviation, point cloud density, point cloud trend, and skewed projection distribution.

[0020] In embodiments of this application, determining the target detection result of obstacles in a candidate region based on grid point cloud information includes:

[0021] Select the meta-mesh and the corresponding neighboring mesh in the candidate region to obtain the mesh combination composed of the meta-mesh and the neighboring mesh;

[0022] The target detection result is obtained by cross-comparing the meta-mesh with the neighboring mesh. The target detection result includes the confidence that the obstacle type is the initial obstacle type.

[0023] If the confidence level meets the preset confidence level conditions, the type of obstacle is determined to be the initial obstacle type.

[0024] In embodiments of this application, determining the target detection result of obstacles in the candidate region based on grid point cloud information further includes:

[0025] If the confidence level of the grid combination does not meet the preset confidence level, increase the number of neighboring grids and return to the step of cross-comparing the meta grid with the neighboring grids to obtain the target detection result, until the confidence level of the grid combination meets the preset confidence level.

[0026] The number of neighboring grids shall not exceed the preset number.

[0027] In embodiments of this application, determining the target detection result of obstacles in the candidate region based on grid point cloud information further includes:

[0028] If the number of neighboring grids reaches a preset number and the confidence level of the grid combination still does not meet the preset confidence level condition, it is determined that the target obstacle type of the grid combination is not the initial obstacle type.

[0029] A second aspect of this application provides a device for identifying obstacles, comprising:

[0030] The point cloud data acquisition module is configured to acquire target point cloud data of the detection area;

[0031] The initial detection module is configured to detect target point cloud data through a point cloud segmentation algorithm to determine candidate regions and initial obstacle types of obstacles in the candidate regions, wherein the detection region includes the candidate regions;

[0032] The grid division module is configured to divide the candidate region into grids and determine the grid point cloud information;

[0033] The target detection module is configured to determine the target detection results of obstacles in the candidate region based on the grid point cloud information;

[0034] The pose information acquisition module is configured to acquire the pose information of the obstacle when the target detection result indicates that the obstacle type is the initial obstacle type.

[0035] A third aspect of this application provides a controller, including:

[0036] The memory is configured to store instructions; and

[0037] The processor is configured to retrieve instructions from memory and, when executing the instructions, to implement the method described above for identifying obstacles.

[0038] A fourth aspect of this application provides a mobile device, including:

[0039] Data acquisition device for collecting data from mobile devices; and

[0040] Based on the controller described above.

[0041] In embodiments of this application, the data acquisition device includes:

[0042] LiDAR, installed at the front end of a mobile device, is used to collect point cloud data of the detection area;

[0043] An inertial measurement unit (IMU) is installed inside a mobile device to collect attitude data of the mobile device.

[0044] The above technical solution acquires target point cloud data of the detection area, and uses a point cloud segmentation algorithm to detect the target point cloud data to determine candidate areas and the initial obstacle types of obstacles within those candidate areas. The detection area includes the candidate areas. Furthermore, the candidate areas are divided into grids, and the grid point cloud information is determined. Based on the grid point cloud information, the target detection results of obstacles within the candidate areas are determined. If the target detection results indicate that the obstacle type matches the initial obstacle type, the obstacle's pose information is acquired. This application determines the initial obstacle type (i.e., the candidate area) based on point cloud data, and further determines the obstacle type based on grid division and other methods within the candidate areas. This improves the comprehensiveness and accuracy of obstacle identification by mobile devices in areas with complex road conditions, effectively avoiding the problem of incomplete data collection for complex road surfaces.

[0045] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0046] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0047] Figure 1 The schematic diagram illustrates a partial structural diagram of the rescue equipment provided according to an embodiment of this application;

[0048] Figure 2 A flowchart illustrating a method for identifying obstacles according to an embodiment of this application is shown schematically.

[0049] Figure 3 The diagram illustrates a determination of the concave-convex relationship of a plane according to an embodiment of this application.

[0050] Figure 4A diagram illustrating the determination of a trench according to an embodiment of this application is shown schematically.

[0051] Figure 5 A schematic diagram illustrating the determination of a slope according to an embodiment of this application is shown.

[0052] Figure 6 The diagram illustrates a judgment diagram of a vertical staircase according to an embodiment of this application;

[0053] Figure 7 A schematic block diagram of a controller according to an embodiment of this application is shown.

[0054] Explanation of reference numerals in the attached figures

[0055] 1. LiDAR Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0057] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0058] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0059] The obstacle identification method provided in this application can be applied to various rescue equipment, or to other types of mobile devices such as automobiles or mobile robots. For simplicity, the following description will primarily focus on the application of the method to rescue equipment.

[0060] Figure 1 The illustration shows a partial structural diagram of a rescue device provided according to an embodiment of this application. For example... Figure 1 As shown, the rescue equipment includes, but is not limited to, unmanned rescue vehicles, fire trucks, engineering rescue vehicles, or rescue-type engineering machinery vehicles. This rescue equipment may include a data acquisition device and a controller (not shown in the figure). The data acquisition device may include a lidar 1 and an inertial measurement unit (IMU) (not shown in the figure). The lidar 1 can be used to detect the position and velocity characteristics of the target. The IMU can be used to measure the three-axis attitude angles and acceleration of the rescue equipment. In one example, the lidar 1 can be positioned at the front end of the rescue equipment. The X-axis of the center point coordinates of the lidar 1 is parallel to the ground plane, and the Z-axis of the center point coordinates of the lidar 1 is parallel to the Z-axis of the center point coordinates of the front end of the rescue equipment, with no obstruction from the laser signal emission point of the lidar 1 by the rescue equipment. This ensures that the lidar 1 is not obstructed during data acquisition, allowing for more accurate data collection. The IMU can be positioned at the center of gravity of the rescue equipment.

