Calibration Method, Device, Equipment and Storage Medium for Radar Detection Area

Abstract

An embodiment of the present invention provides a method, device, equipment and storage medium for calibrating a radar detection area. The method includes: obtaining multiple frames of first point clouds collected by the radar; obtaining a motion area grid map corresponding to the multiple frames of first point clouds; generating a corresponding detection area grid map according to the motion area grid map; and calibrating the radar detection area according to the detection area grid map. Since the detection area is represented in the form of a grid map, the use of road boundary lines to represent the detection area is avoided, and the calibration difficulty of the detection area is reduced. And the automatic calibration of the radar detection area can be completed without manual calibration. And it can quickly identify the point cloud data in the detection area, reduce the computational complexity, and further reduce the requirements for hardware resources.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of radar, and in particular, to a method, device, equipment and storage medium for calibrating a radar detection area. Background Art

[0002] With the rapid development of autonomous driving technology, due to problems such as limited sensing distance and occlusion blind spots in the single-vehicle sensing solution, the development of autonomous driving technology has gradually shifted from single-vehicle intelligence to roadside intelligence. The most core problem of roadside intelligence lies in the roadside sensing system. Currently, the relatively mature and complete solution for roadside sensing is radar roadside sensing. In the roadside radar sensing algorithm, in order to reduce the algorithm complexity, only the point cloud data within the detection area is processed. Therefore, it is necessary to calibrate the detection area of the roadside radar.

[0003] In the prior art, the method for calibrating the detection area of the roadside radar is mainly based on the road boundary line. Specifically, the road boundary line of the detection area is determined artificially by a straight line or a curve to calibrate the detection area of the roadside radar. In the prior art method for calibrating the detection area of the roadside radar, when identifying whether each frame of point cloud data obtained by the roadside radar is the point cloud data within the detection area, it is identified by comparing each frame of point cloud data with the road boundary line.

[0004] Since it is difficult to express the road boundary line by a straight line or a curve in some special-shaped roads, the prior art method for calibrating the detection area of the roadside radar increases the difficulty of calibrating the detection area. Moreover, due to the large amount of data in each frame of point cloud data, comparing each frame of point cloud data with the road boundary line to identify the point cloud data within the detection area in each frame of point cloud data results in a high computational complexity, and thus a high requirement for hardware resources. Summary of the Invention

[0005] The embodiments of the present invention provide a method, device, equipment and storage medium for calibrating a radar detection area, which solves the technical problems that the prior art method for calibrating the detection area of the roadside radar increases the difficulty of calibrating the detection area, and has a high computational complexity when identifying whether the point cloud data is within the detection area, and thus a high requirement for hardware resources.

[0006] In a first aspect, the embodiments of the present invention provide a method for calibrating a radar detection area, including:

[0007] Obtaining multiple frames of first point clouds collected by the radar; obtaining a motion area grid map corresponding to the multiple frames of first point clouds; generating a corresponding detection area grid map according to the motion area grid map; and calibrating the radar detection area according to the detection area grid map.

[0008] Further, for the method described above, obtaining the motion area raster map corresponding to the multiple frames of the first point cloud includes:

[0009] Determining the target motion points in the first point cloud according to the background frame and the first point cloud, where the background frame is multiple frames of point clouds obtained prior to the first point cloud; generating the motion area raster map according to the target motion points in the first point cloud.

[0010] Further, for the method described above, determining the target motion points in the first

[0011] point cloud according to the background frame and the first point cloud includes:

[0012] Comparing the distance values of the point cloud points of each frame of point cloud corresponding to multiple target positions in the background frame to obtain the maximum distance value; using the maximum distance value corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; determining the target motion points in the first point cloud by using the background frame matrix.

[0013] Further, for the method described above, determining the target motion points in the first point cloud according to the background frame and the distance data of the multiple frames of the first point cloud includes:

[0014] Grouping the multiple frames of point clouds of the background frame according to the acquisition order, where each group of point clouds includes N consecutive frames of point clouds; comparing the distance values of the point cloud points of each frame of point cloud corresponding to multiple target positions for each group to obtain the median of the maximum distances; using the median of the maximum distances corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; determining the target motion points in the first point cloud by using the background frame matrix.

