Platform detection method, device, apparatus and medium
By performing vehicle coordinate system transformation, filtering, and fitting clustering on multi-line LiDAR point cloud data, the problem of low platform detection accuracy in existing technologies is solved, providing accurate platform lines unaffected by environmental factors and ensuring the safety of autonomous vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- UISEE TECH BEIJING LTD
- Filing Date
- 2022-10-10
- Publication Date
- 2026-07-14
AI Technical Summary
In existing technologies, platform detection methods based on cameras and lidar have low accuracy and cannot provide accurate platform lines for autonomous vehicles. They are also affected by factors such as weather, lighting, and shadows.
By acquiring multiple raw point cloud data of the vehicle, the vehicle body coordinate system is transformed, the basic platform point cloud data is filtered out, and the platform lines on both sides of the vehicle are extracted based on the lateral distance and fitted clustering.
It enables accurate platform line detection unaffected by weather, lighting, and shadows, improving the accuracy of platform detection and ensuring the safety of autonomous vehicles.
Smart Images
Figure CN115615337B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer technology, and in particular to a platform detection method, apparatus, equipment, and medium. Background Technology
[0002] With the increasing prevalence and widespread use of autonomous vehicles, goods are typically transported between warehouses within a factory using these vehicles, enabling inter-warehouse cargo scheduling. The factory warehouses are equipped with multiple unloading platforms for loading and unloading goods, and multiple parking platforms for autonomous vehicles. As autonomous vehicles navigate the warehouses, they need to periodically detect the edges of the platforms on both sides of the vehicle to determine the platform lines and prevent collisions or scrapes with the platform edges.
[0003] In related technologies, the commonly used platform detection methods are based on cameras or LiDAR to detect the platform edges. However, images acquired by cameras are easily affected by factors such as weather, lighting, and shadows, while LiDAR-based platform edge detection may contain a large number of non-platform edge points. Both of these factors result in low platform detection accuracy and fail to provide accurate platform lines on both sides of autonomous vehicles. Summary of the Invention
[0004] This invention provides a platform detection method, apparatus, equipment, and medium to solve the problem that the platform detection accuracy in related technologies is low, and it cannot provide accurate platform lines on both sides of the vehicle for autonomous vehicles.
[0005] According to one aspect of the present invention, a platform detection method is provided, comprising:
[0006] Multiple raw point cloud data of the vehicle are acquired, and the raw point cloud data are transformed into vehicle coordinate system to obtain transformed point cloud data; wherein, the raw point cloud data is the point cloud data obtained by scanning the surrounding environment of the vehicle by one laser beam of a multi-line lidar.
[0007] The conversion point cloud data is filtered based on the relative height difference between each conversion point cloud data to obtain the basic station point cloud data that matches the conversion point cloud data.
[0008] Based on the lateral distance from the origin of the vehicle body coordinate system to the point cloud data of each basic station, the point cloud data of each basic station is filtered to obtain the point cloud data of each target station.
[0009] By fitting and clustering the point cloud data of each target platform, the platform lines on both sides of the vehicle are obtained.
[0010] According to another aspect of the present invention, a platform detection device is provided, comprising:
[0011] The data acquisition module is used to acquire multiple raw point cloud data of the vehicle, and transform each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data; wherein, the raw point cloud data is the point cloud data obtained by scanning the surrounding environment of the vehicle by one laser beam of a multi-line lidar.
[0012] The first filtering module is used to filter each of the converted point cloud data according to the relative height difference between each of the converted point cloud data, so as to obtain each basic station point cloud data that matches the converted point cloud data.
[0013] The second filtering module is used to filter the basic station point cloud data according to the lateral distance from the origin of the vehicle coordinate system to each basic station point cloud data to obtain the target station point cloud data.
[0014] The data fitting module is used to fit and cluster the point cloud data of each target platform to obtain the platform lines on both sides of the vehicle.
[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:
[0016] At least one processor;
[0017] and a memory communicatively connected to the at least one processor;
[0018] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the platform detection method according to any embodiment of the present invention.
[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the platform detection method according to any embodiment of the present invention.
[0020] The technical solution of this invention acquires multiple raw point cloud data of a vehicle, transforms each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data; then, it filters each transformed point cloud data based on the relative height difference between them to obtain basic platform point cloud data that match the transformed point cloud data; next, it filters each basic platform point cloud data based on the lateral distance from the origin of the vehicle coordinate system to obtain target platform point cloud data; finally, it performs fitting and clustering on each target platform point cloud data to obtain the platform lines on both sides of the vehicle. This solves the problem in related technologies where platform detection schemes have low accuracy and cannot provide accurate platform lines on both sides of the vehicle for autonomous vehicles. It obtains multi-line LiDAR raw point cloud data that is not affected by weather, lighting, shadows, etc., and then performs multiple processing and filtering on the raw point cloud data to accurately extract target platform point cloud data that matches the data characteristics of platform edge point cloud data. Based on the accurately extracted target platform point cloud data, the platform lines on both sides of the vehicle are obtained, ensuring the accuracy of the platform lines on both sides of the vehicle.