[0061] Figure 2 A flowchart illustrating a method for identifying obstacles according to an embodiment of this application is shown schematically. Figure 2 As shown in the figure, this application provides a method for identifying obstacles, which may include the following steps.

[0062] Step 201: Obtain target point cloud data of the detection area;

[0063] Step 202: Detect the target point cloud data using a point cloud segmentation algorithm to determine the candidate region and the initial obstacle type of the obstacles in the candidate region, wherein the detection region includes the candidate region;

[0064] Step 203: Divide the candidate region into grids and determine the grid point cloud information;

[0065] Step 204: Determine the target detection results of obstacles in the candidate area based on the grid point cloud information.

[0066] Step 205: If the target detection result indicates that the obstacle type is the initial obstacle type, obtain the pose information of the obstacle.

[0067] This application acquires target point cloud data of the detection area, detects the target point cloud data using a point cloud segmentation algorithm, and determines candidate areas and the initial obstacle types of obstacles within these candidate areas. The detection area includes the candidate areas. Furthermore, the candidate areas are divided into grids, and grid point cloud information is determined. Based on the grid point cloud information, the target detection result of the obstacles within the candidate areas is determined. If the target detection result indicates that the obstacle type matches the initial obstacle type, the pose information of the obstacle is acquired. This application determines the initial type of the obstacle, i.e., the candidate area, based on point cloud data, and further determines the obstacle type based on methods such as grid division of the candidate areas. This improves the comprehensiveness and accuracy of obstacle identification by mobile devices in areas with complex road conditions, effectively avoiding the problem of incomplete data collection for complex road surfaces.

[0068] The implementation process of this application embodiment is illustrated below with some specific implementation methods. The processor can acquire target point cloud data of the detection area. The detection area is the area in the travel path of the mobile device where the lidar acquires point cloud data. During the travel of the mobile device, the lidar emits a laser signal forward from the mobile device. After the laser signal is reflected by the object being measured, it is received by the laser detector. By analyzing the return time of the laser signal after encountering the object, the relative distance to the object can be calculated. Using the three-dimensional coordinate information of the points on the surface of the object collected during this process, a three-dimensional model of the object can be obtained. Point cloud data is the three-dimensional coordinate information of the points on the surface of the object collected after the lidar emits the laser signal. Furthermore, a world coordinate system can be established beforehand, and point cloud data can be acquired based on this world coordinate system. While the lidar acquires the point cloud data of the detection area, the attitude data of the mobile device can be acquired through an inertial measurement unit, and the point cloud data acquired by the lidar can be corrected based on the attitude data to obtain the target point cloud data. By acquiring the target point cloud data, the problem of deviation in point cloud data caused by changes in the attitude of the mobile device can be reduced, improving the accuracy of obstacle recognition.

[0069] In this embodiment, when determining the type of obstacle, candidate regions and the initial obstacle types of obstacles within these regions can be preliminarily determined. Candidate regions are areas within the detection region where obstacles may exist, determined based on target point cloud data. Obstacle types include, but are not limited to, ravines, slopes, and vertical steps. The processor can preliminarily determine the obstacle type using a point cloud segmentation algorithm. A point cloud segmentation algorithm divides a point cloud based on its spatial, geometric, and textural features, allowing point clouds with similar features to be grouped into the same region. In one example, the point cloud segmentation algorithm can be the Locally Convex Connected Patches (LCCP) algorithm. The LCCP algorithm consists of two parts: 1. Oversegmentation based on hyper-volume clustering; 2. Re-clustering based on concavity and convexity in addition to hyper-volume clustering. The LCCP algorithm only uses spatial and normal information to determine the candidate regions and the initial obstacle types of obstacles within them. The initial obstacle types include ravines, slopes, and vertical steps. The point cloud segmentation algorithm can determine the candidate regions and the initial obstacle types of obstacles within them.

[0070] After determining the candidate region, the processor can divide the candidate region into grids according to a preset initial grid size. The preset initial grid size is a pre-defined grid size value. The processor can divide the candidate region into multiple initial grids according to the preset grid size value. For example, the initial grid size can be set to 0.5 meters, and the candidate region can be divided into multiple initial grids at equal intervals of 0.5 meters. The initial grids contain target point cloud data. After the candidate region is gridded, the grid point cloud information is determined based on the target point cloud data within the grids. The grid point cloud information includes, but is not limited to, at least one of the following: maximum elevation, minimum elevation, standard deviation and point cloud density, point cloud trend, and projection skewness distribution.

[0071] Given the grid point cloud information, the processor can determine the target detection result for obstacles in the candidate region. In one example, after acquiring the target point cloud data, the processor can perform grid generation and determine the grid point cloud information. By determining the meta-grids and corresponding neighboring grids in multiple initial grids, the meta-grids and neighboring grids are cross-compared to obtain the target detection result. The target detection result includes the confidence that the obstacle type is the initial obstacle type. If the target detection result indicates that the obstacle type is the initial obstacle type, the pose information of the obstacle is obtained. The pose information refers to the position and orientation information of the obstacle. By determining the obstacle type twice, the obstacle identification becomes more accurate, and false identification is reduced.

