A method, device, and storage medium for detecting a drivable area

Abstract

The application relates to a drivable area detection method, device, equipment and storage medium, the method comprising: acquiring point cloud data collected by a vehicle-mounted laser radar, converting the point cloud data in a radar coordinate system into first point cloud data in a vehicle coordinate system; performing filtering processing on the first point cloud data to obtain second point cloud data; converting the second point cloud data into a depth image, determining the nearest unpassable pixel point of each column in each column of adjacent two pixel points in all column pixels of the depth image based on the depth values of the two pixel points; converting the unpassable pixel points of all columns in the depth image into third point cloud data, and determining the drivable area of the vehicle based on the third point cloud data. In this way, the boundary points of the drivable area are confirmed based on the neighborhood information of all column pixels of the depth image, the detection of the drivable area of the unstructured road scene is effectively solved, and the robustness, accuracy and real-time performance of the passable space detection are greatly guaranteed.

Description

Technical Field

[0001] This invention relates to the field of environmental perception for autonomous vehicles, and specifically to a method, apparatus, device, and storage medium for detecting drivable areas. Background Technology

[0002] Autonomous driving relies on a vehicle's intelligent perception of its surroundings. Based on road detection data, road scenarios can be categorized into structured and unstructured road scenarios. Unstructured road scenarios include uneven terrain such as ramps and speed bumps.

[0003] Currently, deep learning algorithms are used to detect drivable areas on unstructured roads. However, deep learning algorithms have drawbacks such as dependence on training data, long training time, and high hardware requirements. Summary of the Invention

[0004] One objective of this invention is to provide a method for detecting drivable areas to solve the problem of detecting drivable areas in unstructured road scenarios; a second objective is to provide a device for detecting drivable areas; a third objective is to provide a device for detecting drivable areas; and a fourth objective is to provide a computer-readable storage medium.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An operable area detection method, the method comprising:

[0007] Acquire point cloud data collected by the vehicle-mounted LiDAR and convert the point cloud data in the LiDAR coordinate system into the first point cloud data in the vehicle coordinate system;

[0008] The first point cloud data is filtered to obtain the second point cloud data; wherein, the second point cloud data is the point cloud data excluding noise points within a preset region of interest;

[0009] The second point cloud data is converted into a depth image, and based on the depth values ​​of two adjacent pixels in each column of the depth image, the nearest impassable pixel in each column is determined.

[0010] The impassable pixels in all columns of the depth image are converted into third point cloud data, and the drivable area of ​​the vehicle is determined based on the third point cloud data.

[0011] Based on the above technical means, the boundary points of the drivable area are identified by the neighborhood information of all columns of pixels in the depth image, which effectively solves the problem of drivable area detection in unstructured road scenes and greatly ensures the robustness, accuracy and real-time performance of drivable space detection.

[0012] Furthermore, the step of filtering the first point cloud data to obtain the second point cloud data includes:

[0013] Filter all point cloud data within the preset region of interest from the first point cloud data;

[0014] Filter out the fourth point cloud data from all point cloud data within the preset region of interest, which is within a preset range of the vehicle;

[0015] Noise points are removed from the fourth point cloud data to obtain the fifth point cloud data;

[0016] The fifth point cloud data and the point cloud data other than the fourth point cloud data in all point cloud data within the preset region of interest constitute the second point cloud data.

[0017] Based on the above technical means, the present invention performs filtering processing on the first point cloud data, selects the point cloud data within a preset region of interest, and further removes noise points within the preset region of interest to finally obtain the second point cloud data, thereby reducing the computational load of the point cloud data.

[0018] Furthermore, the step of removing noise points from the fourth point cloud data to obtain the fifth point cloud data includes:

[0019] Obtain the Z-coordinate value and reflected signal intensity value of the first target point in the fourth point cloud data;

[0020] When the Z-coordinate value of the first target point is less than the first Z-th threshold and the reflected signal intensity value is less than the signal intensity threshold, the first target point is removed from the fourth point cloud data as a noise point to obtain the fifth point cloud data.

[0021] Based on the above technical means, the noise points in this invention can be understood as reflection points. Because there may be obstacles with reflective properties on the ground, such as glass or water surfaces, when the laser beam hits an obstacle with reflective properties, it will be reflected to other places and then reflected back to the lidar, resulting in a longer reflection time. This lengthens the distance between the reflection point and the lidar, and it is considered to be a point below the ground. Therefore, it is necessary to remove the reflection points in order to reduce the computational load of point cloud data.

[0022] Furthermore, obtaining all point cloud data within the preset region of interest from the first point cloud data includes:

[0023] Obtain the X, Y, and Z coordinates of the second target point in the first point cloud data;

[0024] When the X coordinate value of the second target point is greater than the X threshold, the Y coordinate value is greater than the Y threshold, and the Z coordinate value is greater than the Z threshold, the second target point is taken as a point within the preset region of interest.