[0015] Further, for the method described above, determining the target motion points in the first point cloud by using the background frame matrix includes:

[0016] Comparing the element value in the background frame matrix with the distance value of the first target point corresponding to the multiple frames of the first point cloud; calculating the difference between the element value in the background frame matrix and the distance value of the point cloud points corresponding to each of the first point clouds; if the difference is greater than a preset distance threshold, determining the corresponding point cloud point as a target motion point.

[0017] Further, for the method described above, the method further includes:

[0018] If the difference is greater than a preset update threshold, using the distance value of the corresponding point cloud point as the element value of the background frame matrix.

[0019] Further, for the method described above, generating the motion area raster map according to the target motion points in the first point cloud includes:

[0020] Initialize the motion area grid map in the motion area grid map coordinate system to obtain an initialized motion area grid map; convert the distance value of the target motion point into the corresponding target motion point position coordinates in the motion area grid map coordinate system; generate the motion area grid map according to the target motion point coordinates and the initialized motion area grid map.

[0021] Further, in the method as described above, obtaining the motion area grid map corresponding to the multiple frames of the first point cloud includes:

[0022] Process the multiple frames of the first point cloud using a deep learning model to obtain the motion area grid map corresponding to the multiple frames of the first point cloud.

[0023] Further, in the method as described above, generating the corresponding detection area grid map according to the motion area grid map includes:

[0024] Use a preset region growing algorithm model to determine the detection area and non-detection area in the motion area grid map; generate the detection area grid map according to the detection area and the non-detection area.

[0025] Further, in the method as described above, the method further includes:

[0026] Judge whether the detection area update condition is satisfied; if the detection area update condition is satisfied, update the detection area grid map.

[0027] In a second aspect, an embodiment of the present invention provides a calibration device for a radar detection area, including:

[0028] A point cloud acquisition module for acquiring multiple frames of the first point cloud collected by the radar; a grid map acquisition module for acquiring the motion area grid map corresponding to the multiple frames of the first point cloud; a grid map generation module for generating a corresponding detection area grid map according to the motion area grid map; and an area calibration module for calibrating the radar detection area according to the detection area grid map.

[0029] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0030] A memory, a processor, and a computer program;

[0031] Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method as described in any one of the first aspects.

[0032] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and the computer program is executed by a processor to implement the method as described in any one of the first aspects.

[0033] An embodiment of the present invention provides a method for calibrating a radar detection area, which includes obtaining multiple frames of first point clouds collected by a radar; obtaining a motion area grid map corresponding to the multiple frames of first point clouds; generating a corresponding detection area grid map according to the motion area grid map; and calibrating the radar detection area according to the detection area grid map. Since the detection area is represented in the form of a grid map, it avoids using road boundary lines to represent the detection area, reducing the difficulty of calibrating the detection area. And it can complete the automatic calibration of the radar detection area without manual calibration. Moreover, representing the detection area in the form of a grid map can quickly identify the point cloud data in the detection area. When identifying the point cloud data in the detection area, only the position coordinates of each frame of point cloud data in the coordinate system of the detection area grid map need to be determined, and then it can be indexed through the detection area grid map whether the position corresponding to each target point is within the detection area, thereby quickly identifying the point cloud data in the detection area, reducing the computational complexity, and further reducing the requirement for hardware resources.

[0034] It should be understood that the content described in the above-mentioned invention content part is not intended to limit the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0036] Figure 1 It is a flowchart of the method for calibrating the radar detection area provided in Embodiment 1 of the present invention;

[0037] Figure 2 It is a flowchart of the method for calibrating the radar detection area provided in Embodiment 2 of the present invention;

[0038] Figure 3 It is a flowchart of step 202 in the method for calibrating the radar detection area provided in Embodiment 2 of the present invention;

[0039] Figure 4 It is a flowchart of step 2021 in the method for calibrating the radar detection area provided in Embodiment 2 of the present invention;