[0072] The above technical solution acquires target point cloud data of the detection area, and uses a point cloud segmentation algorithm to detect the target point cloud data to determine candidate areas and the initial obstacle types of obstacles within those candidate areas. The detection area includes the candidate areas. Furthermore, the candidate areas are divided into grids, and the grid point cloud information is determined. Based on the grid point cloud information, the target detection results of obstacles within the candidate areas are determined. If the target detection results indicate that the obstacle type matches the initial obstacle type, the obstacle's pose information is acquired. This application reduces the complexity of obstacle identification by mobile devices and improves the comprehensiveness and accuracy of obstacle identification by mobile devices in areas with complex post-disaster road conditions.

[0073] In this embodiment of the application, the mobile device includes a lidar and an inertial measurement unit. Step 201, acquiring target point cloud data of the detection area, may include:

[0074] Initial point cloud data of the detection area is acquired using lidar;

[0075] Attitude data of mobile devices are acquired through inertial measurement units;

[0076] The initial point cloud data is corrected based on the attitude data to determine the target point cloud data.

[0077] Specifically, the processor can correct the initial point cloud data acquired by the lidar, reducing the problem of point cloud data deviation caused by changes in the attitude of the mobile device. While the lidar acquires point cloud data of the detection area, the attitude data of the mobile device can be acquired through an inertial measurement unit, and the point cloud data acquired by the lidar can be corrected based on the attitude data to obtain target point cloud data. Among them, the attitude data includes the yaw angle, pitch angle and roll angle of the mobile device. By acquiring target point cloud data, the problem of point cloud data deviation caused by changes in the attitude of the mobile device can be reduced, and the accuracy of obstacle recognition can be improved. The processor can correct the point cloud data acquired by the lidar through a rotation and translation matrix to obtain target point cloud data based on the world coordinate system. The target point cloud data can be calculated by formula (1):

[0078]

[0079] Among them, R XYZ (γ, β, α) is the total rotation transformation matrix, α is the yaw angle, β is the pitch angle, γ is the roll angle, and R... z (α) is the transformation matrix for rotating about the Z-axis by an angle α, R Y (β) is the transformation matrix for rotating about the Y-axis by an angle β, R X (γ) is the transformation matrix for rotating about the X-axis by an angle γ.

[0080] The processor can process the target point cloud data to reduce the amount of point cloud data and improve the speed of subsequent algorithms. In one example, an outlier removal threshold can be preset for the target point cloud data to filter outliers. The outlier removal threshold can be calculated by statistically analyzing the neighborhood of each point and calculating its average distance to all neighboring points. Assuming the result is a Gaussian distribution, the shape of which is determined by the mean and standard deviation, points whose average distance is outside the standard range can be defined as outliers and removed from the target point cloud data. This standard range is determined by the global average distance and variance. Furthermore, the point cloud can be downsampled to achieve a filtering effect, reducing the number of points while maintaining the shape characteristics of the point cloud to improve the speed of the algorithm. Preferably, a VoxelGrid filter can be used to downsample the point cloud. By processing the target point cloud data, the amount of point cloud data can be reduced, and the speed of subsequent algorithms can be improved.

[0081] In this embodiment of the application, step 202, detecting target point cloud data through a point cloud segmentation algorithm to determine candidate regions and the initial obstacle types of obstacles in the candidate regions, may include:

[0082] Based on concavity / convexity, the detection region is segmented using a point cloud segmentation algorithm to obtain multiple faces. The target point cloud data is used to determine if three consecutive faces have heights lower than a first preset height. If three consecutive faces are found to be lower than the first preset height, it is determined whether any two adjacent faces within these three faces are concave. If all adjacent faces are determined to be concave, the region enclosed by these three faces and the ground plane is identified as a candidate region, and the initial obstacle type is determined to be a ditch.

[0083] Obtain the normal vector of each point in the target point cloud data of the detection area and the normal vector of the ground plane; determine the area enclosed by the target point cloud data and the ground plane where the angle between the normal vector and the ground plane normal vector is less than 90 degrees as the candidate area, and determine the initial obstacle type as a slope; or

[0084] Based on concavity and convexity, the detection area is segmented using a point cloud segmentation algorithm to obtain multiple faces. Based on the target point cloud data, it is determined whether there are three consecutive faces whose heights are all higher than a second set height. If it is determined that there are three consecutive faces whose heights are higher than the second set height, it is determined whether the relationship between two adjacent faces in the three consecutive faces is one concave and one convex. If it is determined that the relationship between two adjacent faces in the three consecutive faces is one concave and one convex, the area enclosed by the three consecutive faces and the ground plane is determined as a candidate area, and the initial obstacle type of the obstacle is determined to be a vertical staircase.

