Method and device for detecting drivable area

Abstract

This application discloses a method and apparatus for detecting a drivable area, belonging to the field of autonomous driving technology. The method comprises: obtaining a laser point cloud of a detectable area of ​​an autonomous driving platform; generating a grid map corresponding to the laser point cloud; determining an initial ground grid within each grid of the grid map; determining a target ground grid within the grid map based on the determined initial ground grid; and determining the area corresponding to the determined initial ground grid and the target ground grid as the drivable area. The method provided in this application can relatively accurately detect the drivable area in inclement weather.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a method and device for detecting a drivable area. Background Art

[0002] In autonomous driving, the detection of the vehicle's drivable area is one of the key technologies in unmanned driving technology.

[0003] At present, autonomous vehicles can detect drivable areas by using cameras installed on the autonomous vehicle to capture images of a certain area in front of the vehicle, and then use machine learning models and computer vision technology to identify the images, determine the drivable roads and obstacles that affect the vehicle's driving, and also identify objects such as trees that do not affect the vehicle's driving.

[0004] In the process of implementing this application, the inventors discovered that the related art has at least the following problems:

[0005] The images captured by the above method in adverse environments such as heavy fog and heavy rain cannot accurately identify the drivable area and obstacles. Therefore, the detection accuracy of the above method in adverse environments such as heavy fog and heavy rain is low. Summary of the Invention

[0006] The embodiments of the present application provide a method and apparatus for detecting a drivable area to overcome the problem of large errors in drivable area detection existing in the related art.

[0007] In a first aspect, the present application provides a method for detecting a drivable area, comprising:

[0008] Obtaining laser point clouds of the detectable area of ​​the autonomous driving platform;

[0009] Generating a grid map corresponding to the laser point cloud, wherein each laser point in the laser point cloud corresponds to a grid in the grid map;

[0010] determining an initial ground grid in each grid of the grid map;

[0011] Determining a target ground grid in the grid map based on the initial ground grid;

[0012] The area corresponding to the determined initial ground grid and the target ground grid is used as a drivable area.

[0013] In the embodiments of the present application, the autonomous driving platform can be a self-driving car, an intelligent robot, or the like, and the grid map can be a polar coordinate grid or a rectangular coordinate grid. The autonomous driving platform can be equipped with a laser radar (LIDAR). To effectively detect obstacles in the direction of travel, the LIDAR is typically mounted on the top or front of the autonomous driving platform. While the autonomous driving platform is in motion, the LIDAR emits a laser beam toward a detectable area of ​​the autonomous driving platform's surroundings to scan the surrounding environment. The laser beam reflects off the ground, trees, obstacles, buildings, and the like, and the LIDAR receives the reflected echoes, thereby generating a laser point cloud of the detectable area. A grid map is then generated, ensuring that each laser point falls within a grid of the grid map. An initial grid is then determined within the grid map. Using these initial ground grids as a reference, a target ground grid is determined within the grid map. Finally, the area corresponding to the initial and target ground grids is defined as the drivable area. The above method first acquires a laser point cloud of the area, then rasterizes it. An initial ground grid is then determined within the raster. The initial ground grid is then used as a reference to determine the target ground grid within the raster. This method, which uses lasers to detect drivable areas, can adapt to harsh environments and avoid false detections that can occur with image recognition technology in inclement weather.

[0014] In a possible implementation, before determining the initial ground grid in the grid of the grid map, the method further includes:

[0015] For each grid in the grid map, the laser points in the grid are layered to obtain at least one point layer.

[0016] In the solution shown in the embodiments of this application, for each grid containing laser points, the laser points in the grid can be first layered based on the preset drivable height and the height of the laser points in the grid. The collection of laser points in each layer can be called a point layer. The principle of layering is to ensure that the minimum height difference of laser points between different point layers is greater than the preset drivable height. In this way, when the grid is subsequently inspected, only the point layer with the lowest average height can be inspected, without having to inspect all laser points in the grid, which improves inspection speed.

[0017] In a possible implementation, determining an initial ground grid in each grid of the grid map includes:

[0018] An initial ground grid is determined in each grid of the grid map based on the distance between the grid map and the autonomous driving platform and the height of the laser point in the target point layer having the smallest average height of the laser points in the grid map, the target point layer belonging to the at least one point layer.

[0019] The solution shown in the embodiment of the present application can determine the initial ground grid in each grid in the grid map based on the distance between the grid and the automatic driving platform, and the height of the laser point in the target point layer with the smallest average height of the grid.

[0020] In a possible implementation, determining a target ground grid in the grid map based on the initial ground grid includes:

[0021] The initial ground grid is used as a reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the initial ground grid is used as the reference ground height corresponding to the initial ground grid;

[0022] The grid of the reference grid that meets the preset proximity condition is used as the grid to be detected;

[0023] Determining a target ground grid in the grid to be detected based on a reference ground height corresponding to the reference grid and an average height of laser points in a target point layer that is smallest than an average height of laser points in the grid to be detected;

[0024] The determined target ground grid is used as the reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the target ground grid is used as the reference ground height corresponding to the target ground grid, and the step of using the grid that meets the preset proximity condition of the reference grid as the grid to be detected is performed.

[0025] The solution shown in the embodiment of the present application, after determining the initial ground grid, can use the initial ground grid as a reference grid, and the grids of the initial ground grid that meet the preset proximity conditions as the grids to be detected, to detect whether the grid to be detected is the target ground grid. Then, using the detected target ground grid as a reference grid, the grids of the target ground grid that meet the preset proximity conditions as the grids to be detected, continue to determine whether these grids to be detected are the target ground grids, and so on. In this way, when determining whether each grid is a ground grid, its adjacent, already determined ground grids are used as a reference, which has better accuracy than judging only the basic grid itself.