[0040] Figure 5 It is a schematic diagram of the vehicle motion trajectory in the method for calibrating the radar detection area provided in Embodiment 2 of the present invention;

[0041] Figure 6Another flowchart of step 2021 in the calibration method of the radar detection area provided in the second embodiment of the present invention;

[0042] Figure 7 Schematic diagram of the target moving point cloud corresponding to the vehicle in the calibration method of the radar detection area provided in the second embodiment of the present invention;

[0043] Figure 8 Schematic diagram of the detection area and non-detection area in the calibration method of the radar detection area provided in the second embodiment of the present invention;

[0044] Figure 9 Flowchart of the target detection method provided in the third embodiment of the present invention;

[0045] Figure 10 Flowchart of the calibration method of the detection area of the multi-sensor system provided in the third embodiment of the present invention;

[0046] Figure 11 Schematic diagram of the structure of the radar detection area calibration device provided in the fourth embodiment of the present invention;

[0047] Figure 12 Schematic diagram of the structure of the radar detection area calibration device provided in the fifth embodiment of the present invention;

[0048] Figure 13 Schematic diagram of the structure of the target detection device provided in the sixth embodiment of the present invention;

[0049] Figure 14 Schematic diagram of the structure of the electronic device provided in the seventh embodiment of the present invention. Detailed implementation manners

[0050] The embodiments of the present invention will be described in more detail with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not used to limit the protection scope of the present invention.

[0051] In the description and claims of the embodiments of the present invention and the above-mentioned drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0052] The embodiments of the present invention will be specifically described below with reference to the drawings.

[0053] Embodiment 1

[0054] Figure 1 is a flowchart of a method for calibrating a radar detection area provided in Embodiment 1 of the present invention. As Figure 1 shown, the execution subject of this embodiment is a device for calibrating a radar detection area. The device for calibrating a radar detection area can be integrated in an electronic device. The electronic device can be a radar, a computer, a tablet computer, or other devices with independent computing and processing capabilities. Then, the method for calibrating a radar detection area provided in this embodiment includes the following steps.

[0055] Step 101, obtain multiple frames of first point clouds collected by the radar.

[0056] In this embodiment, the radar can be a radar installed on the roadside. The type of radar can be a lidar, a millimeter-wave radar, or other types of radars. Among them, the roadside radar can be installed beside the road and is used to detect the distance between moving targets within the detection range and the radar. After the radar is installed on the roadside, the detection area of the radar is calibrated.

[0057] In this embodiment, when calibrating the detection area of the radar, first, multiple frames of first point clouds collected by the radar are obtained. Multiple frames of first point clouds can be collected by the acquisition device of the radar.

[0058] As an alternative embodiment, in this embodiment, if the electronic device is not a radar, a communication connection is established between the electronic device and the radar in advance. After the acquisition device of the radar has collected multiple frames of first point clouds, the multiple frames of first point clouds collected by the acquisition device of the radar are obtained by communicating with the radar.

[0059] As another alternative embodiment, in this embodiment, if the electronic device is the radar itself, after the acquisition device of the radar acquires multiple frames of first point cloud data, the multiple frames of first point cloud data can be directly obtained from the acquisition device.

[0060] Among them, each frame of the first point cloud includes the distance value of the point cloud points. The distance value of each frame of the first point cloud can be represented by a matrix. For example, if the radar is an m-line radar and the horizontal resolution is h°, then the number of columns of the matrix of the distance data of each frame of the first point cloud is 360 / h columns, and the number of rows is m rows. The dimension of the matrix corresponding to the distance data of each frame of the first point cloud is m*(360 / h). Each element value in the matrix corresponding to the distance data of each frame of the first point cloud is the distance data corresponding to each first target point in that frame of the first point cloud. The distance data corresponding to each first target point can be expressed as Lij. Where i is 1, 2,..., m, m is the number of lines of the radar, and j = 1, 2,..., n, n is the number of first target points included in each line of the radar.

[0061] Step 102, obtain the motion area grid map corresponding to multiple frames of the first point cloud.

[0062] In this embodiment, first, the target moving points can be determined from multiple frames of the first point cloud, and then the motion area grid map is generated according to the target moving points.

