Sparse grid space scanning method and parking space identification method

Abstract

The invention relates to the technical field of obstacle edge recognition, in particular to a sparse grid space scanning method and a parking space recognition method.The sparse grid space scanning method comprises the steps that the full grid direction and the sparse grid direction are determined under a vehicle coordinate system; acquiring coordinate information of an imaging trace point in the own vehicle coordinate system; all grid information of the coordinate information is reserved in the full grid direction, nearest grid information in the coordinate information is reserved in the sparse grid direction, and edge points are formed. According to the method, the problems of inaccurate radar far-end grid identification and low trace point edge precision caused by adopting polar coordinate system rasterization in existing obstacle edge identification are solved, or the problem that the radar far-end grid identification capability is reserved by adopting rectangular coordinate system rasterization, but large memory space is occupied at the same time is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of obstacle edge recognition, and in particular to a sparse grid space scanning method and a parking space recognition method. Background Art

[0002] With the rapid development of millimeter-wave radar in the automotive industry, the requirements for it are getting higher and higher, and it also plays an indispensable and important role in intelligent driving. Among the various functions of millimeter-wave radar, electronic mapping is indispensable. For example, alarm function, ACC, ACE, parking space and multi-sensor fusion, etc., all require edge recognition of obstacles around the vehicle to obtain the drivable area.

[0003] Among the various obstacle edge recognition algorithms currently popular, most are point trace rasterization. Compared with the traditional polar coordinate system rasterization and rectangular coordinate system rasterization, the two algorithms have their own advantages and disadvantages. The traditional polar coordinate system rasterization occupies less memory space, but the grid is denser when the distance is closer to the radar, but as the distance increases, the grid will become larger and larger, which will increase the difficulty in removing grid isolated points, and it is easy to remove them by mistake, and it will also lose the edge accuracy of the point trace; while the traditional rectangular coordinate system rasterization and polar coordinate system have the opposite performance. The rectangular grid is a uniform grid and has nothing to do with the distance from the radar. It will be much higher in accuracy and precision, but it takes up more memory space, which will increase the difficulty of use for some projects with insufficient memory.

[0004] Therefore, the present invention proposes a sparse grid space scanning method and a parking space recognition method under the goal of considering high-precision obstacle edges and saving computing resources. Summary of the invention

[0005] The invention content of the present invention is to provide a sparse grid space scanning method and a parking space recognition method, which mainly solves the problem of inaccurate radar far-end grid recognition and low point edge precision caused by the use of polar coordinate system rasterization in the existing obstacle edge recognition, or the use of rectangular coordinate system rasterization to retain the radar far-end grid recognition capability, but at the same time it will cause the problem of large memory space occupancy.

[0006] The present invention proposes a sparse grid space scanning method, comprising the following steps:

[0007] Determine the full grid direction and the sparse grid direction in the vehicle coordinate system;

[0008] Obtaining coordinate information of imaging points in the vehicle coordinate system;

[0009] All grid information of the coordinate information is retained along the full grid direction, and the filtered grid information in the coordinate information is retained along the sparse grid direction, and edge points are formed.

[0010] Preferably, the method for determining the full grid direction and the sparse grid in the vehicle coordinate system includes:

[0011] The full grid direction and the sparse grid direction are perpendicular to each other.

[0012] Preferably, the method for determining the full grid direction and the sparse grid in the vehicle coordinate system includes:

[0013] The vehicle coordinate system is established with the center of the vehicle as the origin, the direction of the vehicle's motion as the x-axis, and the direction perpendicular to the motion direction as the y-axis;

[0014] Set the x-axis direction to the full grid direction and the y-axis direction to the sparse grid direction.

[0015] Preferably, retaining all grid information of the coordinate information along the full grid direction, and retaining the nearest grid information of the coordinate information along the sparse grid direction, comprises:

[0016] All grid information of the coordinate information is retained along the full grid direction;

[0017] Determine whether there is an isolated grid in the sparse grid direction according to the coordinate information, and if so, remove the isolated grid and retain the nearest grid information of the coordinate information;

[0018] The grid information retained in the full grid direction and the sparse grid direction is combined to form edge points.

[0019] Preferably, judging whether there is an isolated grid in the sparse grid direction according to the coordinate information includes:

[0020] Determine all grids in the sparse grid direction according to the coordinate information;

[0021] It is determined whether the number of points included in any of the grids and a preset number of grids around it is less than a preset value. If so, the current grid is defined as an isolated grid.