[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1A This is a flowchart of a platform detection method provided in Embodiment 1 of the present invention.
[0024] Figure 1B This is a schematic diagram of raw point cloud data provided in Embodiment 1 of the present invention.
[0025] Figure 1C This is a schematic diagram of platform lines on both sides of a vehicle according to Embodiment 1 of the present invention.
[0026] Figure 2 This is a flowchart of a platform detection method provided in Embodiment 2 of the present invention.
[0027] Figure 3 This is a schematic diagram of the structure of a platform detection device provided in Embodiment 3 of the present invention.
[0028] Figure 4A schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation
[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0030] It should be noted that the terms "target," "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising," "including," and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0031] Figure 1A This is a flowchart of a platform detection method provided in Embodiment 1 of the present invention. This embodiment is applicable to detecting the platform edges on both sides of an autonomous vehicle during its operation to obtain the platform lines on both sides of the vehicle. This method can be executed by a platform detection device, which can be implemented in hardware and / or software and can be configured in an electronic device. For example, the electronic device can be the vehicle controller. Figure 1A As shown, the method includes:
[0032] Step 101: Obtain multiple raw point cloud data of the vehicle, and transform each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data.
[0033] The raw point cloud data is obtained by scanning the surrounding environment of the vehicle using one laser beam from a multi-line lidar.
[0034] Optionally, the vehicle is an autonomous vehicle that drives in the warehouse. It is necessary to detect the platform edges on both sides of the vehicle in a timely manner to obtain the platform lines on both sides of the vehicle, so as to avoid the vehicle body colliding with or scraping the platform edges.
[0035] Optionally, the multi-line lidar is installed on an autonomous vehicle. During vehicle operation, the multi-line lidar emits laser beams to scan the surrounding environment, detecting objects around the vehicle in real time to obtain multiple points (sampling points) on the surfaces of objects around the vehicle, which are then used to construct raw point cloud data.
[0036] Optionally, the sampling points obtained by scanning the surrounding environment of the vehicle with a laser beam constitute a raw point cloud data. The raw point cloud data may include planar point cloud data formed by the laser beam scanning the road surface or platform plane around the vehicle, and platform edge point cloud data formed by the laser beam scanning the platform edge around the vehicle. Multiple sampling points on the road surface or platform plane around the vehicle constitute planar point cloud data. Multiple sampling points on the platform edge around the vehicle constitute platform edge point cloud data.
[0037] Optionally, the multi-line lidar can be a 16-line, 32-line, 64-line, or 128-line lidar. "Line count" represents the number of laser beams the lidar can emit simultaneously. Different laser beams can scan different ranges. The scanning ranges of different laser beams can be non-overlapping or overlapping. For example, the multi-line lidar is a 16-line lidar. The electronic device acquires 16 raw point cloud data points from the vehicle.
[0038] Optionally, each of the original point cloud data is point cloud data in the lidar coordinate system of the multi-line lidar. To facilitate subsequent processing of the original point cloud data based on its positional relationship with the vehicle, it is necessary to transform the original point cloud data into a vehicle coordinate system. The original point cloud data after the vehicle coordinate system transformation are the transformed point cloud data. The vehicle's vehicle coordinate system has its origin (O point) at the projection point from the rear axle center to the ground, its longitudinal axis (y-axis) at the front of the vehicle, its transverse axis (x-axis) at the right side of the vehicle, and its vertical axis (z-axis) upwards.
[0039] Optionally, the original point cloud data can be converted from the lidar coordinate system to the vehicle coordinate system using pre-calibrated transformation parameters from the lidar coordinate system to the vehicle coordinate system.
[0040] Step 102: Filter the converted point cloud data according to the relative height difference between each converted point cloud data to obtain the basic station point cloud data that matches the converted point cloud data.
[0041] Optionally, the step of filtering the converted point cloud data based on the relative height difference between each converted point cloud data to obtain the basic station point cloud data that matches the converted point cloud data includes: performing the following operations for each converted point cloud data: filtering out invalid point cloud data in the converted point cloud data to obtain valid point cloud data; wherein, the invalid point cloud data consists of invalid points whose transmission angle is not within the radar's visible angle range or whose height is greater than a preset height threshold; sorting the valid points in the valid point cloud data according to the transmission angle; grouping the sorted valid points according to a preset window size to obtain multiple windows; determining the attribute status of each window according to the relative height difference between the valid points in each window; wherein, the attribute status of each window is planar data or station candidate data; determining a target continuous window according to the attribute status of each window, preset window conditions, and a preset height range; and determining all valid points included in the target continuous window as basic station point cloud data.