[0063] Among them, the method for determining the target moving points from multiple frames of the first point cloud can be: obtain the background frame corresponding to the radar, and then determine the target moving points according to the background frame and the first point cloud. Or, the method for determining the target moving points from multiple frames of the first point cloud can also be: use a deep learning model to construct a moving target recognition model, and then input multiple frames of the first point cloud into the moving target recognition model to identify the target moving points of the multiple frames of the first point cloud and output the target moving points.

[0064] It can be understood that the method for determining the target moving points from multiple frames of the first point cloud can also be other methods, which are not limited in this embodiment.

[0065] Step 103, generate the corresponding detection area grid map according to the motion area grid map.

[0066] Specifically, in this embodiment, the motion area grid map can be input into a preset region growing algorithm model, and the detection area and non-detection area in the motion area grid map are determined according to the preset region growing strategy in the preset region growing algorithm model, so as to output the detection area grid map.

[0067] Among them, the detection area grid map includes a detection area and a non-detection area, and the values corresponding to the grids in the detection area and the non-detection area are different.

[0068] Step 104: Calibrate the radar detection area according to the detection area grid map.

[0069] Specifically, in this embodiment, after determining the detection area grid map, the detection area grid map is used to represent the radar detection area to achieve the calibration of the radar detection area.

[0070] It can be understood that when representing the detection area of the radar using the detection area grid map, when identifying the point cloud data within the detection area for each frame of the radar point cloud data, after determining the position coordinates of each frame of the point cloud data in the coordinate system of the detection area grid map, it is possible to index through the detection area grid map whether the position corresponding to each target point is within the detection area, thereby quickly identifying the point cloud data within the detection area.

[0071] It can be understood that the coordinate system of the detection area grid map can be the same as the coordinate system of the motion area grid map.

[0072] The method for calibrating the radar detection area provided in this embodiment includes obtaining multiple frames of first point clouds collected by the radar; obtaining the motion area grid maps corresponding to the multiple frames of first point clouds; generating the corresponding detection area grid maps according to the motion area grid maps; and calibrating the radar detection area according to the detection area grid maps. Since the detection area is represented in the form of a grid map, it avoids using the road boundary line to represent the detection area, reducing the difficulty of calibrating the detection area. And it can complete the automatic calibration of the radar detection area without manual calibration. And representing the detection area in the form of a grid map can, when identifying the point cloud data within the detection area, only determine the position coordinates of each frame of the point cloud data in the coordinate system of the detection area grid map, and then index through the detection area grid map whether the position corresponding to each target point is within the detection area, thereby quickly identifying the targets in the point cloud data within the detection area, reducing the computational complexity, and thus reducing the requirements for hardware resources.

[0073] Embodiment 2

[0074] Figure 2 The flowchart of the method for calibrating the radar detection area provided in the second embodiment of the present invention is as follows. Figure 2 As shown, the method for calibrating the radar detection area provided in this embodiment further refines steps 102 - 103 on the basis of the method for calibrating the radar detection area provided in the first embodiment of the present invention, and also includes the step of updating the detection area grid map. Then the method for calibrating the radar detection area provided in this embodiment includes the following steps.

[0075] Step 201: Obtain multiple frames of first point clouds collected by the radar.

[0076] Step 202: Obtain the motion area raster map corresponding to multiple frames of the first point cloud.

[0077] As an alternative implementation, in this embodiment, as Figure 3 shown, Step 202 includes the following steps:

[0078] Step 2021: Determine the target moving points in the first point cloud according to the background frame and the first point cloud.

[0079] Among them, the background frame is multiple frames of point clouds obtained prior to the first point cloud.

[0080] Step 2022: Generate the motion area raster map according to the target moving points in the first point cloud.

[0081] Optionally, in this embodiment, as Figure 4 shown, Step 2021 includes the following steps:

[0082] Step 2021a: Compare the distance values of the point cloud points corresponding to multiple target positions in the background frame to obtain the maximum distance value.