[0022] Preferably, determining whether the number of points included in any of the grids and a preset number of grids around it is less than a preset value includes:

[0023] Taking any of the grids as the center, obtain the number of points in all grids within a preset radius;

[0024] Determine whether the number of points obtained is less than a preset value;

[0025] Among them, all grids within the preset radius are complete grids.

[0026] Preferably, obtaining coordinate information of the imaging point trace in the vehicle coordinate system includes:

[0027] Obtain imaging points within the recognition range of the millimeter-wave radar;

[0028] Obtain the coordinate information of the imaging point trace in the vehicle coordinate system.

[0029] Preferably, obtaining imaging point traces within the recognition range of the millimeter wave radar includes:

[0030] Acquire static point traces within the recognition range of the millimeter wave radar, and perform multi-frame point trace accumulation;

[0031] The imaging point trace is formed after compensating the DR of the vehicle.

[0032] Preferably, the imaging points are distributed in a pre-formed absolute coordinate system;

[0033] Obtaining coordinate information of the imaging point trace in the vehicle coordinate system includes:

[0034] Determine whether the absolute coordinate system where the imaging point trace is located is the same coordinate system as the vehicle coordinate system. If so, directly obtain the coordinate information of the imaging point trace. If not, project the imaging point trace to the vehicle coordinate system and obtain the coordinate information of the projected imaging point trace.

[0035] The present invention also proposes a parking space recognition method, including the above-mentioned sparse grid space scanning method, and further comprising the following steps:

[0036] In the parking state, obtaining the initial edge points according to the sparse grid space scanning method, and fitting the initial edge points to form the parking space edge on the entry side;

[0037] In the parking state, gradually acquiring secondary edge points according to the sparse grid space scanning method, and fitting the secondary edge points to form the parking space edge on the parking side;

[0038] The parking space edge on the entry side and the parking space edge on the stop side are used as two end edges to form a stop area of ​​the current parking space.

[0039] As can be seen from the above, the application of the technical solution provided by the present invention can achieve the following beneficial effects:

[0040] First, the sparse grid spatial scanning method proposed in the present invention uses full rasterization in the direction of focus and sparse rasterization in the vertical direction, which can obtain high-precision and high-accuracy raster information with minimal memory space;

[0041] Second, the sparse grid space scanning method proposed in the present invention mainly considers the accurate identification of the obstacle edges on the left and right sides in the parking state. Therefore, the left and right directions are set as the full grid directions. Under the premise of satisfying the edge point recognition, the amount of calculation data is reduced, thereby reducing the memory occupied, making this method suitable for project calculations with different memory. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 A flowchart of the steps of the sparse grid space scanning method in an embodiment of the present invention;

[0044] Figure 2 This is a simulation effect diagram of the sparse grid space scanning method in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] The existing obstacle edge recognition uses polar coordinate system rasterization, which leads to inaccurate radar far-end grid recognition and low point edge precision. Alternatively, rectangular coordinate system rasterization is used to retain the radar far-end grid recognition capability, but at the same time it will cause a large memory space occupation problem.

[0047] In order to solve the above problem, this embodiment proposes a sparse grid space scanning method, including the following steps:

[0048] S1, determine the full grid direction and the sparse grid direction in the vehicle coordinate system;

[0049] S2, obtaining the coordinate information of the imaging point trace in the vehicle coordinate system;

[0050] S3, retaining all grid information of the coordinate information along the full grid direction, retaining the filtered grid information in the coordinate information along the sparse grid direction, and forming edge points.

[0051] Preferably, in this embodiment, before step S1, there are steps:

[0052] S0, establish the vehicle coordinate system with the center of the vehicle as the origin, the direction of the vehicle's movement as the x-axis, and the direction perpendicular to the movement direction as the y-axis.

[0053] Correspondingly, the full grid direction in step S1 is the x-axis direction, and the y-axis direction is the sparse grid direction, that is, the full grid direction and the sparse grid direction are perpendicular to each other.

[0054] In this embodiment, a specific area can be framed by obtaining the edges in the same direction. Therefore, the method described in this embodiment obtains relatively dense edge points in the direction of the entire grid, and only obtains filtered edge points in the direction of the sparse grid. The filtered edge points in this embodiment are also the nearest edge points, which ensure accurate edge recognition while reducing computing data and occupying memory.