[0025] Based on the above technical means, the preset region of interest in this invention can be understood as the region within a certain range from the vehicle. Only the point cloud data of the region within a certain range from the vehicle is processed, avoiding the unnecessary consumption of computing resources caused by processing all the first point cloud data.

[0026] Further, determining the impassable pixels in each column based on the depth values ​​of two adjacent pixels in each column of the depth image includes:

[0027] Traverse each column of pixels in the depth image from bottom to top, and determine the second angle between the first line segment and the second line segment based on the depth value of each column of two adjacent pixels and the first angle between two adjacent laser beams; wherein, the first line segment is the line segment between the origin of all laser beams and the uppermost pixel among the two adjacent pixels, and the second line segment is the line segment between the two adjacent pixels.

[0028] When the second included angle is greater than the angle threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel.

[0029] When the second included angle is less than the angle threshold, the lower pixel among the two adjacent pixels is determined to be an impassable point.

[0030] According to the above technical means, when the second included angle is greater than the angle threshold, it means that the pixel located higher in the two adjacent pixels does not fall on the ground, and is therefore identified as an impassable pixel; when the second included angle is less than the angle threshold, it means that the pixel located lower in the two adjacent pixels does not fall on the ground, and is therefore identified as an impassable pixel; when the second included angle is equal to the angle threshold, it means that both adjacent pixels fall on the ground, and neither of the two adjacent pixels is an impassable pixel. Determining the boundary points of the drivable area through neighborhood information not only requires less computing power, but also ensures the robustness and accuracy of drivable space detection.

[0031] Furthermore, before determining that the pixel higher in one of the two adjacent pixels is an impassable pixel, the process further includes:

[0032] Convert all pixels within a range of N rows and N columns around the uppermost pixel in the two adjacent pixels into sixth point cloud data;

[0033] Determine the variance of the Z-coordinate values ​​of all points in the sixth point cloud data;

[0034] The step of determining that the pixel higher in one of two adjacent pixels is a non-passable pixel includes:

[0035] When the variance is greater than the variance threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel.

[0036] Furthermore, it also includes: when the variance is less than or equal to the variance threshold, determining the pixel that is higher in the two adjacent pixels as the passable pixel.

[0037] Based on the aforementioned technical means, this invention calculates the variance of the Z-coordinate values ​​of all points in the sixth point cloud data to determine the degree of point cloud height distribution of pixels located higher in a given pair of adjacent pixels. This distribution degree is used to confirm whether a pixel is truly impassable, ensuring the accuracy of drivable area detection. If the variance is greater than a variance threshold, it indicates a discrete distribution, meaning the point cloud is not located on the ground and there is a high probability of impassable pixels. Therefore, pixels located higher in a given pair of adjacent pixels are confirmed as impassable. If the variance is less than or equal to the variance threshold, it indicates a relatively concentrated distribution, meaning the point cloud is likely located on the ground and there is a low probability of impassable pixels. Therefore, pixels located higher in a given pair of adjacent pixels are rejected as impassable.

[0038] A detection device for a drivable area, the detection device comprising:

[0039] The acquisition unit is used to acquire point cloud data collected by the vehicle-mounted LiDAR.

[0040] The conversion unit is used to convert point cloud data in the lidar coordinate system into first point cloud data in the vehicle coordinate system.

[0041] The processing unit is used to filter the first point cloud data to obtain the second point cloud data; wherein the second point cloud data is the point cloud data excluding noise points within a preset region of interest;

[0042] The conversion unit is also used to convert the second point cloud data into a depth image;

[0043] The processing unit is also used to determine the nearest impassable pixel in each column to the vehicle based on the depth values ​​of two adjacent pixels in each column of the depth image.

[0044] The conversion unit is also used to convert all impassable pixels in all columns of the depth image into third point cloud data;

[0045] The processing unit is also used to determine the drivable area of ​​the vehicle based on the third point cloud data.

[0046] A drivable area detection device includes: a processor and a memory configured to store a computer program capable of running on the processor, wherein the processor is configured to execute the steps of the drivable area detection method described above when running the computer program.

[0047] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a method for detecting a drivable area.

[0048] The beneficial effects of this invention are:

[0049] (1) In this invention, the three-dimensional second point cloud data is reduced in dimension and converted into a two-dimensional depth image. The reduction in dimension saves the computing power consumed by traversing the point cloud.