[0083] In this embodiment, the background frame includes multiple frames of point clouds, which are point clouds prior to the first point cloud. Each point cloud point in each of these multiple frames of point clouds includes a distance value. First, determine the target positions of the background frame from the point cloud points in each frame. For example, the point cloud points can be sampled at equal intervals, and the sampled point cloud points are determined as the target positions of the background frame, or each point cloud in the point cloud can be directly determined as the target position of the background frame. Then, compare the distance values of the point cloud points corresponding to each target position in the background frame to obtain the maximum distance value.

[0084] Among them, the maximum distance value represents the distance value corresponding to the static background within the radar detection range, such as the road surface, roadside railing, trees, buildings, etc.

[0085] Step 2021b: Use the maximum distance value corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix.

[0086] In this embodiment, a matrix can be created first, and a mapping relationship between the target positions and the elements of the newly created matrix is established. Then, the maximum distance value corresponding to each target position is placed at the element with the mapping relationship to obtain the element value of each element of the matrix, thereby generating the background frame matrix.

[0087] Step 2021c: Use the background frame matrix to determine the target moving points in the first point cloud.

[0088] Optionally, in this embodiment, Step 2021c can specifically include:

[0089] Compare the element values in the background frame matrix with the distance values of the first target points corresponding to those in the multi-frame first point cloud; calculate the difference between the element values in the background frame matrix and the distance values of the point cloud points corresponding to those in each first point cloud; if the difference is greater than a preset distance threshold, then determine the corresponding point cloud point as a target moving point.

[0090] Specifically, in this embodiment, the element value in the background frame matrix can be expressed as Bij, and the distance value of the point cloud point corresponding to those in the multi-frame first point cloud can be expressed as lij. Then, determine whether the difference between Bij and lij is greater than the preset distance threshold. If it is greater than the preset distance threshold, then determine the corresponding point cloud point as a target moving point.

[0091] Among them, the preset distance threshold can be 10 cm, 20 cm, etc. Preferably, in order to avoid interference caused by the shaking of trees, the preset distance threshold can be 30 cm.

[0092] As Figure 5 shown, in Figure 5 , the target moving points in the first point cloud are determined according to the background frame matrix and the multi-frame first point cloud distance values, including the moving points of the first vehicle and the second vehicle. The target moving points of the first vehicle can determine the movement trajectory of the first vehicle, and the target moving points of the second vehicle can determine the movement trajectory of the second vehicle.

[0093] Optionally, in this embodiment, in step 2021c, after calculating the difference between the element values in the background frame matrix and the distance values of the point cloud points corresponding to those in each first point cloud, it further includes:

[0094] If the difference is greater than a preset update threshold, then use the distance value of the corresponding point cloud point as the element value of the background frame matrix.

[0095] Specifically, in this embodiment, the preset update threshold is the threshold for updating the background frame matrix. If the difference between the element value in the background frame matrix and the distance value of the point cloud point corresponding to those in each first point cloud is greater than the preset update threshold, it means that the representation of the current background by the background frame matrix is not accurate enough, meeting the condition for updating the background frame matrix. Then, when updating the background frame matrix, use the distance value of the corresponding point cloud point to update the element value of the background frame matrix.

[0096] In this embodiment, calculate the difference between the element values in the background frame matrix and the distance values of the point cloud points corresponding to those in each first point cloud. If the difference is greater than the preset update threshold, then use the distance value of the corresponding point cloud point as the element value of the background frame matrix. It can update the background frame matrix when meeting the update condition of the background frame matrix, meeting the requirement of the accuracy of the background frame matrix.

[0097] Or optionally, in this embodiment, as Figure 6As shown in the figure, step 2021 includes the following steps:

[0098] Step 20211: Group the multi-frame point clouds of the background frame according to the acquisition order, and each group of point clouds includes N consecutive frames of point clouds.

[0099] Where N is a positive integer greater than 1.

[0100] In this embodiment, the multi-frame point clouds of the background frame are divided into multiple groups according to the acquisition order. Each group of multi-frame point clouds of the background frame includes N consecutive frames of point clouds, and each frame of point cloud includes the distance value of the point cloud points. Then, in each group of point clouds, the distance values of the point cloud points of N consecutive frames can be expressed as Lijk, where k = 1, 2,..., N.