[0026] In a possible implementation, the grid map is a polar coordinate grid map;

[0027] The determining of the initial ground grid in each grid based on the distance between the grid and the automatic driving platform and the average height of the laser points in the target point layer having the smallest average height of the laser points in the grid comprises:

[0028] In each radial grid of the polar coordinate grid diagram, starting from the grid with the smallest distance from the automatic driving platform, an initial ground grid is determined along the radial direction based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid.

[0029] The solution shown in the embodiment of the present application, for the case of a polar coordinate grid diagram, in the radial direction of each angle in the polar coordinate grid diagram, starting from the grid including the laser point closest to the automatic driving platform, it is determined outward in sequence whether the grid is the initial ground grid until the initial ground grid in the radial direction is determined.

[0030] In a possible implementation, generating a grid map corresponding to the laser point cloud includes:

[0031] The grid map is a polar coordinate grid map;

[0032] The grid of the reference grid that meets the preset proximity condition is the next grid including the laser point in the radial direction of the reference grid.

[0033] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0034] The determining of the initial ground grid in each grid based on the distance between the grid and the automatic driving platform and the average height of the laser points in the target point layer having the smallest average height of the laser points in the grid comprises:

[0035] In the grids within a first preset range around the autonomous driving platform in the rectangular coordinate grid diagram, an initial ground grid is determined based on the heights of the laser points in the target point layer having the smallest average height of the laser points in the grids.

[0036] In the embodiment of the present application, for rectangular coordinates, the initial ground grid can be determined within a first preset range around the autonomous driving platform. The first preset range is a range relatively close to the autonomous driving platform. For example, the first preset range can be set to an 8m*8m range around the autonomous driving platform.

[0037] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0038] The grids that satisfy the preset proximity condition of the reference grid are grids within a second preset range around the reference grid.

[0039] In the solution shown in the embodiment of the present application, the second preset range can be a grid of a 3*3 neighborhood centered on the reference grid, for example Figure 12As shown, R22 is a reference grid, and the eight grids R11, R12, R13, R21, R23, R31, R32 and R33 around it are the second preset range.

[0040] In a possible implementation, for each grid in the grid map, layering the laser points in the grid includes:

[0041] Acquire laser points one by one in the grid in order of height from small to large;

[0042] If the absolute value of the difference between the height of the currently acquired laser point and the height of the previously acquired laser point is less than the preset drivable height, the currently acquired laser point is assigned to the point layer where the previously acquired laser point is located; otherwise, the currently acquired laser point is assigned to a newly created point layer.

[0043] The solution shown in the embodiment of the present application is to sort the laser points in the grid by height. Start from the laser point with the minimum height and acquire them one by one. When the first laser point is acquired, a point layer is created and the first laser point is assigned to the created point layer. Continue to acquire laser points and set the height z of the currently acquired i-th laser point to i and the height z of the previously acquired i-1th laser point i-1 For comparison, if z i and z i-1 If the difference between the two is less than the preset drivable height h, the i-th laser point is assigned to the point layer containing the i-1-th laser point. Otherwise, a new point layer is created and the i-th laser point is assigned to the new point layer. This method can ensure that the minimum height difference between laser points in different point layers is greater than the preset drivable height.

[0044] In a second aspect, a device for detecting a drivable area is provided, the device comprising:

[0045] An acquisition module, used to acquire the laser point cloud of the detectable area of ​​the autonomous driving platform;

[0046] A generating module, configured to generate a grid map corresponding to the laser point cloud, wherein each laser point in the laser point cloud corresponds to a grid in the grid map;

[0047] The determination module is configured to determine an initial ground grid in each grid of the grid map, determine a target ground grid in the grid map based on the initial ground grid, and use an area corresponding to the determined initial ground grid and the target ground grid as a drivable area.

[0048] In a possible implementation, the apparatus further includes:

[0049] For each grid in the grid map, the laser points in the grid are layered to obtain at least one point layer.

[0050] In a possible implementation, the determining module is configured to:

[0051] An initial ground grid is determined in each grid of the grid map based on the distance between the grid map and the autonomous driving platform and the height of the laser point in the target point layer having the smallest average height of the laser points in the grid map, the target point layer belonging to the at least one point layer.

[0052] In a possible implementation, the determining module is configured to:

[0053] The initial ground grid is used as a reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the initial ground grid is used as the reference ground height corresponding to the initial ground grid;

[0054] The grid of the reference grid that meets the preset proximity condition is used as the grid to be detected;

[0055] Determining a target ground grid in the grid to be detected based on a reference ground height corresponding to the reference grid and an average height of laser points in a target point layer that is smallest than an average height of laser points in the grid to be detected;

[0056] The determined target ground grid is used as the reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the target ground grid is used as the reference ground height corresponding to the target ground grid, and the step of using the grid that meets the preset proximity condition of the reference grid as the grid to be detected is performed.

[0057] In a possible implementation, the grid map is a polar coordinate grid map;

[0058] The determining of the initial ground grid in each grid based on the distance between the grid and the automatic driving platform and the average height of the laser points in the target point layer having the smallest average height of the laser points in the grid comprises:

[0059] In each radial grid of the polar coordinate grid diagram, starting from the grid with the smallest distance from the automatic driving platform, an initial ground grid is determined along the radial direction based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid.

[0060] In a possible implementation, the generating module is configured to:

[0061] The grid map is a polar coordinate grid map;

[0062] The grid of the reference grid that meets the preset proximity condition is the next grid including the laser point in the radial direction of the reference grid.

[0063] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0064] The determining module is configured to:

[0065] In the grids within a first preset range around the autonomous driving platform in the rectangular coordinate grid diagram, an initial ground grid is determined based on the heights of the laser points in the target point layer having the smallest average height of the laser points in the grids.

[0066] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0067] The grids that satisfy the preset proximity condition of the reference grid are grids within a second preset range around the reference grid.

[0068] In a possible implementation, the layering module is configured to:

[0069] Acquire laser points one by one in the grid in order of height from small to large;

[0070] If the absolute value of the difference between the height of the currently acquired laser point and the height of the previously acquired laser point is less than the preset drivable height, the currently acquired laser point is assigned to the point layer where the previously acquired laser point is located; otherwise, the currently acquired laser point is assigned to a newly created point layer.