[0050] (2) The present invention identifies the boundary points of the drivable area based on the neighborhood information of all columns of pixels in the depth image, effectively solving the detection of the drivable area in unstructured road scenes, greatly ensuring the robustness, accuracy and real-time performance of the drivable space detection; at the same time, it also saves a lot of hardware computing resources, with low computing power requirements and low hardware requirements. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the first process of the method for detecting drivable areas in an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of a laser beam as exemplified in an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the second process of the method for detecting drivable areas in an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the third process of the method for detecting drivable areas in an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram showing the position of the laser beam emission origin relative to pixel B and pixel A in an embodiment of the present invention;

[0056] Figure 6 This is a schematic diagram of the fourth process of the method for detecting drivable areas in an embodiment of the present invention;

[0057] Figure 7 This is a schematic diagram of the composition structure of the detection device for the drivable area in an embodiment of the present invention;

[0058] Figure 8 This is a schematic diagram of the structure of the detection device for the drivable area in an embodiment of the present invention. Detailed Implementation

[0059] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0060] This invention provides a method for detecting drivable areas. Figure 1 This is a schematic diagram of the first process of the method for detecting drivable areas in an embodiment of the present invention, as shown below. Figure 1 As shown, the detection method for this drivable area may specifically include:

[0061] Step 101: Acquire point cloud data collected by the vehicle-mounted LiDAR and convert the point cloud data in the LiDAR coordinate system into the first point cloud data in the vehicle coordinate system.

[0062] LiDAR (Light Detection and Ranging) is an active sensor that senses its surroundings by emitting a laser beam and measuring the time and intensity of the beam's return, thus acquiring information about the distance and location of obstacles. LiDAR boasts advantages such as highly reliable obstacle detection capabilities, high detection accuracy, and high resolution, making it increasingly widely used in automotive sensing systems.

[0063] Figure 2 This is a schematic diagram of a laser beam as exemplified in an embodiment of the present invention, such as... Figure 2 As shown, the central rectangular part represents a vehicle equipped with a lidar. The lidar consists of multiple lasers. Assuming the lidar is perpendicular to the ground and there are no other obstacles besides the ground, the beams emitted by the multiple lasers hit the ground, forming points. After rotating 360 degrees, these points form a line, which, from a top-down view, resembles a circle of concentric circles.

[0064] The lidar coordinate system is a three-dimensional Cartesian coordinate system with the laser emission center as the origin. The vehicle coordinate system is a three-dimensional Cartesian coordinate system with the vehicle center as the origin.

[0065] Based on the transformation relationship between the lidar coordinate system and the vehicle coordinate system, the point cloud data in the lidar coordinate system is converted into the first point cloud data in the vehicle coordinate system.

[0066] Step 102: Filter the first point cloud data to obtain the second point cloud data; wherein, the second point cloud data is the point cloud data excluding noise points within a preset region of interest.

[0067] Since laser beams have a long range, this invention reduces the computational load of point cloud data by filtering the first point cloud data, selecting point cloud data within a preset Region of Interest (ROI), and further removing noise points within the preset ROI to finally obtain the second point cloud data.

[0068] The preset region of interest can be understood as the area within a certain range of the vehicle.

[0069] Noise points can be understood as reflection points. Because there may be obstacles with reflective properties on the ground, such as glass or water surfaces, when the laser beam hits an obstacle with reflective properties, it will be reflected to other places, and then reflected back to the lidar from other places, which will prolong the reflection time and thus lengthen the distance between the reflection point and the lidar. It is considered to be a point below the ground, so it is necessary to remove the reflection points.

[0070] Step 103: Convert the second point cloud data into a depth image, and based on the depth values ​​of the two adjacent pixels in each column of the depth image, determine the impassable pixel in each column that is closest to the vehicle.

[0071] Based on the transformation relationship between the vehicle coordinate system (i.e., the three-dimensional Cartesian coordinate system) and the two-dimensional Cartesian coordinate system, the second point cloud data in the vehicle coordinate system is converted into a depth image. The depth image includes length information, width information, and depth information. In this invention, the three-dimensional point cloud is reduced in dimensionality and converted into a two-dimensional depth image, which saves the computational power consumed by traversing the point cloud.

[0072] In this invention, to determine the nearest impassable pixel in each column of the depth image to the vehicle, it is necessary to traverse the pixels in each column of the depth image from bottom to top.

[0073] In this invention, the pixels in each column of the depth image are traversed from bottom to top. When the impassable pixels in each column are determined based on the depth values ​​of two adjacent pixels in each column of the depth image, the traversal of the pixels in that column is ended. This can avoid the unnecessary consumption of computing resources caused by blindly processing all pixels.

[0074] Step 104: Convert all impassable pixels in all columns of the depth image into third point cloud data, and determine the drivable area of ​​the vehicle based on the third point cloud data.