[0055] More specifically, step S3 includes:

[0056] S31, retaining all grid information of coordinate information along the full grid direction;

[0057] S32, judging whether there is an isolated grid in the sparse grid direction according to the coordinate information, if so, removing the isolated grid and retaining the nearest grid information of the coordinate information, otherwise retaining the nearest grid information of the coordinate information;

[0058] S33, combining the grid information retained in the full grid direction and the sparse grid direction to form edge points.

[0059] Preferably, the judgment condition of an isolated grid is to judge whether the number of points included in any grid and a preset number of grids around it is less than a preset value, and if so, the current grid is defined as an isolated grid. Preferably, but not limited to, the preset number of grids around it can be defined as grids within a radius range. Since the grids in this application are square, the grids within the preset range are equivalent to the same number of grids in the up, down, left, and right directions.

[0060] Among them, the judgment condition of the isolated grid in the sparse grid direction is that the isolated grid is judged by taking the grid as the center, and the number of points in multiple grids (for example, 4 grids each) above, below, left and right, including the central grid, is a total of 17 grids. If the number of points is greater than 6, the central grid is not an isolated grid, and the grid information is retained. Preferably, but not limited to, the number of points in the judgment condition can be set according to the threshold of the data characteristics, and different projects can be modified according to the point cloud density characteristics of the radar. Take the nearest several (for example, 3) grids in the sparse grid direction, and after judging the isolated grids of the 3 grid information, only the nearest non-isolated grid is retained. According to the judgment result of the isolated grid, the 3 grid information can retain 0, 1, 2 or 3, and then filter according to the retained grid information to further retain the nearest only grid.

[0061] In this embodiment, the number of points of the preset value may be concentrated on one grid, or scattered in the peripheral and central grids, and the grid located in the center may be determined as a non-isolated grid.

[0062] In this embodiment, the judgment of isolated grids can obtain complete edge points of obstacles around the vehicle.

[0063] More specifically, step S2 includes:

[0064] S21, obtaining imaging points within the recognition range of the millimeter wave radar;

[0065] S22, obtaining coordinate information of the imaging point trace in the vehicle coordinate system.

[0066] Among them, the recognition range of the millimeter wave radar in step S21 is actually to screen imaging points within 15m, that is, the 15m range can effectively cover the parking space data around the vehicle position.

[0067] Wherein, step S21 includes:

[0068] S211, acquiring stationary point traces within the recognition range of the millimeter wave radar, and accumulating multiple frames of point traces;

[0069] S212, forming imaging points after compensating the DR of the vehicle.

[0070] Preferably, an absolute coordinate system with imaging points is established when the DR of the own vehicle is cleared. The own vehicle coordinate system in step S1 is actually a corresponding Cartesian coordinate system established according to the rasterized area. Therefore, it is necessary to determine whether the absolute coordinate system and the own vehicle coordinate system are the same coordinate system. If not, the imaging points need to be projected to the own vehicle coordinate system to obtain the lateral and longitudinal coordinate information in the own vehicle coordinate system.

[0071] In this embodiment, the target information within 15m in front, behind, left and right is scanned with the center of the vehicle as the origin, and the grid size is 0.2mx0.2m. Since the obstacle edge is mainly considered on the left and right of the vehicle, high precision is required for the obstacle edges on the left and right sides. Therefore, the design considers the full grid in front and back to save the edge length of the obstacle, and considers sparse grids on the left and right. The nearest 3 grids are retained on the left, and the nearest 3 grids are also retained on the right to eliminate false targets (only 3 are retained because the nearest grid in most projects is a false target, and it is basically impossible that the nearest 3 grids are all false). The specific data is shown in the following table:

[0072]

[0073]

[0074] As shown in the table above, the total number of sparse rectangular grids is only 4% of the total number of traditional rectangular grids, which is equivalent to a 96% saving in memory consumption, which is a very large amount for actual industrial projects. Due to the reduction in the total number of grids, the computing power will also decrease when removing false targets, which is very practical.

[0075] The sparse grid space scanning method described in the above embodiment can effectively identify edge points with high precision, thereby providing accuracy for subsequent identification of parking spaces.

[0076] Therefore, based on the content of the above embodiment, a parking space recognition method can be expanded, including the following steps:

[0077] In the parking state, the initial edge points are obtained according to the sparse grid space scanning method, and the initial edge points are fitted to form the parking space edge on the entry side;

[0078] In the parking state, the secondary edge points are gradually obtained according to the sparse grid space scanning method, and the secondary edge points are fitted to form the parking space edge on the stopping side;

[0079] The parking space edge on the entry side and the parking space edge on the stop side are used as the two end edges to form the stop area of ​​the current parking space.