[0071] A third aspect provides a decision controller, comprising a processor and a memory;

[0072] The memory stores one or more programs, and the one or more programs are configured to be executed by the processor to implement the method for detecting the drivable area described in the first aspect above.

[0073] In a fourth aspect, a computer-readable storage medium is provided, comprising computer-readable instructions. When the computer-readable storage medium is run on a decision controller, the decision controller executes the method for detecting a drivable area as described in the first aspect.

[0074] In a fifth aspect, a computer program product comprising instructions is provided. When the computer program product is run on a decision controller, the decision controller executes the method for detecting the drivable area described in the first aspect.

[0075] The technical solution provided by this application includes at least the following beneficial effects:

[0076] In this application, we first need to obtain a laser point cloud, then divide the obtained laser points into a grid, determine the initial ground grid in the grid, and then use each initial ground grid as a reference to continue to determine the target ground grid. Finally, the area corresponding to the determined initial ground grid and the target ground grid is used as the drivable area. In this way, since the determination of the drivable area is based on the laser points, the laser has better penetration and is minimally affected by the environment. Even in harsh environments such as heavy fog and heavy rain, the drivable area can be detected more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is a schematic diagram of the structure of a decision controller provided in an embodiment of the present application;

[0078] Figure 2 This is a flow chart of a method for detecting a drivable area provided in an embodiment of the present application;

[0079] Figure 3 is a schematic diagram of a laser radar coordinate system provided in an embodiment of the present application;

[0080] Figure 4 This is a schematic diagram of a polar coordinate grid diagram provided in an embodiment of the present application;

[0081] Figure 5 is a schematic diagram of a rectangular coordinate grid diagram provided in an embodiment of the present application;

[0082] Figure 6 This is a schematic diagram of a point layer provided in an embodiment of the present application;

[0083] Figure 7 This is a flow chart of a method for detecting a drivable area provided in an embodiment of the present application;

[0084] Figure 8 This is a flow chart of a method for detecting a drivable area provided in an embodiment of the present application;

[0085] Figure 9 is a schematic diagram of a rectangular coordinate grid diagram provided in an embodiment of the present application;

[0086] Figure 10 This is a flow chart of a method for detecting a drivable area provided in an embodiment of the present application;

[0087] Figure 11 This is a flow chart of a method for detecting a drivable area provided in an embodiment of the present application;

[0088] Figure 12 is a schematic diagram of a rectangular coordinate grid diagram provided in an embodiment of the present application;

[0089] Figure 13It is a structural diagram of a device for detecting a drivable area provided in an embodiment of the present application. DETAILED DESCRIPTION

[0090] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0091] The present application provides a method for detecting a drivable area. This method can be applied to autonomous driving platforms such as self-driving cars and intelligent robots. This method can be implemented by the decision controller of various autonomous driving platforms. Each autonomous driving platform can be equipped with a perception system, such as a lidar. The decision controller uses data collected by the perception system to determine the drivable area and thereby control the autonomous driving platform's movement.

[0092] like Figure 1 FIG. 1 is a schematic diagram of a decision controller 100 provided in an embodiment of the present application. Figure 1 In the embodiment, the decision controller may include a processor 101 and a memory 102. The processor 101 may be a central processing unit (CPU). The processor 101 may refer to one processor or may include multiple processors. The memory 102 may include a volatile memory, such as a random access memory (RAM); the memory may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, etc.; the memory may also include a combination of the above types of memory. The memory 102 may refer to one memory or may include multiple memories. In one embodiment, the memory 102 stores computer-readable instructions, which may be executed by the processor 101 to implement the method for detecting a drivable area provided in an embodiment of the present application.

[0093] The present application embodiment provides a method for detecting a drivable area, such as Figure 2 As shown, the processing flow of the method may include the following steps:

[0094] Step 201: Obtain a laser point cloud of a detectable area of ​​an autonomous driving platform.

[0095] Among them, the automatic driving platform can be a self-driving car, an intelligent robot, etc.

[0096] In practice, the autonomous driving platform can be equipped with a laser radar. To effectively detect obstacles in the direction of travel, the laser radar is usually installed on the top or front of the autonomous driving platform. The laser radar can be a multi-line laser radar, a micro-electro-mechanical system (MEMS) laser radar, or a flash laser radar.

[0097] During operation, the autonomous platform uses a laser radar to scan the surrounding environment by emitting laser beams into its detectable area. The laser beams reflect off the ground, trees, obstacles, and buildings, and the radar receives the reflected echoes, generating a laser point cloud of the detectable area. The following describes how to obtain the 3D coordinates of each laser point in the laser point cloud.

[0098] After processing the reflected echo, the emission angle and scanning distance d of the laser line corresponding to the echo can be obtained, where the emission angle includes the horizontal angle α and the vertical angle β. Figure 3 The figure shows the schematic diagram of the laser point in the laser radar coordinate system. From this, the three-dimensional coordinates of the laser point on the object in the laser radar coordinate system can be calculated as: (x, y, z), where:

[0099] x=cosα·sinβ·d

[0100] y=cosα·cosβ·d

[0101] z=sinβ·d

[0102] Then, the three-dimensional coordinates of the laser point in the laser radar coordinate system are converted to the automatic driving platform coordinate system. In the case where the automatic driving platform is an autonomous driving vehicle, the autonomous driving platform coordinate system can take the midpoint of the rear axle of the autonomous driving vehicle as the origin, the driving direction as the positive direction of the horizontal (X) axis, the left side as the positive direction of the vertical (Y) axis, and the vertical upward as the positive direction of the vertical (Z) axis. Of course, the autonomous driving platform coordinate system can also be calibrated in other ways, which is not limited in this embodiment of the present application. The method for converting the three-dimensional coordinates of the laser point in the laser radar coordinate system to the automatic driving platform coordinate system can be as follows:

[0103] P′=R·P+T, where P′ is the three-dimensional coordinate of the laser point in the autonomous driving platform coordinate system, P is the three-dimensional coordinate of the laser point in the autonomous driving platform coordinate system, R is the rotation matrix, and T is the translation matrix. R and T can be obtained by pre-calibrating the lidar coordinate system and the autonomous driving platform coordinate system. Here, calibration can be achieved using a plane segmentation matching algorithm estimated by a random sample consensus (RANSAC) model. Of course, other methods can also be used for calibration, and the embodiments of this application do not limit the calibration method.