[0042] Optionally, invalid point cloud data in the converted point cloud data is filtered out to obtain valid point cloud data. This includes performing the following operations for each converted point cloud data: detecting whether the emission angle of each sampling point in the converted point cloud data is within the radar's visible angle range and whether its height is less than or equal to a preset height threshold; if the emission angle of the sampling point is within the radar's visible angle range and its height is less than or equal to the preset height threshold, then the sampling point is determined to be a valid point in the converted point cloud data; if the emission angle of the sampling point is not within the radar's visible angle range, or its height is greater than the preset height threshold, then the sampling point is determined to be an invalid point in the converted point cloud data; invalid points in the converted point cloud data are deleted to obtain valid point cloud data. The height of the sampling point is the distance corresponding to the vertical coordinate of the sampling point. The emission angle of the sampling point is the rotation angle of the laser beam when the multi-line lidar emits the laser beam to obtain the sampling point. The preset height threshold can be the product of the platform height and a preset coefficient. For example, the platform height is 45 cm, the preset coefficient is 1.1, and the preset height threshold is 49.5 cm.
[0043] Optionally, the radar's line-of-sight range is the same as that of the multi-line lidar. The radar's line-of-sight range is related to the installation location of the multi-line lidar.
[0044] For example, the radar field of view of a multi-line LiDAR installed on the roof of an autonomous vehicle is [0°, 360°]. The radar field of view of a multi-line LiDAR installed on the front left side of an autonomous vehicle is [-120°, +120°].
[0045] Optionally, the preset height threshold is the sum of the platform height and a preset redundancy value. The platform height is the actual height of the platform in the warehouse. For example, the platform height is 45 cm, and the preset redundancy value is 4.5 cm. The preset redundancy value can be 10% of the platform height.
[0046] Optionally, the valid points in the valid point cloud data are sorted according to the emission angle, including: sorting the valid points in the valid point cloud data from smallest to largest emission angle.
[0047] Optionally, the multi-line lidar mechanically rotates each laser beam 360 degrees to obtain the original point cloud data corresponding to each laser beam. For example, such as... Figure 1B As shown, the original point cloud data are roughly distributed in concentric circles with the LiDAR installation location as the center.
[0048] Optionally, the preset window size can be set according to business needs. For example, the preset window size is 9. A certain valid point cloud dataset includes 320 valid points. Based on the preset window size, all the sorted valid points are grouped, starting with the first valid point, with every 9 valid points forming a window, resulting in 35 sequentially arranged windows. The remaining 5 valid points form the last window. Thus, based on the preset window size, all the sorted valid points are grouped into 36 windows.
[0049] Optionally, determining the attribute state of each window based on the relative height difference between valid points in each window includes: performing the following operations for each window: calculating the relative height difference of each valid point relative to the previous valid point according to the order of the valid points; summing the relative height differences corresponding to each adjacent valid point to obtain the cumulative relative height difference of the window; and determining the attribute state of the window using the following formula:
[0050]
[0051] Where attri(x) is the attribute state of the window, height is the cumulative relative height difference of the window, win_height_thres is the preset height difference threshold, FLAT is the planar data, and UP is the platform alternative data.
[0052] Optionally, the cumulative relative height difference of the window is the sum of the relative height differences of all valid points within the window. The first valid point in the window has no corresponding preceding valid point. The relative height difference between the first valid point and the preceding valid point is 0. A preset height difference threshold can be set according to business requirements.
[0053] Optionally, the window's attribute status is used to characterize that the valid point cloud data in the window is road surface point cloud data formed by laser beam scanning of the road surface or platform edge around the vehicle. Ground data is used to characterize that the valid point cloud data in the window is road surface point cloud data formed by laser beam scanning of the road surface. Platform alternative data is used to characterize that the valid point cloud data in the window is platform edge point cloud data formed by laser beam scanning of the platform edge.
[0054] Optionally, determining the target continuous windows based on the attribute status of each window, preset window conditions, and preset height range includes: filtering continuous windows that meet preset window conditions according to the window sorting; wherein, the preset window conditions are that the attribute status of the first and / or last window in the continuous window is planar data, excluding the first and / or last window whose attribute status is planar data, the attribute status of the other windows in the continuous window is platform candidate data; summing the cumulative relative height differences of each window in each continuous window to obtain the cumulative relative height difference of each continuous window; and determining the continuous windows whose cumulative relative height differences are within the preset height range as target continuous windows.
[0055] In a specific example, the preset window conditions include: the attribute status of the first and last windows in a continuous window is planar data; excluding the first and / or last windows whose attribute status is planar data, the attribute status of the other windows in the continuous window is platform candidate data, which can be expressed by the following formula:
[0056] state = FLAT*1 + UP*n + FLAT*1,
[0057] Where state represents a continuous window, the first FLAT*1 in the formula represents the attribute state of the first window in the continuous window as planar data, the second FLAT*1 in the formula represents the attribute state of the last window in the continuous window as planar data, and UP*n in the formula represents the attribute state of the n other windows in the continuous window as platform candidate data.
[0058] In another specific example, the preset window conditions further include: the first window in the continuous window is planar data, the attribute status of the other windows in the continuous window is planar data, and the attribute status of the other windows is platform candidate data, which can be expressed by the following formula:
[0059] state = FLAT * 1 + UP * n,
[0060] Where state represents a continuous window, FLAT*1 in the formula is used to represent that the attribute state of the first window in the continuous window is planar data, and UP*n in the formula is used to represent that the attribute state of the n other windows in the continuous window is platform candidate data.