[0075] Based on the transformation relationship between the vehicle coordinate system (i.e., the 3D Cartesian coordinate system) and the 2D Cartesian coordinate system, all impassable pixels in all columns of the depth image are converted into third-point cloud data. The third-point cloud data are connected to form the boundary of the drivable area, and the vehicle is automatically driven to travel within the drivable area based on the boundary of the drivable area.

[0076] Here, the execution entity for steps 101 to 104 can be the processor of the detection device for the drivable area.

[0077] In this embodiment of the invention, the boundary points of the drivable area are identified based on the neighborhood information of all columns of pixels in the depth image, which effectively solves the problem of drivable area detection in unstructured road scenes and greatly ensures the robustness, accuracy and real-time performance of drivable space detection.

[0078] Based on the above embodiments, this invention further provides a method for detecting drivable areas. Figure 3 This is a schematic diagram of the second process of the method for detecting drivable areas in an embodiment of the present invention, as shown below. Figure 3 As shown, the detection method for this drivable area may specifically include:

[0079] Step 301: Acquire point cloud data collected by the vehicle-mounted LiDAR and convert the point cloud data in the LiDAR coordinate system into the first point cloud data in the vehicle coordinate system.

[0080] Step 302: Filter out all point cloud data within the preset region of interest from the first point cloud data.

[0081] Since laser beams have a long range, this invention aims to reduce the computational load of point cloud data by selecting all point cloud data within a preset region of interest from the initial point cloud data. This way, subsequent processing only needs to address the point cloud data within the preset region of interest. The preset region of interest can be understood as an area within a certain range of the vehicle.

[0082] In some embodiments, step 302 may specifically include: obtaining the X coordinate value, Y coordinate value and Z coordinate value of the second target point in the first point cloud data;

[0083] When the X coordinate value of the second target point is less than the X threshold, the Y coordinate value is less than the Y threshold, and the Z coordinate value is less than the Z threshold, the second target point is taken as a point within the preset region of interest; all the second target points in the first point cloud data constitute all point cloud data within the preset region of interest.

[0084] The second target point is any point in the first cloud data.

[0085] The X threshold is the maximum X-coordinate value of a point within a preset region of interest, the Y threshold is the maximum Y-coordinate value of a point within a preset region of interest, and the Z threshold is the maximum Z-coordinate value of a point within a preset region of interest. These parameters (X threshold, Y threshold, and Z threshold) can be verified and adjusted to obtain empirical values ​​based on actual test data.

[0086] Step 303: Filter out the fourth point cloud data that is within a preset range from the vehicle from all point cloud data in the preset region of interest.

[0087] Combination Figure 2 The vehicle's preset range can be understood as the range consisting of the preset ring numbers closest to the vehicle.

[0088] The present invention only performs noise reduction on the fourth point cloud data within a preset range from the vehicle (step 304) because the noise pattern of the point cloud data outside the preset range from the vehicle within the preset region of interest is not obvious and is prone to misoperation.

[0089] Step 304: Remove noise points from the fourth point cloud data to obtain the fifth point cloud data.

[0090] Noise points can be understood as reflection points. Because there may be obstacles with reflective properties on the ground, such as glass or water surfaces, when the laser beam hits an obstacle with reflective properties, it will be reflected to other places, and then reflected back to the lidar from other places, which will prolong the reflection time and thus lengthen the distance between the reflection point and the lidar. It is considered to be a point below the ground, so it is necessary to remove the reflection points.

[0091] In some embodiments, step 304 may specifically include: obtaining the Z coordinate value and reflected signal intensity value of the first target point in the fourth point cloud data;

[0092] When the Z-coordinate value of the first target point is less than the first Z-th threshold and the reflected signal intensity value is less than the signal intensity threshold, the first target point is removed from the fourth point cloud data as a noise point to obtain the fifth point cloud data.

[0093] The first target point is any point in the fourth cloud data.

[0094] This invention uses a Reflected Noise Removal (RNR) method based on a three-dimensional lidar reflection model to eliminate reflection noise in point clouds. This method iterates through a fourth set of point cloud data within a preset distance from the vehicle, removing data points whose Z-coordinate values ​​are less than a first Z-th threshold and whose reflected signal intensity values ​​are less than a signal intensity threshold. These points are considered to be reflection noise points with very low intensity and low height.

[0095] Combination Figure 2 The vehicle's preset range can be understood as the range consisting of the preset ring numbers closest to the vehicle.

[0096] The preset number of rings, the first Z threshold, and the signal strength threshold can be verified and adjusted to obtain empirical values ​​based on actual test data. The preset number of rings can be set to 20 and the signal strength threshold can be set to 0.2.

[0097] Step 305: The fifth point cloud data and the point cloud data other than the fourth point cloud data in all point cloud data in the preset region of interest are combined to form the second point cloud data.