[0104] Here, it should also be noted that P′ is the three-dimensional coordinate of the laser point in the autonomous driving platform coordinate system, and the vertical coordinate of P′ represents the height of the laser point in the autonomous driving platform coordinate system.

[0105] In one possible implementation, since the laser radar is usually installed on top of an autonomous driving platform, some laser lines may scan the autonomous driving platform. The laser points obtained by the laser lines scanning the autonomous driving platform can be removed. The removal method can be as follows:

[0106] A bounding box is created under the autonomous platform. This bounding box can be a minimum rectangular parallelepiped that completely encloses the platform. Laser points within the laser point cloud that fall within this bounding box are removed. Specifically, laser points whose horizontal, vertical, and vertical coordinates all fall within the bounding box's coordinate range are removed. Furthermore, since objects above the autonomous platform will not affect its normal operation, laser points obtained by scanning these objects can also be removed. Specifically, laser points whose horizontal and vertical coordinates all fall within the bounding box's coordinate range are removed.

[0107] Step 202: Generate a raster map corresponding to the laser point cloud.

[0108] Each laser point in the laser point cloud corresponds to a grid in the grid map. A grid may have no laser points, one or more laser points.

[0109] In implementation, the laser point cloud is first subjected to dimensionality reduction. Specifically, the laser point cloud is projected onto a two-dimensional plane at Z = 0 in the autonomous driving platform coordinate system. The height of each laser point is stored as feature data for that laser point. A grid map is then generated on the laser point cloud within the two-dimensional plane, and the laser points are divided into grids within the grid map. The grid map can be a polar coordinate grid map, a rectangular coordinate grid map, or other grid maps. Polar coordinate grids and rectangular coordinate grids are described separately below.

[0110] 1. Polar Grid Chart

[0111] A polar coordinate grid image can be generated for the area within a preset range around the autonomous driving platform. A polar coordinate system is established with the origin of the autonomous driving platform coordinate system as the origin. Under the polar coordinate system, the polar coordinate grid image is divided according to the preset angular resolution and the preset radial resolution. For example, Figure 4 A polar coordinate grid diagram is shown in Figure 4 The polar coordinate grid is generated with an angle of 0.5 degrees, a radial resolution of 0.2m within a radial distance of 30 meters (m), a radial resolution of 0.5m from 30m to 60m, and a radial resolution of 1m above 60m. Figure 4 In the polar coordinate grid diagram, grids A, B, and C are located in the same radial direction, wherein grid B is the next grid of grid A in the radial direction, and grid C is the next grid of grid B in the radial direction.

[0112] 2. Rectangular Coordinate Grid

[0113] A rectangular coordinate grid map can be generated for the area within a preset range around the autonomous driving platform, where each grid in the rectangular coordinate grid map is a rectangular grid of the same size. For example, Figure 5 A rectangular coordinate grid diagram is shown. Figure 5 A rectangular coordinate grid map is generated for the area 80m in front and behind and 60m in left and right of the automatic driving platform. Each grid in the grid map is a square with a side length of 0.2m.

[0114] Step 203: Determine an initial ground grid in each grid of the grid map.

[0115] The distance between the grid and the autonomous driving platform can be the distance from the center to the autonomous driving platform.

[0116] In practice, for each grid containing laser points, the laser points can be stratified based on the preset drivable height and the heights of the laser points within the grid. Each layer of laser points is referred to as a point layer. The stratification principle is to ensure that the minimum height difference between laser points in different point layers is greater than the preset drivable height. Based on this stratification principle, various stratification methods can be used, several of which are listed below.

[0117] Method 1:

[0118] Acquire laser points one by one. When the first laser point is acquired, create a point layer and assign the first laser point to the created point layer. Continue to acquire laser points. When the i-th laser point is acquired, determine whether there is a target point layer in the existing point layer that meets the following conditions:

[0119] Hh<z i <Lh, where H is the maximum height of the laser point in any existing point layer, L is the minimum height of the laser point in the existing point layer, zi is the height of the i-th laser point, and h is the preset drivable height, which can be set to a value slightly larger than the height of the automatic driving platform, for example, it can be set to the height of the automatic driving platform plus 0.1m.

[0120] If a target that meets the above conditions is determined to have a point layer, the i-th laser point is assigned to the target's existing point layer. If multiple targets that meet the above conditions are determined to have point layers, these multiple target point layers are merged and the i-th laser point is assigned to the merged point layer.

[0121] Method 2:

[0122] Sort the laser points in the grid by height. Start acquiring the laser points one by one from the laser point with the minimum height. When acquiring the first laser point, create a point layer and assign the first laser point to the created point layer. Continue acquiring laser points and assign the height z of the currently acquired i-th laser point to the point layer. i and the height z of the previously acquired i-1th laser point i-1 For comparison, if z i and z i-1 If the difference is less than the preset drivable height h, the i-th laser point is assigned to the point layer where the i-1-th laser point is located; otherwise, a new point layer is created and the i-th laser point is assigned to the new point layer.

[0123] After stratifying the laser points in the grid, the initial ground grid can be determined for each grid in the grid map based on the distance between the grid map and the autonomous driving platform and the height of the laser points in the target point layer with the minimum average height of the laser points in the grid map. The following describes the methods for determining the initial ground grid for polar and rectangular grid maps, respectively.