[0080] In summary, the sparse grid space scanning method proposed in this embodiment is provided with a full grid direction and a sparse grid direction that are perpendicular to each other. By sacrificing the calculation in an unimportant direction, the number of operations can be effectively reduced and the memory occupied can be reduced. At the same time, the accurate recognition of edge points is guaranteed and the original function of edge point recognition is completed.

[0081] The above-described implementation methods do not constitute a limitation on the protection scope of the technical solution. Any modification, equivalent replacement and improvement made within the spirit and principle of the above-described implementation methods shall be included in the protection scope of the technical solution.

Claims

1. A sparse grid space scanning method, characterized in that: The following steps are involved: Determine the full grid direction and the sparse grid direction in the vehicle coordinate system; Obtaining coordinate information of imaging points in the vehicle coordinate system; All grid information of the coordinate information is retained along the full grid direction, and the filtered grid information in the coordinate information is retained along the sparse grid direction, and edge points are formed.

2. A sparse grid space scanning method according to claim 1, characterized in that: The method for determining the full grid direction and the sparse grid in the vehicle coordinate system includes: The full grid direction and the sparse grid direction are perpendicular to each other.

3. A sparse grid space scanning method according to claim 2, characterized in that: The method for determining the full grid direction and the sparse grid in the vehicle coordinate system includes: The vehicle coordinate system is established with the center of the vehicle as the origin, the direction of the vehicle's motion as the x-axis, and the direction perpendicular to the motion direction as the y-axis; Set the x-axis direction to the full grid direction and the y-axis direction to the sparse grid direction.

4. A sparse grid space scanning method according to any one of claims 1 to 3, characterized in that: Retaining all grid information of the coordinate information along the full grid direction, and retaining the nearest grid information of the coordinate information along the sparse grid direction, including: All grid information of the coordinate information is retained along the full grid direction; Determine whether there is an isolated grid in the sparse grid direction according to the coordinate information, if yes, remove the isolated grid and retain the nearest grid information of the coordinate information, if no, save the nearest grid information of the coordinate information; The grid information retained in the full grid direction and the sparse grid direction is combined to form edge points.

5. A sparse grid space scanning method according to claim 4, characterized in that: Judging whether there is an isolated grid in the sparse grid direction according to the coordinate information includes: Determine all grids in the sparse grid direction according to the coordinate information; It is determined whether the number of points included in any of the grids and a preset number of grids around it is less than a preset value. If so, the current grid is defined as an isolated grid.

6. A sparse grid space scanning method according to claim 5, characterized in that: Determining whether the number of points included in any of the grids and a preset number of grids around it is less than a preset value includes: Taking any of the grids as the center, obtain the number of points in all grids within a preset radius; Determine whether the number of points obtained is less than a preset value; Among them, all grids within the preset radius are complete grids.

7. A sparse grid space scanning method according to any one of claims 1 to 6, characterized in that: Obtaining coordinate information of the imaging point trace in the vehicle coordinate system includes: Obtain imaging points within the recognition range of the millimeter-wave radar; Obtain the coordinate information of the imaging point trace in the vehicle coordinate system.

8. A sparse grid space scanning method according to claim 7, characterized in that: Within the recognition range of the millimeter-wave radar, imaging points are obtained, including: Acquire static point traces within the recognition range of the millimeter wave radar, and perform multi-frame point trace accumulation; The imaging point trace is formed after compensating the DR of the vehicle.

9. A sparse grid space scanning method according to claim 8, characterized in that: The imaging points are distributed in a pre-formed absolute coordinate system; Obtaining coordinate information of the imaging point trace in the vehicle coordinate system includes: Determine whether the absolute coordinate system where the imaging point trace is located is the same coordinate system as the vehicle coordinate system. If so, directly obtain the coordinate information of the imaging point trace. If not, project the imaging point trace to the vehicle coordinate system and obtain the coordinate information of the projected imaging point trace.

10. A parking space recognition method, comprising the sparse grid space scanning method according to any one of claims 1 to 9, characterized in that: The following steps are involved: In the parking state, obtaining the initial edge points according to the sparse grid space scanning method, and fitting the initial edge points to form the parking space edge on the entry side; In the parking state, gradually acquiring secondary edge points according to the sparse grid space scanning method, and fitting the secondary edge points to form the parking space edge on the parking side; The parking space edge on the entry side and the parking space edge on the stop side are used as two end edges to form a stop area of ​​the current parking space.

Images