[0098] Step 306: Convert the second point cloud data into a depth image, and based on the depth values ​​of the two adjacent pixels in each column of the depth image, determine the impassable pixel in each column that is closest to the vehicle.

[0099] Based on the transformation relationship between the vehicle coordinate system (i.e., the three-dimensional Cartesian coordinate system) and the two-dimensional Cartesian coordinate system, the second point cloud data in the vehicle coordinate system is converted into a depth image. The depth image includes length information, width information, and depth information. In this invention, the three-dimensional point cloud is reduced in dimensionality and converted into a two-dimensional depth image, which saves the computational power consumed by traversing the point cloud.

[0100] In this invention, to determine the nearest impassable pixel in each column of the depth image to the vehicle, it is necessary to traverse the pixels in each column of the depth image from bottom to top.

[0101] In this invention, the pixels in each column of the depth image are traversed from bottom to top. When the impassable pixels in each column are determined based on the depth values ​​of two adjacent pixels in each column of the depth image, the traversal of the pixels in that column is ended. This can avoid the unnecessary consumption of computing resources caused by blindly processing all pixels.

[0102] Step 307: Convert all impassable pixels in all columns of the depth image into third point cloud data, and determine the drivable area of ​​the vehicle based on the third point cloud data.

[0103] Based on the transformation relationship between the vehicle coordinate system (i.e., the 3D Cartesian coordinate system) and the 2D Cartesian coordinate system, all impassable pixels in all columns of the depth image are converted into third-point cloud data. The third-point cloud data are connected to form the boundary of the drivable area, and the vehicle is automatically driven to travel within the drivable area based on the boundary of the drivable area.

[0104] Based on the above embodiments, this invention further provides a method for detecting drivable areas. Figure 4 This is a schematic diagram of the third process of the drivable area detection method in an embodiment of the present invention, as shown below. Figure 4 As shown, the detection method for this drivable area may specifically include:

[0105] Step 401: Acquire point cloud data collected by the vehicle-mounted LiDAR and convert the point cloud data in the LiDAR coordinate system into the first point cloud data in the vehicle coordinate system.

[0106] Step 402: Filter the first point cloud data to obtain the second point cloud data; wherein, the second point cloud data is the point cloud data excluding noise points within a preset region of interest.

[0107] Step 403: Convert the second point cloud data into a depth image.

[0108] Based on the transformation relationship between the vehicle coordinate system (i.e., the three-dimensional Cartesian coordinate system) and the two-dimensional Cartesian coordinate system, the second point cloud data in the vehicle coordinate system is converted into a depth image. The depth image includes length information, width information, and depth information. In this invention, the three-dimensional point cloud is reduced in dimensionality and converted into a two-dimensional depth image, which saves the computational power consumed by traversing the point cloud.

[0109] In this invention, to determine the impassable pixel closest to the vehicle in each column of the depth image, it is necessary to traverse the image pixels column by column starting from the starting point of the bottom row of the depth image. Here, the starting point of the row is the starting point of the horizontal rotation of the lidar laser beam.

[0110] Step 404: Traverse each column of pixels in the depth image from bottom to top. Based on the depth value of two adjacent pixels in each column and the first angle between two adjacent laser beams, determine the second angle between the first line segment and the second line segment. The first line segment is the line segment between the origin of all laser beams and the uppermost pixel among two adjacent pixels, and the second line segment is the line segment between two adjacent pixels.

[0111] In the depth image, each pair of adjacent pixels in each column is measured by laser beams emitted from two adjacent lasers mounted vertically side by side on a lidar system. The first included angle between the two adjacent laser beams is a fixed value, specified and provided by the lidar manufacturer, and is assumed to be α here.

[0112] Combination Figure 5 As shown, starting from the bottom of a column in the depth image and traversing upwards, the first pixel is designated as point B, and the next adjacent pixel as point A. Assuming point O is the origin of all laser beams, the laser beams at points A and B are OA and OB, respectively, forming a triangle. OA is designated as d1, calculated based on the X, Y, and Z coordinates of point A. OB is designated as d2, calculated based on the X, Y, and Z coordinates of point B. Here, the depth value... The first included angle between two adjacent laser beams is α (a known value). Assuming the second included angle between OA (i.e. the first line segment) and AB (i.e. the second line segment) is β, β can be calculated using the following formula (1) based on the knowledge of triangle geometry.

[0113]

[0114] Step 405: Is the second included angle greater than the angle threshold? If the second included angle is greater than the angle threshold, proceed to step 406; if the second included angle is less than the angle threshold, proceed to step 407.

[0115] Step 406: Determine that the pixel located higher in a pair of adjacent pixels is the impassable pixel.

[0116] Step 407: Determine that the lower pixel among two adjacent pixels is the impassable pixel.