[0124] 1. Polar coordinate grid map, such as Figure 7 As shown, the following processing can be performed in this case.

[0125] In the radial direction of each angle in the grid diagram, starting from the grid containing the laser point closest to the autonomous driving platform, the grids are sequentially determined outwards to determine whether they are the initial ground grids, until the initial ground grid in the radial direction is determined. The determination method can be as follows.

[0126] 701. Obtain a grid with the smallest distance from the automatic driving platform in each radial direction in the grid map as a grid to be detected.

[0127] 702. Obtain a target point layer having the minimum average height of laser points in the grid to be detected.

[0128] 703. Determine whether the heights of all laser points in the target point layer are within a preset height range, and whether the maximum height difference of the laser points in the target point layer is less than a first height difference threshold.

[0129] Among them, the preset height range can be [h0-Δh, h0+Δh], h0 is the height of the origin of the automatic driving platform coordinate system from the ground, which can be called the calibrated ground height and can be obtained through factory calibration. The basis for setting the preset height range can be: the undulating ground within a certain height range does not affect the driving of the automatic driving platform. For example, the height fluctuation range Δh can be set to 0.2m. The basis for setting the first height difference threshold can be: because obstacles are usually relatively high, they will affect the passage of vehicles, for example, 1m, then the maximum height difference of the laser point scanned on the obstacle may reach about 1m, and for the ground that meets the road slope specifications, the maximum height difference of the laser point scanned on the ground can be about 0.1m. In addition, since there is a certain error in the laser radar ranging, the first height difference threshold can be set relatively large, for example, it can be set between 0.1m and 0.2m.

[0130] 704. If the heights of all laser points in the target point layer are within a preset height range and the maximum height difference of the laser points in the target point layer is less than a first height difference threshold, the grid to be detected is determined as the initial ground grid, and the average height of the laser points in the target point layer is determined as the reference ground height corresponding to the initial ground grid.

[0131] 705. If there is a laser point in the target point layer whose height is not within the preset height range, or the maximum height difference of the laser points in the target point layer is not less than the first height difference threshold, determine whether the absolute value of the first difference between the minimum height of the laser points in the target point layer and the calibrated ground height is greater than the preset drivable height.

[0132] 706. If the absolute value of the first difference is greater than the preset drivable height, the grid to be detected is determined to be a high-altitude grid, and the next grid including the laser point in the radial direction of the high-altitude grid is used as the grid to be detected, and the process goes to step 702.

[0133] Due to the limitation of the laser radar scanning angle, the following situation may occur: the ground around the driving platform cannot be scanned by the laser radar, but there are suspended objects above the ground that are not scanned by the laser radar, such as banners, which can be scanned by the laser radar.

[0134] Then, the grid corresponding to this area will only contain the laser point of the suspended object. Because the suspended object is hanging very high, the corresponding laser point height will be much higher than the calibrated ground height, so this grid will not meet the ground grid condition in step 701. In step 706, by determining that the first difference between the minimum laser point height and the calibrated ground height is greater than the preset drivable height, it can be determined that although the area corresponding to this grid contains a suspended object, the autonomous driving platform is passable. Therefore, this grid can be determined as a high-altitude grid and not judged as an obstacle.

[0135] 707 . If the absolute value of the first difference is not less than the preset drivable height, determine the to-be-detected grid as an obstacle grid.

[0136] Corresponding to the situation in step 706 above, if the suspended object is a railing, then in step 707, by determining whether the first difference between the minimum height of the laser spot and the calibrated ground height is less than the preset drivable height, it can be identified that there is a suspended object in the area corresponding to the grid, and the height between the suspended object and the ground is not high enough for the autonomous driving platform to pass. In this case, the grid can be determined to be an obstacle grid.

[0137] Second, the rectangular coordinate grid map, such as Figure 8 As shown, the following processing can be performed in this case.

[0138] 801. Determine a grid including a laser point within a first preset range around the autonomous driving platform as a grid to be detected.

[0139] The first preset range is the range closer to the autonomous driving platform. Figure 9 As shown, the first preset range can be set to a range of 8m*8m around the automatic driving platform.

[0140] 802. Obtain a target point layer having the minimum average height of laser points in the grid to be detected.

[0141] 803 : Determine whether the heights of all laser points in the target point layer are within a preset height range, and whether the maximum height difference of the laser points in the target point layer is less than a first height difference threshold.

[0142] 804. If the heights of the laser points in the target point layer are all within the preset height range, and the maximum height difference of the laser points in the target point layer is less than the first height threshold, the average height of the laser points in the target point layer is determined as the reference ground height corresponding to the grid to be detected, the grid to be detected is determined as the initial ground grid, and marked as detected.

[0143] 805. If there is a laser point in the target point layer whose height is not within the preset height range, or the maximum height difference of the laser points in the target point layer is not less than the first height difference threshold, determine whether the absolute value of the first difference between the minimum height of the laser points in the target point layer and the calibrated ground height is less than the preset drivable height.

[0144] 806. If the absolute value of the first difference is less than the preset drivable height, the grid to be detected is determined as an obstacle grid and marked as detected.

[0145] If the absolute value of the first difference is greater than the preset drivable height, the grid to be detected is not marked as detected, so that the grid to be detected can be re-checked later.

[0146] The above method can be used to determine the ground grid around the automatic driving platform in the rectangular coordinate grid diagram and the polar coordinate grid diagram.

[0147] Step 204: Based on the initial ground grid, determine the target ground grid in the grid map.

[0148] After determining the initial ground grid, you can use the initial ground grid as a reference to determine whether the grid adjacent to the initial ground grid is the target ground grid, the upper air grid, or the obstacle grid. The following describes the methods for determining the target ground grid, the upper air grid, or the obstacle grid for polar and rectangular grid plots, respectively.

[0149] 1. Polar coordinate grid map, such as Figure 10 As shown, the following processing can be performed in this case.