[0085] Specifically, based on concavity and convexity, the detection region can be segmented using point cloud segmentation algorithms to obtain multiple faces. When performing super-volume clustering on the point cloud, for over-segmented point clouds, it is necessary to calculate the concavity and convexity relationships between different faces. These relationships can be determined using the Extended Convexity Criterion (CC) and the Sanity Criterion (SC). Figure 3 This diagram schematically illustrates a method for determining the concavity and convexity relationship of a plane according to an embodiment of this application. For example... Figure 3 As shown, assume there are two adjacent planes, plane 1 and plane 2. Let α1 be the angle between the centerline vectors of plane 1 and the adjacent plane, and α2 be the angle between the centerline vectors of plane 2 and the adjacent plane. When α1 > α2, the relationship between the planes is concave; when α1 < α2, the relationship is convex. In practical applications, a limit can be set to filter out small misjudgments of concavity / convexity. For example, a limit of 5 degrees can be set, so that only when α1 - α2 > 5 degrees can the relationship between the planes be determined to be concave. Furthermore, if the concavity / convexity relationship of two planes is the same, then the concavity / convexity relationship of their common adjacent plane is the same as that of the two planes themselves. If there are cases where adjacent planes are not connected, the CC criterion cannot accurately determine this; therefore, the processor can use the SC criterion to determine whether there is a single plane among the adjacent planes. Figure 3 As shown, θ is the angle between the cross product of the plane normal vectors and the vector connecting the centers of adjacent planes. When determining θ, the smaller angle is chosen as θ. Whether there is a single plane among two adjacent planes is related to the angle θ. When the angle θ is 0 degrees, there is a single plane among two adjacent planes. The cross product of the plane normal vectors can be determined by formula (2), and then θ is determined according to the angle between the cross product of the plane normal vectors and the vector connecting the centers of adjacent planes:

[0086]

[0087] in, The cross product of the plane normal vectors. Let be the normal vector of plane 1. Let be the normal vector of plane 2.

[0088] The centerline vectors of adjacent planes can be determined using formula (3):

[0089]

[0090] in, The vectors of the centerlines of adjacent planes Let be the vector pointing towards the center in plane 1. Let be the vector pointing towards the center in plane 2.

[0091] To determine if an obstacle's initial obstacle type is a ravine, it must satisfy the condition that two adjacent faces out of three consecutive faces are concave. Figure 4 A schematic diagram illustrating the determination of a trench according to an embodiment of this application is shown. Figure 4 As shown in this embodiment, the system determines whether three consecutive faces have heights lower than a first preset height based on the target point cloud data. Since the first preset height is used to determine whether an obstacle is a ditch, it is set to a negative value. If three consecutive faces have heights lower than the first preset height, adjacent faces are compared, and the convex-concave relationship between the adjacent faces is determined using the LCCP algorithm. When it is detected that two adjacent faces of three consecutive faces in the detection area are concave, the area enclosed by the three consecutive faces and the ground plane is determined as a candidate area, and the initial obstacle type in the candidate area is determined to be a ditch. The area of ​​the enclosed area should be larger than a preset area to filter out small ditches.

[0092] To determine that the initial obstacle type is a slope, the angle between the normal vector of each point in the target point cloud data and the normal vector of the ground plane must be greater than 0 degrees and less than 90 degrees. Figure 5 A schematic diagram illustrating the determination of a slope according to an embodiment of this application is shown. Figure 5 As shown in this embodiment, the processor can acquire the normal vector of each point in the target point cloud data of the detection area and the normal vector of the ground plane. The processor can determine the normal vector of the ground plane using an inertial measurement unit. Thus, the angle β between the normal vector and the normal vector of the ground plane can be obtained. If this angle β is greater than 0 degrees and less than 90 degrees, the area enclosed by the target point cloud data and the ground plane with an angle β greater than 0 degrees and less than 90 degrees is determined as a candidate area, and the initial obstacle type of obstacles in the candidate area is determined to be a slope. For example, if multiple consecutive points in the target point cloud data have a 45-degree angle between their normal vectors and the normal vector of the ground plane, the initial obstacle type of the obstacle can be determined to be a slope. The area of ​​the enclosed region should be greater than a preset area to filter out small slopes.

[0093] To determine if an initial obstacle type is a vertical staircase, it must satisfy the condition that there are three consecutive surfaces in which two adjacent surfaces are in a concave relationship and one is in a convex relationship. Figure 6 This diagram schematically illustrates a determination of a vertical staircase according to an embodiment of this application. For example... Figure 6As shown in this embodiment, the system determines whether three consecutive faces have heights higher than a second preset height based on the target point cloud data. Since the second preset height is used to determine whether an obstacle is a vertical staircase, it is set to a positive value. If three consecutive faces have heights higher than the second preset height, the system determines whether the relationship between two adjacent faces is concave and convex. If the relationship is concave and convex, the area enclosed by the three consecutive faces and the ground plane is defined as a candidate area, and the initial obstacle type is determined to be a vertical staircase. The area of ​​the enclosed region should be larger than a preset area to filter out small vertical staircases.

[0094] This application embodiment first determines the initial obstacle type of the obstacle through a point cloud segmentation algorithm, and then further determines the obstacle type in the candidate region based on the grid point cloud information. This can accurately determine the obstacle type and reduce the possibility of misjudgment by the mobile device.

[0095] In this embodiment of the application, the grid point cloud information may include at least one of the following:

[0096] Maximum elevation, minimum elevation, standard deviation, point cloud density, point cloud trend, and skewed projection distribution.