[0117] The angle threshold refers to the angle between OA and the ground when point A lands on the ground. The angle threshold can be set to 12°.

[0118] Here, when the second included angle β is greater than the angle threshold, point A is considered an impassable pixel (point A is the pixel that is higher in the two adjacent pixels); when the second included angle β is less than the angle threshold, point B is considered an impassable pixel (point B is the pixel that is lower in the two adjacent pixels).

[0119] In this invention, when the second included angle equals the angle threshold, it indicates that both points B and A fall on the ground, meaning neither pixel is an impassable pixel. Then, based on pixel A and the next pixel, it is determined whether an impassable pixel exists, and so on.

[0120] Step 408: Convert all impassable pixels in all columns of the depth image into third point cloud data, and determine the drivable area of ​​the vehicle based on the third point cloud data.

[0121] In this invention, to further confirm whether the impassable pixels determined in step 406 are truly impassable pixels, the following steps are included before determining that the pixel higher in a pair of adjacent pixels is an impassable pixel:

[0122] Step 601: Convert all pixels within a range of N rows and N columns around the uppermost pixel in two adjacent pixels into sixth point cloud data.

[0123] The pixel coordinates of the uppermost pixel in two adjacent pixels are known, namely the row number and column number. Here, a point count threshold N is set, and all pixels within a range of N rows and N columns around the pixel are treated as a set U. All pixels in set U are then converted into sixth point cloud data.

[0124] Step 602: Determine the variance of the Z coordinate values ​​of all points in the sixth point cloud data.

[0125] Step 603: Determine whether the variance is greater than the variance threshold; if the variance is greater than the variance threshold, proceed to step 604; if the variance is less than or equal to the variance threshold, proceed to step 605.

[0126] Step 604: Determine that the pixel located higher in a pair of adjacent pixels is the impassable pixel.

[0127] Step 605: Determine the pixel that is higher in one of two adjacent pixels as the passable pixel.

[0128] Variance is a measure of the dispersion of a random variable or a set of data in probability theory and statistics. This invention determines the degree of point cloud height distribution by calculating the variance of the Z-coordinate values ​​of all points in the sixth point cloud data.

[0129] In this invention, a variance threshold is preset. If the variance is greater than the variance threshold, it is considered that the height distribution of the point cloud in the pixel set is discrete, the point cloud is not located on the ground, and there is a high probability that there are impassable pixels. Therefore, the pixel in the upper position of two adjacent pixels is identified as an impassable pixel, and the information of the pixel is stored in the container FreeSpacePt. The traversal of this column of pixels ends, and the next column of pixels is traversed. If the variance is less than or equal to the variance threshold, it is considered that the height distribution of the point cloud in the pixel set is relatively concentrated, the point cloud is likely located on the ground, and there is a low probability that there are impassable pixels. Therefore, the pixel in the upper position of two adjacent pixels is rejected as an impassable pixel, and the next pixel in the column is traversed to see if it is an impassable pixel.

[0130] It should be noted that the present invention will further confirm whether the impassable pixels determined in step 407 are truly impassable pixels. The specific method and steps are as described in steps 601 to 605, and will not be repeated here.

[0131] To implement the method of the embodiments of the present invention, based on the same inventive concept, the embodiments of the present invention also provide a drivable area detection device. Figure 7 This is a schematic diagram of the structure of the detection device for the drivable area in an embodiment of the present invention, as shown below. Figure 7 As shown, the detection device 70 for the drivable area includes:

[0132] Acquisition unit 701 is used to acquire point cloud data collected by vehicle-mounted lidar;

[0133] The conversion unit 702 is used to convert point cloud data in the lidar coordinate system into first point cloud data in the vehicle coordinate system.

[0134] The processing unit 703 is used to filter the first point cloud data to obtain second point cloud data; wherein, the second point cloud data is point cloud data excluding noise points within a preset region of interest;

[0135] The conversion unit 702 is further configured to convert the second point cloud data into a depth image;

[0136] The processing unit 703 is further configured to determine the impassable pixels in each column based on the depth values ​​of two adjacent pixels in each column of all columns of the depth image.

[0137] The conversion unit 702 is also used to convert all columns of impassable pixels in the depth image into third point cloud data;

[0138] The processing unit 703 is also used to determine the drivable area of ​​the vehicle based on the third point cloud data.

[0139] In this embodiment of the invention, the boundary points of the drivable area are identified based on the neighborhood information of all columns of pixels in the depth image, which effectively solves the problem of drivable area detection in unstructured road scenes and greatly ensures the robustness, accuracy and real-time performance of drivable space detection.