[0061] In another specific example, the preset window conditions further include: the attribute state of the last window in the continuous window is planar data, and excluding the last window whose attribute state is planar data, the attribute state of the other windows in the continuous window is platform candidate data, which can be expressed by the following formula:
[0062] state = UP * n + FLAT * 1
[0063] Where state represents a continuous window, FLAT*1 in the formula represents the attribute state of the last window in the continuous window as planar data, and UP*n in the formula represents the attribute state of the n other windows in the continuous window as platform candidate data.
[0064] Optionally, the target continuous window is a continuous window containing point cloud data of the platform edge obtained by scanning the platform edge with a laser beam. Multiple windows arranged in sequence constitute a continuous window. The preset window conditions are used to assist in determining whether a continuous window is the target continuous window. The cumulative relative height of the continuous window is the sum of the cumulative relative height differences of each window in the continuous window.
[0065] Optionally, the upper limit of the preset height range is the product of the platform height and a first coefficient. The lower limit of the preset height range is the product of the platform height and a second coefficient. The platform height is the actual height of the platform in the warehouse. The first coefficient is greater than the second coefficient. For example, the first coefficient is 1.1 and the second coefficient is 0.9.
[0066] Optionally, all valid points included in the target continuous window are determined as the base platform point cloud data. The base platform point cloud data is the platform edge point cloud data obtained by laser beam scanning of the platform edge.
[0067] Therefore, based on the relative height difference between the conversion point cloud data, the sampling points in each conversion point cloud data are initially screened to obtain the platform edge point cloud data obtained by laser beam scanning of the platform edge in each conversion point cloud data.
[0068] Step 103: Based on the lateral distance between each basic station point cloud data and the origin of the vehicle coordinate system, filter each basic station point cloud data to obtain each target station point cloud data.
[0069] Optionally, the step of filtering the basic station point cloud data based on the lateral distance from each basic station point cloud data to the origin of the vehicle coordinate system to obtain target station point cloud data includes: determining the quadrant to which the valid points in each basic station point cloud data belong in the vehicle coordinate system; determining the lateral distance from the valid points in each basic station point cloud data to the origin of the vehicle in the horizontal axis direction; determining the minimum distance among the lateral distances from the valid points in the basic station point cloud data contained in each quadrant to the origin of the vehicle in the horizontal axis direction as the lower distance threshold corresponding to each quadrant; and determining the target station point cloud data based on the lateral distance, the lower distance threshold, and a preset redundancy value.
[0070] Optionally, the vehicle coordinate system includes four quadrants: the first quadrant (x > 0, y > 0), the second quadrant (x < 0, y > 0), the third quadrant (x < 0, y < 0), and the fourth quadrant (x > 0, y < 0). Based on the x and y coordinates of the valid points in the basic station point cloud data, the quadrant to which each valid point in the basic station point cloud data belongs in the vehicle coordinate system is determined to be the first, second, third, or fourth quadrant. The distance corresponding to the x-coordinate is taken as the lateral distance. The lateral distance from the valid point in each basic station point cloud data to the vehicle origin along the horizontal axis is the absolute value of the x-coordinate of the valid point in each basic station point cloud data.
[0071] Optionally, determining the target station point cloud data based on the lateral distance, the lower limit threshold of the distance, and the preset redundancy value includes: performing the following operation for each basic station point cloud data: if the lateral distance from the valid point in the basic station point cloud data to the origin of the vehicle body in the lateral direction is less than or equal to the sum of the lower limit threshold of the distance corresponding to the quadrant to which it belongs and the preset redundancy value, then the valid point in the basic station point cloud data is determined as the target station point cloud data.
[0072] Optionally, the preset redundancy value can be set according to business needs. Typically, the road surface point cloud data formed by the laser beam scanning of the platform edges around the vehicle has a relatively short lateral distance to the vehicle. The target platform point cloud data is selected from the basic platform point cloud data to better match the distance characteristics of the platform edge point cloud data.
[0073] Therefore, based on the lateral distance from the valid points in each of the basic station point cloud data to the origin of the vehicle coordinate system, the valid points in each of the basic station point cloud data are further filtered to obtain point cloud data that better matches the distance characteristics of the station edge point cloud data.
[0074] Step 104: Fit and cluster the point cloud data of each target platform to obtain the platform lines on both sides of the vehicle.
[0075] Optionally, the step of fitting and clustering the point cloud data of each target platform to obtain the platform lines on both sides of the vehicle includes: dividing the valid points in the point cloud data of each target platform into a left point cloud data set or a right point cloud data set according to the sign of the x-coordinate of the valid points; comparing the number of data in the left point cloud data set and the right point cloud data set; if the number of data in the left point cloud data set is greater than the number of data in the right point cloud data set, then determining the left point cloud data set as the first point cloud data set and the right point cloud data set as the second point cloud data set; using a random sampling consensus algorithm to fit and cluster the valid points in the first point cloud data set to obtain the corresponding left platform line; adding the target platform point cloud data that does not meet the fitting threshold in the first point cloud data set to the second point cloud data set; using a random sampling consensus algorithm to fit and cluster the valid points in the second point cloud data set to obtain the corresponding right platform line.