[0097] Specifically, grid point cloud information can be determined through target point cloud data within the grid. After acquiring the target point cloud data, the processor can perform grid division and determine the grid point cloud information. Grid point cloud information includes at least one of the following: maximum elevation, minimum elevation, standard deviation, point cloud density, point cloud trend, and projected skewness distribution. Elevation refers to the height of a point relative to a reference plane. The maximum elevation refers to the absolute height of the highest point of the target point cloud data relative to the reference plane. The minimum elevation refers to the height of the lowest point of the target point cloud data relative to the reference plane. The reference plane can be selected as the ground plane. The standard deviation reflects the dispersion of the target point cloud data. Point cloud density is an indicator of data resolution, and in this embodiment, it reflects the density of the target point cloud data.

[0098] Point cloud trends can be determined using trend tests. In one example, the Cox-Stuart test can be used to determine the point cloud trend. The Cox-Stuart test is based on the sign test, which directly considers the trend of data change. If the data shows an upward trend, the values ​​of later data points will be significantly larger than those of earlier data points; conversely, if the data shows a downward trend, the values ​​of later data points will be significantly smaller than those of earlier data points. In other words, the overall trend of the target point cloud data can be determined by the sign of the difference in coordinates between each point in the target point cloud data. Projection skewness refers to the skewed distribution of the projection of the target point cloud data. Determining the grid point cloud information facilitates further determination of the type of obstacle in the candidate region.

[0099] In this embodiment of the application, step 204, determining the target detection result of obstacles in the candidate region based on the grid point cloud information, may include:

[0100] Select the meta-mesh and the corresponding neighboring mesh in the candidate region to obtain the mesh combination composed of the meta-mesh and the neighboring mesh;

[0101] The target detection result is obtained by cross-comparing the meta-mesh with the neighboring mesh. The target detection result includes the confidence that the obstacle type is the initial obstacle type.

[0102] If the confidence level meets the preset confidence level conditions, the type of obstacle is determined to be the initial obstacle type.

[0103] Specifically, the processor can first divide the candidate region into grids according to a preset initial grid size. The preset initial grid size is a pre-defined grid size value. The processor can divide the candidate region into multiple grids according to the preset grid size value. After grid division, the processor can select a meta-grid and its corresponding neighboring grids from these grids, thus obtaining a grid combination consisting of the meta-grid and the neighboring grids. The meta-grid and the neighboring grids are cross-compared to obtain the target detection result. This target detection result includes a confidence score that the obstacle type matches the initial obstacle type. The confidence score that the obstacle type matches the initial obstacle type can be determined based on the grid point cloud information of the grid combination being cross-compared.

[0104] In one specific embodiment, the processor can divide the candidate region into multiple grids, select two adjacent grids from these grids, and choose the first grid as the meta-grid and the other grid as the neighboring grid. The processor can determine the grid point cloud information using the target point cloud data within the two grids. With the grid point cloud information determined, the processor cross-compares the point cloud information of the meta-grid and the neighboring grid in a predetermined order. The processor can determine the point cloud trend using trend testing. After determining the point cloud trend, the processor determines the confidence level based on the comparison rule function and proceeds to the next step. For example, when the point cloud trend meets the conditions of a planar trend or a rapidly decreasing trend, the processor can determine whether the preset conditions are met based on the minimum value and standard deviation of the point cloud within the grid in a planar trend or the maximum value and projected skewness distribution of the point cloud within the grid in a rapidly decreasing trend. If the preset conditions are met, the confidence level is updated, and grids are selected again as a grid combination for cross-comparison. When selecting grids again as a grid combination, grids whose confidence level has reached the preset confidence level will not be selected as meta-grids. Based on the comparison rule function, the combined grid point cloud information of the judgment grid can be used to obtain the confidence level that the obstacle type is the initial obstacle type. If the confidence level meets the preset confidence condition, the obstacle type is determined to be the initial obstacle type. By making a secondary judgment on the obstacle type based on the grid point cloud information and its confidence level, the accuracy of obstacle identification can be improved.

[0105] In this embodiment of the application, step 204, determining the target detection result of obstacles in the candidate region based on the grid point cloud information, may further include:

[0106] If the confidence level of the grid combination does not meet the preset confidence level, increase the number of neighboring grids and return to the step of cross-comparing the meta grid with the neighboring grids to obtain the target detection result, until the confidence level of the grid combination meets the preset confidence level.

[0107] The number of neighboring grids shall not exceed the preset number.

[0108] Specifically, if the confidence level of a grid combination does not meet the preset confidence level condition, the processor can increase the number of neighboring grids for iteration. Increasing the number of neighboring grids for iteration means adding more neighboring grids to replace the original neighboring grids in the grid combination, and cross-comparing them with the meta-grid until the confidence level reaches the preset confidence level condition or the number of neighboring grids used in iteration reaches a preset number. By increasing the number of neighboring grids to iterate on grid combinations whose confidence level does not meet the preset condition, the grid point cloud information of the grid combination can be further determined, and the confidence level can be updated, thereby reducing errors in the process of determining obstacle types.

[0109] In this embodiment of the application, step 204, determining the target detection result of obstacles in the candidate region based on the grid point cloud information, may further include:

[0110] If the number of neighboring grids reaches a preset number and the confidence level of the grid combination still does not meet the preset confidence level condition, it is determined that the target obstacle type of the grid combination is not the initial obstacle type.