[0140] In some embodiments, the processing unit 703 is further configured to filter out all point cloud data within the preset region of interest from the first point cloud data;

[0141] Filter out the fourth point cloud data from all point cloud data within the preset region of interest, which is within a preset range of the vehicle;

[0142] Noise points are removed from the fourth point cloud data to obtain the fifth point cloud data;

[0143] The fifth point cloud data and the point cloud data other than the fourth point cloud data in all point cloud data within the preset region of interest constitute the second point cloud data.

[0144] In some embodiments, the processing unit 703 is further configured to obtain the Z coordinate value and the reflected signal intensity value of the first target point in the fourth point cloud data;

[0145] When the Z-coordinate value of the first target point is less than the first Z-th threshold and the reflected signal intensity value is less than the signal intensity threshold, the first target point is removed from the fourth point cloud data as a noise point to obtain the fifth point cloud data.

[0146] In some embodiments, the processing unit 703 is further configured to obtain the X coordinate value, Y coordinate value and Z coordinate value of the second target point in the first point cloud data;

[0147] When the X coordinate value of the second target point is less than the X threshold, the Y coordinate value is less than the Y threshold, and the Z coordinate value is less than the Z threshold, the second target point is taken as a point within the preset region of interest.

[0148] All the second target points in the first point cloud data constitute all point cloud data within the preset region of interest.

[0149] In some embodiments, the processing unit 703 is further configured to traverse each column of pixels in the depth image from bottom to top, and determine a second angle between a first line segment and a second line segment based on the depth value of two adjacent pixels in each column and a first angle between two adjacent laser beams; wherein, the first line segment is the line segment between the origin of all laser beams and the uppermost pixel among the two adjacent pixels, and the second line segment is the line segment between the two adjacent pixels;

[0150] When the second included angle is greater than the angle threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel.

[0151] When the second included angle is less than the angle threshold, the lower pixel among the two adjacent pixels is determined to be an impassable point.

[0152] In some embodiments, the processing unit 703 is further configured to convert all pixels within a range of N rows and N columns around the uppermost pixel in the two adjacent pixels into sixth point cloud data.

[0153] Determine the variance of the Z-coordinate values ​​of all points in the sixth point cloud data;

[0154] The step of determining that the pixel higher in one of two adjacent pixels is a non-passable pixel includes:

[0155] When the variance is greater than the variance threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel.

[0156] In some embodiments, the processing unit 703 is further configured to determine, when the variance is less than or equal to the variance threshold, the pixel that is higher in the two adjacent pixels as a passable pixel.

[0157] This invention also provides another device for detecting drivable areas. Figure 8 This is a schematic diagram of the structure of the detection equipment for the drivable area in an embodiment of the present invention, as shown below. Figure 8 As shown, the detection device 80 for the drivable area includes: a processor 801 and a memory 802 configured to store computer programs capable of running on the processor;

[0158] When the processor 801 is configured to run a computer program, it executes the method steps described in the foregoing embodiments.

[0159] Of course, in practical applications, such as Figure 8As shown, the various components in the detection device 80 for the drivable area are coupled together via a bus system 803. It is understood that the bus system 803 is used to enable communication between these components. In addition to a data bus, the bus system 803 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 8 The general labeled all buses as Bus System 803.

[0160] In practical applications, the aforementioned processor can be at least one of the following: Application-Specific Integrated Circuit (ASIC), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field-Programmable Gate Array (FPGA), controller, microcontroller, and microprocessor. It is understood that, for different devices, the electronic devices used to implement the functions of the aforementioned processor can also be other types, and this embodiment of the invention does not impose specific limitations.

[0161] The aforementioned memory can be volatile memory, such as random-access memory (RAM); or non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); or a combination of the above types of memory, and provides instructions and data to the processor.

[0162] In an exemplary embodiment, the present invention also provides a computer-readable storage medium for storing a computer program.

[0163] Optionally, the computer-readable storage medium can be applied to any of the methods in the embodiments of the present invention, and the computer program causes the computer to execute the corresponding processes implemented by the processor in the various methods of the embodiments of the present invention. For the sake of brevity, these will not be described in detail here.

[0164] In the several embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0165] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Furthermore, in the various embodiments of the present invention, all functional units can be integrated into one processing module, or each unit can be a separate unit, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units. Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0167] The methods disclosed in the several method embodiments provided by this invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0168] The features disclosed in the several product embodiments provided by this invention can be arbitrarily combined without conflict to obtain new product embodiments.