[0150] 1001. Determine the determined initial ground grid as a reference grid;

[0151] 1002. Acquire the next grid including the laser point in the radial direction of the reference grid as the grid to be detected.

[0152] 1003. Obtain a target point layer having the minimum average height of the laser points in the grid to be detected.

[0153] 1004. Calculate the slope of the target point layer relative to the reference grid based on the average height of the laser points in the target point layer and the reference ground height corresponding to the reference grid.

[0154] The slope calculation method can be as follows:

[0155] s=(z current -z last ) / (r current -r last )

[0156] Among them, s is the slope of the target point layer relative to the reference grid, z current is the average height of the laser points in the target point layer, z last is the reference ground height corresponding to the reference grid, r current is the distance from the center of the grid to be detected to the origin, r last is the distance from the center of the reference grid to the origin.

[0157] 1005. Determine whether the absolute value of the slope is less than a preset slope threshold, and whether the maximum height difference of the laser points in the target point layer is less than a first height difference threshold.

[0158] The preset slope threshold is set in consideration of the road slope specification and the ranging error of the lidar, and can be set between 0.1 and 0.15, such as 0.15.

[0159] 1006. If the absolute value of the slope is less than the preset slope threshold, and the maximum height difference of the laser points in the target point layer is less than the first height difference threshold, the grid to be detected is determined as the target ground grid, the average height of the laser points in the target point layer is determined as the reference ground height corresponding to the target ground grid, and the slope is determined as the reference ground slope corresponding to the target ground grid. The target ground grid is determined as the reference grid, and the process proceeds to step 1002.

[0160] 1007 : If the slope is not less than the preset slope threshold, or the maximum height difference of the laser points in the target point layer is not less than the first height difference threshold, determine whether the slope is greater than zero.

[0161] 1008. If the slope is greater than zero, determine the estimated ground height of the grid to be detected based on the reference ground height and the reference ground slope corresponding to the reference grid.

[0162] The estimated ground height can be calculated as follows:

[0163] z current ′=z last +s last ×(r current -r last )

[0164] Among them, z current ′ is the estimated ground height of the grid to be detected, z last is the reference ground height corresponding to the reference grid, r current is the distance from the center of the grid to be detected to the origin, r last is the distance from the center of the reference grid to the origin, s last It should be noted that if the reference grid is the ground grid closest to the autonomous driving platform, its corresponding reference ground slope can be set to 0.

[0165] 1009. Determine whether the absolute value of a second difference between the estimated ground height and the minimum height of the laser point in the target point layer is greater than a preset drivable height.

[0166] 1010. If the absolute value of the second difference between the estimated ground height and the minimum height of the laser point in the target point layer is greater than the preset drivable height, the grid to be detected is determined to be a high-altitude grid, the estimated ground height is determined to be the reference ground height corresponding to the high-altitude grid, and the reference ground slope corresponding to the reference grid is determined to be the reference ground slope corresponding to the high-altitude grid. The high-altitude grid is determined to be the reference ground grid, and the process proceeds to step 1002.

[0167] Due to the limitation of the laser radar scanning angle, the following situation may occur: the laser radar may not be able to scan the ground far away from the automatic driving platform, but the suspended objects on the ground can be scanned. In this case, the calculated slope will be greater than 0. If the suspended object is very high and does not affect the passage, such as a banner, then the absolute value of the second difference between the calculated estimated ground height and the minimum height of the laser point in the target point layer will be greater than the preset drivable height, then the grid can be determined as a high-altitude grid. In this case, the area corresponding to the grid actually has ground, but it has not been scanned. In this case, the estimated ground height can be used as the reference ground height corresponding to the high-altitude grid, and the high-altitude grid can be used as a reference to determine the type of grid in the next radial direction.

[0168] 1011. If the absolute value of the second difference is less than the preset drivable height, the grid to be detected is determined to be an obstacle grid. If it is determined to be an obstacle grid, the grids following the obstacle grid in the radial direction are no longer detected.

[0169] 1012. If the slope is less than zero, determine the grid to be detected as an obstacle grid.

[0170] For the case where the slope is less than 0, it is possible that the area corresponding to the grid to be detected is a downhill or sunken ground, and the slope is too large or the depth of the sunken ground is already impassable. In this case, the grid to be detected can be determined as an obstacle grid.

[0171] Second, the rectangular coordinate grid map, such as Figure 11 As shown, the following processing can be performed in this case.

[0172] 1101. Use the determined initial ground grid as a reference grid.

[0173] 1102. Determine grids within a second preset range around the reference grid that are not marked as detected as grids to be detected.

[0174] The second preset range may be a 3*3 neighborhood grid centered on the reference grid. Figure 12 As shown, R22 is a reference grid, and the eight grids R11, R12, R13, R21, R23, R31, R32 and R33 around it are the second preset range.

[0175] 1103. Determine whether the grid to be detected includes a laser point.

[0176] 1104. If the grid to be detected does not include a laser point, the reference ground height corresponding to the reference grid is determined as the reference ground height corresponding to the grid to be detected. The grid to be detected is determined as a ground grid and marked as detected. The target ground grid is used as the reference grid, and the process proceeds to step 1102.

[0177] 1105. If the grid to be detected includes laser points, obtain a target point layer having the minimum average height of the laser points in the grid to be detected.

[0178] 1106. Determine whether the absolute value of a third difference between an average height of the laser points in the target point layer and a reference ground height corresponding to the reference grid is less than a second height threshold, and whether the maximum height difference of the laser points in the target point layer is less than a first height difference threshold.

[0179] 1107. If the absolute value of the third difference between the average height of the laser points in the target point layer and the reference ground height corresponding to the reference grid is less than the second height threshold, and the maximum height difference of the laser points in the target point layer is less than the first threshold, the average height of the laser points in the target point layer is determined as the reference ground height corresponding to the grid to be detected, the grid to be detected is determined as the target ground grid, and marked as detected. The target ground grid is used as the reference grid, and the process proceeds to step 1102.