[0111] Specifically, the processor can preset a number of neighboring grids. If the confidence level of a grid combination does not meet the preset confidence level, the processor can increase the number of neighboring grids for iteration. Increasing the number of neighboring grids for iteration means increasing the number of neighboring grids, replacing the neighboring grids in the original grid combination, and cross-comparing them with the meta-grid until the confidence level reaches the preset confidence level or the number of neighboring grids in the iteration reaches the preset number. If the number of neighboring grids reaches the preset number and the confidence level of the grid combination still does not meet the preset confidence level, it is determined that the target obstacle type within the grid combination is not the initial obstacle type. By determining the preset number of neighboring grids, the problem of iterating the grid combination indefinitely when the confidence level does not meet the preset confidence level can be avoided.

[0112] In one specific embodiment, when the initial obstacle type for obstacle detection is a slope, the normal vector of each point in the point cloud data within the candidate region forms a certain angle with the ground normal vector. Regions with angled normal vectors within the candidate region are identified, and the existence of a slope is determined using the elevation difference from a linear scan. If there is an elevation difference between the mean of the scanned grid and the difference in the slope of their plane fitting is within a threshold, the existence of a slope is confirmed. The slope location can be determined through iteration, thereby assisting the mobile device in planning its power and movement strategies.

[0113] When the initial obstacle type detected is a vertical, traversable staircase, the convexity and soundness criteria are extended, and the angle between the line connecting the centers of adjacent grids and the normal vector is used to determine the convexity / concavity relationship between two grids. Furthermore, the number of neighboring grids can be increased to iteratively fuse the grid combinations, continuously comparing the maximum elevation value in the grid point cloud information with the recorded ground height. When there is a certain elevation difference between the maximum elevation value and the ground, the obstacle is determined to be a vertical staircase, and the height of the vertical staircase is recorded. Moreover, obstacle-crossing instructions can be formulated using special devices on mobile devices.

[0114] This application provides a device for identifying obstacles, which may include:

[0115] The point cloud data acquisition module is configured to acquire target point cloud data of the detection area;

[0116] The initial detection module is configured to detect target point cloud data through a point cloud segmentation algorithm to determine candidate regions and initial obstacle types of obstacles in the candidate regions, wherein the detection region includes the candidate regions;

[0117] The grid division module is configured to divide the candidate region into grids and determine the grid point cloud information;

[0118] The target detection module is configured to determine the target detection results of obstacles in the candidate region based on the grid point cloud information;

[0119] The pose information acquisition module is configured to acquire the pose information of the obstacle when the target detection result indicates that the obstacle type is the initial obstacle type.

[0120] Specifically, the point cloud data acquisition module can acquire target point cloud data of the detection area. In this embodiment, the initial detection module can first preliminarily determine the initial obstacle type of the obstacle in the candidate area. The initial detection module can first preliminarily determine the initial obstacle type of the obstacle through a point cloud segmentation algorithm. The mesh division module can first divide the candidate area into meshes according to a preset initial mesh size. The preset initial mesh size is a pre-set value for the mesh size. The processor can divide the candidate area into multiple initial meshes according to the preset mesh size value. The initial meshes contain target point cloud data. When the candidate area is meshed, the mesh point cloud information is determined based on the target point cloud data within the mesh. The point cloud information includes at least one of the following: maximum elevation, minimum elevation, standard deviation and point cloud density, point cloud trend, and projection skewness distribution. When the mesh point cloud information is determined, the target detection module can determine the target detection result of the obstacle in the candidate area. When the target detection result indicates that the obstacle type is the initial obstacle type, the pose information acquisition module acquires the pose information of the obstacle.

[0121] After acquiring the target point cloud data, the point cloud data acquisition module can perform mesh generation and determine the mesh point cloud information using the mesh generation module. The target detection module determines the meta-mesh and corresponding neighboring meshes in multiple initial meshes. The meta-mesh and neighboring meshes are cross-compared to obtain the target detection result. The target detection result includes the confidence level that the obstacle type matches the initial obstacle type. If the target detection result indicates that the obstacle type matches the initial obstacle type, the pose information acquisition module acquires the obstacle's pose information. This secondary determination of the obstacle type makes obstacle identification more accurate and reduces false identifications.

[0122] Figure 7 A schematic block diagram of a controller according to an embodiment of this application is shown. Figure 7 As shown in the figure, this application provides a controller that may include:

[0123] Memory 710 is configured to store instructions; and

[0124] The processor 720 is configured to retrieve instructions from memory and, when executing the instructions, to implement the method for identifying obstacles as described above.

[0125] Specifically, in this embodiment of the application, the processor 720 can be configured to:

[0126] Acquire target point cloud data of the detection area;

[0127] The target point cloud data is detected by a point cloud segmentation algorithm to determine the candidate region and the initial obstacle type of the obstacle in the candidate region, wherein the detected region contains the candidate region;

[0128] The candidate region is divided into grids and the grid point cloud information is determined;

[0129] The target detection results of obstacles in the candidate area are determined based on the grid point cloud information;

[0130] If the target detection result indicates that the obstacle type is the initial obstacle type, obtain the obstacle pose information.

[0131] Furthermore, the processor 720 can also be configured as follows:

[0132] Initial point cloud data of the detection area is acquired using lidar;

[0133] Attitude data of mobile devices are acquired through inertial measurement units;

[0134] The initial point cloud data is corrected based on the attitude data to determine the target point cloud data.

[0135] Furthermore, the processor 720 can also be configured as follows:

[0136] Based on concavity / convexity, the detection region is segmented using a point cloud segmentation algorithm to obtain multiple faces. The target point cloud data is used to determine if three consecutive faces have heights lower than a first preset height. If three consecutive faces are found to be lower than the first preset height, it is determined whether any two adjacent faces within these three faces are concave. If all adjacent faces are determined to be concave, the region enclosed by these three faces and the ground plane is identified as a candidate region, and the initial obstacle type is determined to be a ditch.