[0169] The features disclosed in the several method or device embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0170] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting a drivable area, characterized in that, The method includes: Acquire point cloud data collected by the vehicle-mounted LiDAR and convert the point cloud data in the LiDAR coordinate system into the first point cloud data in the vehicle coordinate system; The first point cloud data is filtered to obtain the second point cloud data; wherein, the second point cloud data is the point cloud data excluding noise points within a preset region of interest; The second point cloud data is converted into a depth image, and based on the depth values ​​of two adjacent pixels in each column of the depth image, the nearest impassable pixel in each column is determined. The impassable pixels in all columns of the depth image are converted into third point cloud data, and the drivable area of ​​the vehicle is determined based on the third point cloud data. The step of determining the nearest impassable pixel to the vehicle in each column based on the depth values ​​of two adjacent pixels in each column of the depth image includes: determining the nearest impassable pixel to the vehicle in each column based on the depth values ​​of two adjacent pixels in each column of the depth image and the first angle between two adjacent laser beams.

2. The detection method according to claim 1, characterized in that, The step of filtering the first point cloud data to obtain the second point cloud data includes: Filter all point cloud data within the preset region of interest from the first point cloud data; Filter out the fourth point cloud data from all point cloud data within the preset region of interest, which is within a preset range of the vehicle; Noise points are removed from the fourth point cloud data to obtain the fifth point cloud data; The fifth point cloud data and the point cloud data other than the fourth point cloud data in all point cloud data within the preset region of interest constitute the second point cloud data.

3. The detection method according to claim 2, characterized in that, The step of removing noise points from the fourth point cloud data to obtain the fifth point cloud data includes: Obtain the Z-coordinate value and reflected signal intensity value of the first target point in the fourth point cloud data; When the Z-coordinate value of the first target point is less than the first Z-th threshold and the reflected signal intensity value is less than the signal intensity threshold, the first target point is removed from the fourth point cloud data as a noise point to obtain the fifth point cloud data.

4. The detection method according to claim 2, characterized in that, The step of filtering all point cloud data within the preset region of interest from the first point cloud data includes: Obtain the X, Y, and Z coordinates of the second target point in the first point cloud data; When the X coordinate value of the second target point is less than the X threshold, the Y coordinate value is less than the Y threshold, and the Z coordinate value is less than the Z threshold, the second target point is taken as a point within the preset region of interest; all the second target points in the first point cloud data constitute all point cloud data within the preset region of interest.

5. The detection method according to any one of claims 1 to 4, characterized in that, The step of determining the nearest impassable pixel to the vehicle in each column based on the depth values ​​of two adjacent pixels in each column of the depth image includes: Traverse each column of pixels in the depth image from bottom to top, and determine the second angle between the first line segment and the second line segment based on the depth value of each column of two adjacent pixels and the first angle between two adjacent laser beams; wherein, the first line segment is the line segment between the origin of all laser beams and the uppermost pixel among the two adjacent pixels, and the second line segment is the line segment between the two adjacent pixels. When the second included angle is greater than the angle threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel. When the second included angle is less than the angle threshold, the lower pixel among the two adjacent pixels is determined to be an impassable point; the angle threshold refers to the included angle between the first line segment and the ground when the upper pixel among the two adjacent pixels falls on the ground.

6. The detection method according to claim 5, characterized in that, Before determining that the pixel higher in one of the two adjacent pixels is an impassable pixel, the process also includes: Convert all pixels within a range of N rows and N columns around the uppermost pixel in the two adjacent pixels into sixth point cloud data; Determine the variance of the Z-coordinate values ​​of all points in the sixth point cloud data; The step of determining that the pixel higher in one of two adjacent pixels is a non-passable pixel includes: When the variance is greater than the variance threshold, the pixel that is higher in the two adjacent pixels is determined to be an impassable pixel.

7. The detection method according to claim 6, characterized in that, Also includes: When the variance is less than or equal to the variance threshold, the pixel that is higher in the two adjacent pixels is determined to be a passable pixel.

8. A device for detecting a drivable area, characterized in that, The detection device includes: The acquisition unit is used to acquire point cloud data collected by the vehicle-mounted LiDAR. The conversion unit is used to convert point cloud data in the lidar coordinate system into first point cloud data in the vehicle coordinate system. The processing unit is used to filter the first point cloud data to obtain the second point cloud data; wherein the second point cloud data is the point cloud data excluding noise points within a preset region of interest; The conversion unit is also used to convert the second point cloud data into a depth image; The processing unit is also used to determine the impassable pixels in each column based on the depth values ​​of two adjacent pixels in each column of the depth image. The conversion unit is also used to convert all impassable pixels in all columns of the depth image into third point cloud data; The processing unit is also configured to determine the drivable area of ​​the vehicle based on the third point cloud data; The processing unit is specifically used to determine the nearest impassable pixel in each column to the vehicle based on the depth value of two adjacent pixels in each column of the depth image and the first included angle between two adjacent laser beams.

9. A detection device for drivable areas, characterized in that, The detection device includes: a processor and a memory configured to store computer programs capable of running on the processor. Wherein, when the processor is configured to run the computer program, it performs the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 7.

Images