[0101] Step 20212: Compare the distance values of the point cloud points of each frame corresponding to multiple target positions in groups to obtain the maximum distance median.

[0102] Step 20213: Use the maximum distance median corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix.

[0103] Specifically, each element value Bij in the background frame matrix can be initially set to 0. First, determine the target positions of the background frame from the point cloud points in each group of multi-frame point clouds of the background frame. Sort the distance values of the point cloud points corresponding to the target positions in each group of multi-frame point clouds of the background frame from large to small, and determine the distance median of each corresponding point cloud point as Cij. If the distance median Cij of a certain corresponding point cloud point in the current group is greater than the corresponding element value Bij in the background frame matrix, then update the element value Bij in the background frame matrix to Cij. For the distance values of the point cloud points corresponding to the next group of point clouds, if the distance median Cij of a certain corresponding point cloud point in the next group is greater than the corresponding element value Bij in the background frame matrix, then still update the element value Bij in the background frame matrix to Cij. Therefore, the element value in the background frame matrix is the maximum distance median among the distance values of the point cloud points corresponding to all groups of point clouds.

[0104] In this embodiment, when determining the target moving points in the first point cloud based on the background frame and the multi-frame first point cloud distance data, two different methods can be used to determine the target moving points in the first point cloud, which improves the flexibility of determining the target moving points in the first point cloud.

[0105] Step 20214: Use the background frame matrix to determine the target moving points in the first point cloud.

[0106] In this embodiment, the implementation manner of step 20214 is similar to that of inventive step 2021c, and will not be elaborated here one by one.

[0107] Alternatively, as another optional implementation, in this embodiment, step 202 includes the following steps:

[0108] Process multiple frames of the first point cloud using a deep learning model to obtain a motion area grid map corresponding to the multiple frames of the first point cloud.

[0109] Specifically, in this embodiment, the deep learning model is a motion target recognition model.

[0110] First, use training samples to train the motion target recognition model. The training samples can be multiple frames of point cloud data with labeled motion points. After training the motion target recognition model with the training samples until convergence, a motion target recognition model trained to convergence is obtained. Then, input the multiple frames of the first point cloud data into the motion target recognition model trained to convergence, identify the target motion points in the multiple frames of the first point cloud data through the motion target recognition model, and output the target motion points. Finally, generate a motion area grid map based on the target motion points in the first point cloud.

[0111] In this embodiment, when obtaining the motion area grid map corresponding to the multiple frames of the first point cloud, the target motion points in the first point cloud can be determined by comparing the distance values between the background frame and the corresponding point cloud points in the first point cloud, or the target motion points in the first point cloud can be identified using a deep learning model. Then, a motion area grid map is generated based on the target motion points, enabling multiple methods to determine the target motion points in the first point cloud and improving the flexibility of determining the target motion points in the first point cloud.

[0112] As an optional implementation, in this embodiment, as Figure 7 shown, step 2022 of generating a motion area grid map based on the target motion points in the first point cloud specifically includes the following steps:

[0113] Step 2022a, initialize the motion area grid map in the coordinate system of the motion area grid map to obtain an initialized motion area grid map.

[0114] Further, in this embodiment, the size of each grid in the motion area grid map can be preset. The size of the grid can be determined according to the horizontal resolution of the lidar, so that each target position can be mapped to the grid in the motion area grid map after coordinate transformation. For example, the size of each grid in the motion area grid map is K meters. Then, determine the detection range of the lidar, and establish a coordinate system for the motion area grid map according to the detection range of the lidar and the size of each grid. For example, take the point at the upper left corner of the motion area grid map as the coordinate origin of the grid map coordinate system to establish the coordinate system for the motion area grid map. In the established coordinate system of the motion area grid map, if the detection range of the lidar is S meters and the size of each grid in the motion area grid map is K meters, then the length of the motion area grid map can be 2*S / K grid lengths, and the width can be 2*S / K grid widths.