[0137] Obtain the normal vector of each point in the target point cloud data of the detection area and the normal vector of the ground plane; determine the area enclosed by the target point cloud data and the ground plane where the angle between the normal vector and the ground plane normal vector is less than 90 degrees as the candidate area, and determine the initial obstacle type as a slope; or

[0138] Based on concavity and convexity, the detection area is segmented using a point cloud segmentation algorithm to obtain multiple faces. Based on the target point cloud data, it is determined whether there are three consecutive faces whose heights are all higher than a second set height. If it is determined that there are three consecutive faces whose heights are higher than the second set height, it is determined whether the relationship between two adjacent faces in the three consecutive faces is one concave and one convex. If it is determined that the relationship between two adjacent faces in the three consecutive faces is one concave and one convex, the area enclosed by the three consecutive faces and the ground plane is determined as a candidate area, and the initial obstacle type of the obstacle is determined to be a vertical staircase.

[0139] In this embodiment of the application, the grid point cloud information may further include at least one of the following:

[0140] Maximum elevation, minimum elevation, standard deviation, point cloud density, point cloud trend, and skewed projection distribution.

[0141] Furthermore, the processor 720 can also be configured as follows:

[0142] Select the meta-mesh and the corresponding neighboring mesh in the candidate region to obtain the mesh combination composed of the meta-mesh and the neighboring mesh;

[0143] The target detection result is obtained by cross-comparing the meta-mesh with the neighboring mesh. The target detection result includes the confidence that the obstacle type is the initial obstacle type.

[0144] If the confidence level meets the preset confidence level conditions, the type of obstacle is determined to be the initial obstacle type.

[0145] Furthermore, the processor 720 can also be configured as follows:

[0146] If the confidence level of the grid combination does not meet the preset confidence level, increase the number of neighboring grids and return to the step of cross-comparing the meta grid with the neighboring grids to obtain the target detection result, until the confidence level of the grid combination meets the preset confidence level.

[0147] The number of neighboring grids shall not exceed the preset number.

[0148] Furthermore, the processor 720 can also be configured as follows:

[0149] If the number of neighboring grids reaches a preset number and the confidence level of the grid combination still does not meet the preset confidence level condition, it is determined that the target obstacle type of the grid combination is not the initial obstacle type.

[0150] The above technical solution acquires target point cloud data of the detection area, and uses a point cloud segmentation algorithm to detect the target point cloud data to determine candidate areas and the initial obstacle types of obstacles within those candidate areas. The detection area includes the candidate areas. Furthermore, the candidate areas are divided into grids, and the grid point cloud information is determined. Based on the grid point cloud information, the target detection results of obstacles within the candidate areas are determined. If the target detection results indicate that the obstacle type matches the initial obstacle type, the obstacle's pose information is acquired. This application reduces the complexity of obstacle identification by mobile devices and improves the comprehensiveness and accuracy of obstacle identification by mobile devices in areas with complex post-disaster road conditions.

[0151] This application embodiment also provides a mobile device, which may include:

[0152] Data acquisition device for collecting data from mobile devices; and

[0153] Based on the controller described above.

[0154] In this embodiment of the application, the data acquisition device may include:

[0155] LiDAR, installed at the front end of a mobile device, is used to collect point cloud data of the detection area;

[0156] An inertial measurement unit (IMU) is installed inside a mobile device to collect attitude data of the mobile device.

[0157] Specifically, such as Figure 1 As shown, taking a mobile device as an example of rescue equipment, the rescue equipment may include a data acquisition device and a controller (not shown in the figure). The data acquisition device may include a lidar 1 and an inertial measurement unit (IMU) (not shown in the figure). The lidar 1 can be used to detect the position and velocity characteristics of the target. The IMU is used to measure the three-axis attitude angles and acceleration of the rescue equipment. In one example, the lidar 1 can be positioned at the front end of the rescue equipment. The X-axis of the center point coordinates of the lidar 1 is parallel to the ground plane, and the Z-axis of the center point coordinates of the lidar 1 is parallel to the Z-axis of the center point coordinates of the front end of the rescue equipment, and the rescue equipment does not obstruct the laser signal emission of the lidar 1. The IMU can be positioned at the center of gravity of the rescue equipment.

[0158] In this embodiment, the method of setting the position of the data acquisition device described above ensures that the lidar 1 is not obstructed by rescue equipment when acquiring data, thus enabling more accurate data acquisition. Furthermore, it improves the reliability of the measurement data from the inertial measurement unit.

[0159] This application also provides a machine-readable storage medium storing instructions that cause a machine to perform the above-described method for identifying obstacles.