[0076] Optionally, based on the sign of the x-coordinate of the valid points in the point cloud data of each target station, the valid points in the point cloud data of each target station are divided into the left point cloud data set or the right point cloud data set. This includes performing the following operations for each target station point cloud data: if the x-coordinate of the valid point in the target station point cloud data is positive, then the valid point is divided into the right point cloud data set; if the x-coordinate of the valid point in the target station point cloud data is negative, then the valid point is divided into the left point cloud data set.
[0077] Optionally, the number of data points in the left point cloud dataset is the total number of valid points contained in the left point cloud dataset. The number of data points in the right point cloud dataset is the total number of valid points contained in the right point cloud dataset.
[0078] Optionally, based on the sign of the x-coordinate of the valid points in the point cloud data of each target platform, the valid points in the point cloud data of each target platform are divided into a left point cloud data set or a right point cloud data set. The left point cloud data set with a larger number of data points is determined as the first point cloud data set, and the right point cloud data set with a smaller number of data points is determined as the second point cloud data set. The first point cloud data set with a larger number of data points may contain valid points with negative x-coordinates, but which actually correspond to the right edge of the platform of the vehicle. These valid points should belong to the second point cloud data set. During the fitting process of the target platform point cloud data in the first point cloud data set, these valid points usually do not meet the fitting threshold. Therefore, the Random Sample Consensus (RANSAC) algorithm is first used to fit and cluster the valid points in the first point cloud data set to obtain the corresponding left platform line. Then, the valid points in the first point cloud data set that do not meet the fitting threshold are added to the second point cloud data set. Therefore, valid points that should belong to the second point cloud dataset are reassigned to it, making the division between the left and right point cloud datasets more accurate and ensuring the accuracy of the subsequently generated platform lines. Finally, a random sampling consensus algorithm is used to fit and cluster the valid points in the added second point cloud dataset to obtain the corresponding right platform line. The left platform line is the platform edge line on the left side of the vehicle. The right platform line is the platform edge line on the right side of the vehicle.
[0079] Optionally, after comparing the number of data points in the left point cloud data set and the right point cloud data set, the method further includes: if the number of data points in the left point cloud data set is less than the number of data points in the right point cloud data set, then the right point cloud data set is determined as the first point cloud data set, and the left point cloud data set is determined as the second point cloud data set; using a random sampling consensus algorithm, the valid points in the first point cloud data set are fitted and clustered to obtain the corresponding right platform line; valid points in the first point cloud data set that do not meet the fitting threshold are added to the second point cloud data set; using a random sampling consensus algorithm, the valid points in the second point cloud data set are fitted and clustered to obtain the corresponding left platform line.
[0080] Optionally, based on the sign of the x-coordinate of the valid points in the point cloud data of each target platform, the valid points are divided into a left point cloud data set or a right point cloud data set. The right point cloud data set, with a larger number of data points, is determined as the first point cloud data set, and the left point cloud data set, with a smaller number of data points, is determined as the second point cloud data set. The first point cloud data set, with a larger number of data points, may contain valid points with positive x-coordinates that actually correspond to the platform edge on the left side of the vehicle. These valid points should belong to the second point cloud data set. During the fitting process of the target platform point cloud data in the first point cloud data set, these valid points usually do not meet the fitting threshold. Therefore, a random sampling consensus algorithm is first used to fit and cluster the valid points in the first point cloud data set to obtain the corresponding right platform line. Then, the valid points in the first point cloud data set that do not meet the fitting threshold are added to the second point cloud data set. Therefore, valid points that should belong to the second point cloud dataset are reassigned to it, making the division between the left and right point cloud datasets more accurate and ensuring the accuracy of the subsequently generated platform lines. Finally, a random sampling consensus algorithm is used to fit and cluster the valid points in the added second point cloud dataset to obtain the corresponding left platform line. For example, the platform lines on both sides of the vehicle are as follows: Figure 1C As shown.
[0081] The technical solution of this invention acquires multiple raw point cloud data of a vehicle, transforms each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data; then, it filters each transformed point cloud data based on the relative height difference between them to obtain basic platform point cloud data that match the transformed point cloud data; next, it filters each basic platform point cloud data based on the lateral distance from the origin of the vehicle coordinate system to obtain target platform point cloud data; finally, it performs fitting and clustering on each target platform point cloud data to obtain the platform lines on both sides of the vehicle. This solves the problem in related technologies where platform detection schemes have low accuracy and cannot provide accurate platform lines on both sides of the vehicle for autonomous vehicles. It obtains multi-line LiDAR raw point cloud data that is not affected by weather, lighting, shadows, etc., and then performs multiple processing and filtering on the raw point cloud data to accurately extract target platform point cloud data that matches the data characteristics of platform edge point cloud data. Based on the accurately extracted target platform point cloud data, the platform lines on both sides of the vehicle are obtained, ensuring the accuracy of the platform lines on both sides of the vehicle.