[0180] 1108. If the absolute value of the third difference between the average height of the laser points in the target point layer and the reference ground height corresponding to the reference grid is not less than the second height threshold, or the maximum height difference of the laser points in the target point layer is not less than the first threshold, determine whether the average height of the laser points in the target point layer is greater than the reference ground height corresponding to the reference grid.

[0181] 1109. If the average height of the laser points in the target point layer is greater than the reference ground height corresponding to the reference grid, determine whether the absolute value of the fourth difference between the minimum height of the laser points in the target point layer and the reference ground height corresponding to the reference grid is greater than the preset drivable height.

[0182] 1110. If the absolute value of the fourth difference is greater than the preset drivable height, the average height of the laser points in the target point layer is determined as the reference ground height corresponding to the grid to be detected. The grid to be detected is determined as a high-altitude grid and marked as detected. The determined high-altitude grid is used as the reference grid, and the process proceeds to step 1102.

[0183] 1111. If the absolute value of the fourth difference is less than the preset drivable height, determine the to-be-detected grid as an obstacle grid and mark it as detected.

[0184] 1112. If the average height of the laser points in the target point layer is less than the reference ground height corresponding to the reference grid, the grid to be detected is determined as an obstacle grid and marked as detected.

[0185] The above method can detect the target ground grid in the grid map, and can also detect the high-altitude grid and obstacle grid.

[0186] Step 205: The area corresponding to the determined initial ground grid and the target ground grid is used as a drivable area.

[0187] In implementation, for the case where the ground grid (the ground grid includes the initial ground grid and the target ground grid), the high-altitude grid and the obstacle grid are determined in the rectangular coordinate grid map, the area corresponding to the above-mentioned ground grid and the high-altitude grid can be directly used as the drivable area. For the case where the ground grid, the high-altitude grid and the obstacle are determined in the polar coordinate grid map, they can be first converted to the rectangular coordinate grid map. The conversion method can be: directly generate a direct coordinate grid map on the polar coordinate grid map generated above, connect the center of the obstacle grid in the polar coordinate grid map with the origin of the polar coordinate grid map, and the grids in the rectangular grid map that the connecting line passes through can be marked as drivable grids, and the areas corresponding to these drivable grids are the drivable areas.

[0188] Based on the same technical concept, the embodiment of the present application also provides a device for detecting a drivable area, such as Figure 13 As shown, the device includes:

[0189] An acquisition module 1310 is used to acquire a laser point cloud of a detectable area of ​​the autonomous driving platform, and specifically implements the acquisition function in step 201 above, as well as other implicit steps;

[0190] A generation module 1320 is configured to generate a grid map corresponding to the laser point cloud, wherein each laser point in the laser point cloud corresponds to a grid in the grid map, and specifically implement the generation function in the above step 202, as well as other implicit steps;

[0191] Determination module 1330 is configured to determine an initial ground grid in each grid of the grid map, determine a target ground grid in the grid map based on the initial ground grid, and use the area corresponding to the determined initial ground grid and the target ground grid as a drivable area. Specifically, this module can implement the determination functions in steps 202-205 above, as well as other implicit steps.

[0192] In a possible implementation, the apparatus further includes:

[0193] For each grid in the grid map, the laser points in the grid are layered to obtain at least one point layer.

[0194] In a possible implementation, the determining module 1330 is configured to:

[0195] An initial ground grid is determined in each grid of the grid map based on the distance between the grid map and the autonomous driving platform and the height of the laser point in the target point layer having the smallest average height of the laser points in the grid map, the target point layer belonging to the at least one point layer.

[0196] In a possible implementation, the determining module 1330 is configured to:

[0197] The initial ground grid is used as a reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the initial ground grid is used as the reference ground height corresponding to the initial ground grid;

[0198] The grid of the reference grid that meets the preset proximity condition is used as the grid to be detected;

[0199] Determining a target ground grid in the grid to be detected based on a reference ground height corresponding to the reference grid and an average height of laser points in a target point layer that is smallest than an average height of laser points in the grid to be detected;

[0200] The determined target ground grid is used as the reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the target ground grid is used as the reference ground height corresponding to the target ground grid, and the step of using the grid that meets the preset proximity condition of the reference grid as the grid to be detected is executed.

[0201] In a possible implementation, the grid map is a polar coordinate grid map;

[0202] The determining of the initial ground grid in each grid based on the distance between the grid and the automatic driving platform and the average height of the laser points in the target point layer having the smallest average height of the laser points in the grid comprises:

[0203] In each radial grid of the polar coordinate grid diagram, starting from the grid with the smallest distance from the automatic driving platform, an initial ground grid is determined along the radial direction based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid.

[0204] In a possible implementation, the generating module 1320 is configured to:

[0205] The grid map is a polar coordinate grid map;

[0206] The grid of the reference grid that meets the preset proximity condition is the next grid including the laser point in the radial direction of the reference grid.

[0207] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0208] The determining module 1330 is configured to:

[0209] In the grids within a first preset range around the autonomous driving platform in the rectangular coordinate grid diagram, an initial ground grid is determined based on the heights of the laser points in the target point layer having the smallest average height of the laser points in the grids.

[0210] In a possible implementation, the grid image is a rectangular coordinate grid image;

[0211] The grids that satisfy the preset proximity condition of the reference grid are grids within a second preset range around the reference grid.

[0212] In a possible implementation, the layering module is configured to:

[0213] Acquire laser points one by one in the grid in order of height from small to large;

[0214] If the absolute value of the difference between the height of the currently acquired laser point and the height of the previously acquired laser point is less than the preset drivable height, the currently acquired laser point is assigned to the point layer where the previously acquired laser point is located; otherwise, the currently acquired laser point is assigned to a newly created point layer.

[0215] It should be noted that the acquisition module 1310 , the generation module 1320 and the determination module 1330 may be implemented by a processor, or by a processor in conjunction with a memory, or by the processor executing program instructions in the memory.