[0160] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0162] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0164] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0165] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0166] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0167] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0168] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for identifying obstacles, characterized in that, Applied to mobile devices, the method includes: Acquire target point cloud data of the detection area; The target point cloud data is detected by a point cloud segmentation algorithm to determine candidate regions and the initial obstacle types of obstacles in the candidate regions, wherein the detected regions include the candidate regions; The candidate region is divided into grids and the grid point cloud information is determined; The target detection result of the obstacle in the candidate region is determined based on the grid point cloud information; If the target detection result indicates that the obstacle type is the initial obstacle type, the pose information of the obstacle is obtained. The initial obstacle type includes at least one of ravine, slope and vertical steps. The target detection result is obtained after secondary determination of the initial obstacle type. The step of determining the target detection result of the obstacle in the candidate region based on the grid point cloud information includes: In the candidate region, a meta-mesh and its corresponding neighboring mesh are selected to obtain a mesh combination consisting of the meta-mesh and the neighboring mesh; The target detection result is obtained by cross-comparing the meta-mesh with the neighborhood mesh, and the target detection result includes the confidence that the type of the obstacle is the initial obstacle type; If the confidence level meets the preset confidence level conditions, the type of the obstacle is determined to be the initial obstacle type.

2. The method according to claim 1, characterized in that, The mobile device includes a lidar and an inertial measurement unit, and the acquisition of target point cloud data of the detection area includes: The initial point cloud data of the detection area is acquired using the lidar. The attitude data of the mobile device is acquired through the inertial measurement element; The initial point cloud data is corrected based on the attitude data to determine the target point cloud data.

3. The method according to claim 1, characterized in that, The step of detecting the target point cloud data using a point cloud segmentation algorithm to determine candidate regions and the initial obstacle types of obstacles within those candidate regions includes: Based on the concavity / convexity property, the detection region is segmented using a point cloud segmentation algorithm to obtain multiple faces. The target point cloud data is used to determine if three consecutive faces have heights lower than a first preset height. If three consecutive faces are found to be lower than the first preset height, it is determined whether any two adjacent faces of these three consecutive faces are concave. If any adjacent faces of these three consecutive faces are found to be concave, the region enclosed by these three consecutive faces and the ground plane is defined as a candidate region, and the initial obstacle type is determined to be a ravine. Obtain the normal vector of each point in the target point cloud data of the detection area and the normal vector of the ground plane; determine the area enclosed by the target point cloud data and the ground plane where the angle between the normal vector and the normal vector of the ground plane is less than 90 degrees as a candidate area, and determine the initial obstacle type of the obstacle as a slope; or Based on concavity and convexity, the detection area is segmented using a point cloud segmentation algorithm to obtain multiple faces; based on the target point cloud data, it is determined whether there are three consecutive faces whose heights are all higher than a second set height; if it is determined that there are three consecutive faces whose heights are higher than the second set height, it is determined whether the relationship between two adjacent faces in the three consecutive faces is one concave and one convex; if it is determined that the relationship between two adjacent faces in the three consecutive faces is one concave and one convex, the area enclosed by the three consecutive faces and the ground plane is determined as a candidate area, and the initial obstacle type of the obstacle is determined to be a vertical staircase.

4. The method according to claim 1, characterized in that, The grid point cloud information includes at least one of the following: Maximum elevation, minimum elevation, standard deviation, point cloud density, point cloud trend, and skewed projection distribution.

5. The method according to claim 1, characterized in that, The step of determining the target detection result of the obstacle in the candidate region based on the grid point cloud information further includes: If the confidence level of the grid combination does not meet the preset confidence level condition, the number of the neighboring grids is increased, and the step of cross-comparing the meta-grid with the neighboring grids to obtain the target detection result is returned, until the confidence level of the grid combination meets the preset confidence level condition. The number of neighborhood grids does not exceed a preset number.

6. The method according to claim 5, characterized in that, The step of determining the target detection result of the obstacle in the candidate region based on the grid point cloud information further includes: If the number of neighboring grids reaches the preset number and the confidence level of the grid combination still does not meet the preset confidence level condition, it is determined that the target obstacle type of the grid combination is not the initial obstacle type.

7. A device for identifying obstacles, characterized in that, include: The point cloud data acquisition module is configured to acquire target point cloud data of the detection area; An initial detection module is configured to detect the target point cloud data using a point cloud segmentation algorithm to determine candidate regions and initial obstacle types of obstacles in the candidate regions, wherein the detection region includes the candidate regions. The grid division module is configured to divide the candidate region into grids and determine the grid point cloud information; The target detection module is configured to determine the target detection result of the obstacle in the candidate region based on the grid point cloud information; The pose information acquisition module is configured to acquire the pose information of the obstacle when the target detection result indicates that the type of the obstacle is the initial obstacle type. The initial obstacle type includes at least one of ravine, slope and vertical steps. The target detection result is the result obtained after secondary determination of the initial obstacle type. The step of determining the target detection result of the obstacle in the candidate region based on the grid point cloud information includes: In the candidate region, a meta-mesh and its corresponding neighboring mesh are selected to obtain a mesh combination consisting of the meta-mesh and the neighboring mesh; The target detection result is obtained by cross-comparing the meta-mesh with the neighborhood mesh, and the target detection result includes the confidence that the type of the obstacle is the initial obstacle type; If the confidence level meets the preset confidence level conditions, the type of the obstacle is determined to be the initial obstacle type.

8. A controller, characterized in that, include: The memory is configured to store instructions; as well as The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the method for identifying obstacles according to any one of claims 1 to 6.

9. A mobile device, characterized in that, include: Data acquisition device, used to collect data from mobile devices; as well as The controller according to claim 8.

10. The mobile device according to claim 9, characterized in that, The data acquisition device includes: A lidar, installed at the front end of the mobile device, is used to collect point cloud data of the detection area; An inertial measurement unit (IMU) is installed inside the mobile device to collect attitude data of the mobile device.

Images