[0082] Figure 2 This is a flowchart of a platform detection method provided in Embodiment 2 of the present invention. This embodiment of the present invention can be combined with various optional solutions from one or more of the above embodiments. For example... Figure 2 As shown, the method includes:
[0083] Step 201: Obtain multiple raw point cloud data of the vehicle, and transform each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data.
[0084] The raw point cloud data is obtained by scanning the surrounding environment of the vehicle using one laser beam from a multi-line lidar.
[0085] Step 202: Filter the converted point cloud data according to the relative height difference between each converted point cloud data to obtain the basic station point cloud data that matches the converted point cloud data.
[0086] Step 203: Based on the lateral distance between each of the basic station point cloud data and the origin of the vehicle coordinate system, filter each of the basic station point cloud data to obtain the target station point cloud data.
[0087] Step 204: Based on the sign of the x-coordinate of the valid points in the point cloud data of each target station, divide the valid points in the point cloud data of each target station into the left point cloud data set or the right point cloud data set.
[0088] Step 205: Compare the number of data in the left point cloud data set and the right point cloud data set.
[0089] Step 206: If the number of data in the left point cloud data set is greater than the number of data in the right point cloud data set, then the left point cloud data set is determined as the first point cloud data set, and the right point cloud data set is determined as the second point cloud data set.
[0090] Step 207: Use the random sampling consensus algorithm to fit and cluster the valid points in the first point cloud data set to obtain the corresponding left platform line.
[0091] Step 208: Add the valid points in the first point cloud data set that do not meet the fitting threshold to the second point cloud data set.
[0092] Step 209: Use the random sampling consensus algorithm to fit and cluster the valid points in the second point cloud data set to obtain the corresponding right platform line.
[0093] Optionally, based on the sign of the x-coordinate of the valid points in the point cloud data of each target station, the valid points in the point cloud data of each target station are divided into a left point cloud data set or a right point cloud data set. The left point cloud data set, which has a larger number of data points, is determined as the first point cloud data set, and the right point cloud data set, which has a smaller number of data points, is determined as the second point cloud data set. The first point cloud data set, which has a larger number of data points, may contain valid points with negative x-coordinates that actually correspond to the right edge of the station platform of the vehicle. These valid points should be assigned to the second point cloud data set, as they typically do not meet the fitting threshold during the fitting process of the valid points in the first point cloud data set.
[0094] Therefore, the random sampling consensus algorithm is first used to fit and cluster the valid points in the first point cloud data set to obtain the corresponding left platform line. Then, the valid points in the first point cloud data set that do not meet the fitting threshold are added to the second point cloud data set.
[0095] The technical solution of this invention reclassifies valid points that should belong to the second point cloud data set to the second point cloud data set, making the division of the left and right point cloud data sets more accurate and ensuring the accuracy of the platform lines on both sides of the vehicle.
[0096] Figure 3 This is a schematic diagram of a platform detection device provided in Embodiment 3 of the present invention. The device can be configured in an electronic device. Figure 3 As shown, the device includes: a data acquisition module 301, a first screening module 302, a second screening module 303, and a data fitting module 304.
[0097] The system includes a data acquisition module 301, which acquires multiple raw point cloud data of the vehicle and transforms each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data. The raw point cloud data is obtained by scanning the surrounding environment of the vehicle using a single laser beam from a multi-line lidar. A first filtering module 302 filters each transformed point cloud data based on the relative height difference between them to obtain basic platform point cloud data that matches the transformed point cloud data. A second filtering module 303 filters each basic platform point cloud data based on the lateral distance from the origin of the vehicle coordinate system to obtain target platform point cloud data. A data fitting module 304 performs fitting and clustering on the target platform point cloud data to obtain platform lines on both sides of the vehicle.
[0098] The technical solution of this invention acquires multiple raw point cloud data of a vehicle, transforms each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data; then, it filters each transformed point cloud data based on the relative height difference between them to obtain basic platform point cloud data that match the transformed point cloud data; next, it filters each basic platform point cloud data based on the lateral distance from the origin of the vehicle coordinate system to obtain target platform point cloud data; finally, it performs fitting and clustering on each target platform point cloud data to obtain the platform lines on both sides of the vehicle. This solves the problem in related technologies where platform detection schemes have low accuracy and cannot provide accurate platform lines on both sides of the vehicle for autonomous vehicles. It obtains multi-line LiDAR raw point cloud data that is not affected by weather, lighting, shadows, etc., and then performs multiple processing and filtering on the raw point cloud data to accurately extract target platform point cloud data that matches the data characteristics of platform edge point cloud data. Based on the accurately extracted target platform point cloud data, the platform lines on both sides of the vehicle are obtained, ensuring the accuracy of the platform lines on both sides of the vehicle.
[0099] In an optional embodiment of the present invention, the first filtering module 302 is optionally configured to: perform the following operations for each converted point cloud data: filter out invalid point cloud data in the converted point cloud data to obtain valid point cloud data; wherein, the invalid point cloud data consists of invalid points whose emission angle is not within the radar's visible angle range or whose height is greater than a preset height threshold; sort the valid points in the valid point cloud data according to the emission angle; group the sorted valid points according to a preset window size to obtain multiple windows; determine the attribute status of each window according to the relative height difference between the valid points in each window; wherein, the attribute status of each window is planar data or station candidate data; determine a target continuous window according to the attribute status of each window, preset window conditions, and preset height range; and determine all valid points included in the target continuous window as basic station point cloud data.