[0216] It should also be noted that the apparatus for detecting a drivable area provided in the above embodiment is merely exemplified by the division of the aforementioned functional modules when detecting a drivable area. In actual applications, the aforementioned functions can be distributed among different functional modules as needed, i.e., the internal structure of the decision controller can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the apparatus for detecting a drivable area provided in the above embodiment and the method embodiment for detecting a drivable area are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0217] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the device, the process or function described in the embodiment of the present application is generated in whole or in part. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by the device or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (such as a floppy disk, hard disk and tape, etc.), an optical medium (such as a digital video disk (DVD), etc.), or a semiconductor medium (such as a solid-state hard disk, etc.).

[0218] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0219] The above description is only one embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for detecting a drivable area, characterized in that: The method comprises: Obtaining laser point clouds of the detectable area of ​​the autonomous driving platform; Generating a grid map corresponding to the laser point cloud, wherein each laser point in the laser point cloud corresponds to a grid in the grid map; For each grid in the grid map, layering the laser points in the grid to obtain at least one point layer; When the grid map is a polar coordinate grid map, the initial ground grid is determined in each radial grid of the polar coordinate grid map, starting from the grid with the smallest distance from the automatic driving platform and along the radial direction based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid; when the grid map is a rectangular coordinate grid map, the initial ground grid is determined in grids within a first preset range around the automatic driving platform in the rectangular coordinate grid map based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid, the target point layer belonging to the at least one point layer; The initial ground grid is used as a reference grid, and the average height of the laser points in the target point layer with the smallest average height of the laser points in the initial ground grid is used as the reference ground height corresponding to the initial ground grid; The grid of the reference grid that meets the preset proximity condition is used as the grid to be detected; Determining a target ground grid in the grid to be detected based on a reference ground height corresponding to the reference grid and an average height of laser points in a target point layer that is smallest than an average height of laser points in the grid to be detected; The determined target ground grid is used as a reference grid, the average height of the laser points in the target point layer with the smallest average height of the laser points in the target ground grid is used as the reference ground height corresponding to the target ground grid, and the grid that satisfies the preset proximity condition of the reference grid is used as the grid to be detected; The area corresponding to the determined initial ground grid and the target ground grid is used as a drivable area.

2. The method according to claim 1, characterized in that In the case where the grid diagram is a polar coordinate grid diagram, the grid of the reference grid that meets the preset proximity condition is the next grid including the laser point in the radial direction of the reference grid.

3. The method according to claim 1, characterized in that In the case that the grid image is a rectangular coordinate grid image, the grids of the reference grid that meet the preset proximity condition are grids within a second preset range around the reference grid.

4. The method according to any one of claims 1 to 3, characterized in that For each grid in the grid map, layering the laser points in the grid includes: Acquire laser points one by one in the grid in order of height from small to large; If the absolute value of the difference between the height of the currently acquired laser point and the height of the previously acquired laser point is less than the preset drivable height, the currently acquired laser point is assigned to the point layer where the previously acquired laser point is located; otherwise, the currently acquired laser point is assigned to a newly created point layer.

5. A device for detecting a drivable area, characterized in that: The device comprises: An acquisition module, used to acquire the laser point cloud of the detectable area of ​​the autonomous driving platform; A generating module, configured to generate a grid map corresponding to the laser point cloud, wherein each laser point in the laser point cloud corresponds to a grid in the grid map; A stratification module, configured to stratify the laser points in each grid in the grid map to obtain at least one point layer; The determination module is used to, when the grid map is a polar coordinate grid map, determine the initial ground grid in each radial grid of the polar coordinate grid map, starting from the grid with the smallest distance from the automatic driving platform, and along the radial direction based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid; when the grid map is a rectangular coordinate grid map, determine the initial ground grid based on the height of the laser point in the target point layer with the smallest average height of the laser points in the grid in the grid within a first preset range around the automatic driving platform in the rectangular coordinate grid map, the target point layer belongs to the at least one point layer, and the initial ground grid is used as a reference grid, and the laser point in the target point layer with the smallest average height of the laser points in the initial ground grid is used as a reference grid. The average height of the light spots is used as the reference ground height corresponding to the initial ground grid; the grid of the reference grid that meets the preset proximity condition is used as the grid to be detected; based on the reference ground height corresponding to the reference grid and the average height of the laser points in the target point layer that is smallest in the average height of the laser points in the grid to be detected, the target ground grid is determined in the grid to be detected; the determined target ground grid is used as the reference grid, and the average height of the laser points in the target point layer that is smallest in the average height of the laser points in the target ground grid is used as the reference ground height corresponding to the target ground grid, and the process proceeds to use the grid of the reference grid that meets the preset proximity condition as the grid to be detected, and the area corresponding to the determined initial ground grid and the target ground grid is used as the drivable area.

6. The device according to claim 5, characterized in that In the case where the grid diagram is a polar coordinate grid diagram, the grid of the reference grid that meets the preset proximity condition is the next grid including the laser point in the radial direction of the reference grid.

7. The device according to claim 5, characterized in that In the case that the grid image is a rectangular coordinate grid image, the grids of the reference grid that meet the preset proximity condition are grids within a second preset range around the reference grid.

8. The device according to any one of claims 5 to 7, characterized in that The layered module is used to: Acquire laser points one by one in the grid in order of height from small to large; If the absolute value of the difference between the height of the currently acquired laser point and the height of the previously acquired laser point is less than the preset drivable height, the currently acquired laser point is assigned to the point layer where the previously acquired laser point is located; otherwise, the currently acquired laser point is assigned to a newly created point layer.

9. A decision controller, characterized in that: The decision controller includes a processor and a memory; The memory stores one or more programs, and the one or more programs are configured to be executed by the processor to implement the method for detecting a drivable area according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes computer-readable instructions, which, when the computer-readable storage medium is run on a decision controller, causes the decision controller to execute the method for detecting a drivable area according to any one of claims 1 to 4.

Images