[0100] Optionally, when the first filtering module 302 performs the operation of determining the attribute state of each window based on the relative height difference between valid points in each window, it is specifically configured to: perform the following operations for each window: calculate the relative height difference of each valid point relative to the previous valid point according to the sorting of valid points; sum the relative height differences corresponding to each adjacent valid point to obtain the cumulative relative height difference of the window; and determine the attribute state of the window using the following formula:
[0101]
[0102] Where attri(x) is the attribute state of the window, height is the cumulative relative height difference of the window, win_height_thres is the preset height difference threshold, FLAT is the planar data, and UP is the platform alternative data.
[0103] Optionally, when the first filtering module 302 performs the operation of determining the target continuous windows based on the attribute status of each window, preset window conditions, and preset height range, it is specifically used to: filter each continuous window that meets the preset window conditions according to the window sorting; wherein, the preset window conditions are that the attribute status of the first window and / or the last window in the continuous window is planar data, and excluding the attribute status of the first window and / or the last window which is planar data, the attribute status of the other windows in the continuous window is platform candidate data; sum the cumulative relative height differences of each window in each continuous window to obtain the cumulative relative height difference of each continuous window; and determine the continuous windows whose cumulative relative height differences are within the preset height range as target continuous windows.
[0104] Optionally, the second filtering module 303 is specifically used for: determining the quadrant to which the valid points in each of the basic station point cloud data belong in the vehicle coordinate system; determining the lateral distance from the valid points in each of the basic station point cloud data to the vehicle origin in the horizontal axis direction; determining the minimum distance among the lateral distances from the valid points in the basic station point cloud data contained in each quadrant to the vehicle origin in the horizontal axis direction as the lower limit threshold for distance corresponding to each quadrant; and determining the target station point cloud data based on the lateral distance, the lower limit threshold for distance, and a preset redundancy value.
[0105] Optionally, when the second filtering module 303 performs the operation of determining the target station point cloud data based on the lateral distance, the lower limit threshold of the distance, and the preset redundancy value, it is specifically used to perform the following operation for each basic station point cloud data: if the lateral distance from the valid point in the basic station point cloud data to the origin of the vehicle body in the lateral direction is less than or equal to the sum of the lower limit threshold of the distance corresponding to the quadrant to which it belongs and the preset redundancy value, then the valid point in the basic station point cloud data is determined as the target station point cloud data.
[0106] Optionally, the data fitting module 304 is specifically used for: dividing the valid points in the point cloud data of each target platform into a left point cloud data set or a right point cloud data set according to the sign of the x-coordinate of the valid points; comparing the number of data in the left point cloud data set and the right point cloud data set; if the number of data in the left point cloud data set is greater than the number of data in the right point cloud data set, then determining the left point cloud data set as the first point cloud data set and the right point cloud data set as the second point cloud data set; using a random sampling consensus algorithm to perform fitting and clustering on the valid points in the first point cloud data set to obtain the corresponding left platform line; adding the valid points in the first point cloud data set that do not meet the fitting threshold to the second point cloud data set; and using a random sampling consensus algorithm to perform fitting and clustering on the valid points in the second point cloud data set to obtain the corresponding right platform line.
[0107] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0108] The above-mentioned platform detection device can execute the platform detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the platform detection method.
[0109] Figure 4 A schematic diagram of an electronic device 10, which can be used to implement the platform detection method of embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.
[0110] like Figure 4As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or a computer program constructed from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.
[0111] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0112] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as station detection methods.
[0113] In some embodiments, the platform detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is built into RAM 13 and executed by processor 11, one or more steps of the platform detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the platform detection method by any other suitable means (e.g., by means of firmware).
[0114] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0115] Computer programs used to implement the platform detection method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0116] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0117] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0118] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0119] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.
[0120] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.
[0121] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A platform inspection method, characterized in that, include: Multiple raw point cloud data of the vehicle are acquired, and the raw point cloud data are transformed into vehicle coordinate system to obtain transformed point cloud data; wherein, the raw point cloud data is the point cloud data obtained by scanning the surrounding environment of the vehicle by one laser beam of a multi-line lidar. The converted point cloud data is filtered based on the relative height difference between each converted point cloud data to obtain basic station point cloud data that matches the converted point cloud data. Specifically, for each converted point cloud data, the following operations are performed: invalid point cloud data is removed to obtain valid point cloud data; the invalid point cloud data consists of invalid points whose transmission angle is not within the radar's visible angle range or whose height exceeds a preset height threshold; the valid points in the valid point cloud data are sorted according to their transmission angle; the sorted valid points are grouped according to a preset window size to obtain multiple windows; the attribute status of each window is determined based on the relative height difference between the valid points in each window; the attribute status of each window is either planar data or station candidate data; a target continuous window is determined based on the attribute status of each window, preset window conditions, and a preset height range; all valid points included in the target continuous window are determined as basic station point cloud data. Based on the lateral distance from the origin of the vehicle body coordinate system to the point cloud data of each basic station, the point cloud data of each basic station is filtered to obtain the point cloud data of each target station. By fitting and clustering the point cloud data of each target platform, the platform lines on both sides of the vehicle are obtained.
2. The method according to claim 1, characterized in that, The step of determining the attribute state of each window based on the relative height difference between valid points in each window includes: Perform the following operations for each window: Based on the sorting of valid points, calculate the relative height difference of each valid point relative to the previous valid point; The cumulative relative height difference of the window is obtained by summing the relative height differences corresponding to each adjacent valid point; Use the following formula to determine the property state of the window: , in, The attribute state of the window. The cumulative relative height difference of the window. To preset the height difference threshold, For planar data, This is alternative data for the platform.
3. The method according to claim 1, characterized in that, The step of determining the target continuous windows based on the attribute states of each window, preset window conditions, and preset height range includes: Based on the window order, filter consecutive windows that meet preset window conditions; wherein, the preset window conditions are that the attribute status of the first and / or last window in the consecutive window is planar data, and excluding the first and / or last window whose attribute status is planar data, the attribute status of the other windows in the consecutive window is platform candidate data. The cumulative relative height difference of each window in each of the continuous windows is summed to obtain the cumulative relative height difference of each of the continuous windows; Continuous windows with cumulative relative height differences within a preset height range are defined as target continuous windows.
4. The method according to claim 1, characterized in that, The step of filtering the basic station point cloud data based on the lateral distance from the origin of the vehicle coordinate system to obtain the target station point cloud data includes: Determine the quadrant to which the valid points in the point cloud data of each basic station belong in the vehicle body coordinate system; Determine the lateral distance from the vehicle body origin to the valid points in the point cloud data of each basic station in the horizontal axis direction; The minimum distance from the horizontal axis of the valid points in the basic station point cloud data contained in each quadrant to the origin of the vehicle body is determined as the lower limit threshold of the distance for each quadrant. The target station point cloud data is determined based on the lateral distance, the lower limit threshold of the distance, and the preset redundancy value.
5. The method according to claim 4, characterized in that, The step of determining the target station point cloud data based on the lateral distance, the lower limit threshold of the distance, and a preset redundancy value includes: Perform the following operations on the point cloud data of each basic station: If the lateral distance from the valid point in the basic station point cloud data to the origin of the vehicle body in the horizontal direction is less than or equal to the sum of the lower limit threshold of the quadrant to which it belongs and the preset redundancy value, then the valid point in the basic station point cloud data is determined as the target station point cloud data.
6. The method according to claim 1, characterized in that, The step of fitting and clustering the point cloud data of each target platform to obtain the platform lines on both sides of the vehicle includes: Based on the sign of the x-coordinate of the valid points in the point cloud data of each target station, the valid points in the point cloud data of each target station are divided into the left point cloud data set or the right point cloud data set. Compare the number of data points in the left point cloud dataset and the right point cloud dataset; If the number of data in the left point cloud data set is greater than the number of data in the right point cloud data set, then the left point cloud data set is determined as the first point cloud data set, and the right point cloud data set is determined as the second point cloud data set. Using a random sampling consensus algorithm, the valid points in the first point cloud data set are fitted and clustered to obtain the corresponding left platform line. Add valid points in the first point cloud data set that do not meet the fitting threshold to the second point cloud data set; Using a random sampling consensus algorithm, the valid points in the second point cloud data set are fitted and clustered to obtain the corresponding right platform line.
7. A platform detection device, characterized in that, include: The data acquisition module is used to acquire multiple raw point cloud data of the vehicle, and transform each raw point cloud data into a vehicle coordinate system to obtain transformed point cloud data; wherein, the raw point cloud data is the point cloud data obtained by scanning the surrounding environment of the vehicle by one laser beam of a multi-line lidar. The first filtering module is used to filter the converted point cloud data according to the relative height difference between each converted point cloud data to obtain basic station point cloud data that matches the converted point cloud data. Specifically, for each converted point cloud data, the following operations are performed: invalid point cloud data is filtered out to obtain valid point cloud data; the invalid point cloud data consists of invalid points whose transmission angle is not within the radar's visible angle range or whose height is greater than a preset height threshold; the valid points in the valid point cloud data are sorted according to the transmission angle; the sorted valid points are grouped according to a preset window size to obtain multiple windows; the attribute status of each window is determined according to the relative height difference between the valid points in each window; the attribute status of each window is either planar data or station candidate data; a target continuous window is determined according to the attribute status of each window, preset window conditions, and a preset height range; all valid points included in the target continuous window are determined as basic station point cloud data. The second filtering module is used to filter the basic station point cloud data according to the lateral distance from the origin of the vehicle coordinate system to each basic station point cloud data to obtain the target station point cloud data. The data fitting module is used to fit and cluster the point cloud data of each target platform to obtain the platform lines on both sides of the vehicle.
8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the platform detection method according to any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the platform detection method according to any one of claims 1-6.