[0115] For example, if the detection range of the lidar is 100 meters and the size of each grid is 0.1 meters, then in the coordinate system of the motion area grid map, the position of the lidar is (1000, 1000), the length of the motion area grid map is 2000 grid lengths, and the width is also 2000 grid widths.

[0116] After establishing the coordinate system for the motion area grid map, initialize the motion area grid map. Specifically, initialize the value of each grid in the motion area grid map. In this embodiment, the value of each grid is initialized to 0.

[0117] Step 2022b, convert the distance value of the target motion point into the position coordinate of the corresponding target motion point in the coordinate system of the motion area grid map.

[0118] Further, in this embodiment, first convert the distance value lij of the target motion point from the polar coordinate system to the position coordinate (Pijx, Pijy) in the rectangular coordinate system, and then convert the position coordinate in the rectangular coordinate system into the position coordinate of the corresponding target motion point in the coordinate system of the motion area grid map. Then, the position coordinate (Pijx, Pijy) in the rectangular coordinate system is converted into the position coordinate of the corresponding target motion point in the coordinate system of the motion area grid map as (Pijx / K + S / K, Pijy / K + S / K).

[0119] For example, if the distance data lij of the target motion point is converted from the polar coordinate system to the position coordinate in the rectangular coordinate system as (50.2, 40.3), K = 0.1, and S = 100, then the position coordinate (50.2, 40.3) in the rectangular coordinate system is converted into the position coordinate of the corresponding target motion point in the coordinate system of the motion area grid map as (1502, 1403).

[0120] Step 2022c, generate a motion area grid map according to the target motion point coordinates and the initialized motion area grid map.

[0121] Further, in this embodiment, first, the initial value at the position coordinates of the target moving point in the initialized motion area grid map is changed to 1, and the initial values at the position coordinates of the non-target moving points in the initialized motion area grid map are kept unchanged and still 0. Then, the dimension of the generated motion area grid map is (2*S / K, 2*S / K). The value at the position coordinates of the target moving point is 1, and the value at the position of the non-target moving point is 0. As Figure 7 shown, it is Figure 5 the target moving point cloud corresponding to the first vehicle trajectory shown and the target moving point cloud corresponding to the second vehicle. The values at the positions of the target moving point cloud corresponding to the first vehicle trajectory and the target moving point cloud corresponding to the second vehicle are 1.

[0122] Step 203, use a preset region growing algorithm model to determine the detection region and the non-detection region in the motion area grid map.

[0123] Step 204, generate a detection region grid map according to the detection region and the non-detection region.

[0124] Further, in this embodiment, the motion area grid map is input into the preset region growing algorithm model, and the preset region growing algorithm model performs the segmentation of the detection region and the non-detection region. Specifically, if the area of the motion area is less than the preset area threshold, it may be an individual interference region, then determine that the motion area is a non-detection region. If the area of the motion area is greater than the preset area threshold, then determine that the motion area is a detection region. After the preset region growing algorithm model completes the segmentation of the detection region and the non-detection region, set the value corresponding to each grid in the detection region to 1, and set the value of the non-detection region to 0. As Figure 8 shown, the detection region is Figure 8 the black-filled area in Figure 8 and the non-detection region is

[0125] the white-filled area in. And merge the detection region and the non-detection region to generate a detection region grid map, and output the detection region grid map. Store the output detection region grid map in the lidar to realize the calibration of the detection region of the lidar.

[0126] Step 205, determine whether the detection region update condition is satisfied.

[0127] Further, in this embodiment, after the lidar is installed and the detection area is calibrated, the installation position may shift due to external factors in the subsequent process. Therefore, in order to ensure that the detection area of the lidar is accurate, it is necessary to update the detection area of the lidar. After the lidar is installed and the first detection area is calibrated, it is determined whether the detection area update condition is met. If the detection area update condition is met, the detection area is updated.

[0128] Specifically, in this embodiment, the detection area can be updated periodically. Therefore, determining whether the detection area update condition is met can be determining whether the detection area update period has been reached. If the detection area update period has been reached, it is determined that the detection area update condition is met.

[0129] Step 206, if the detection area update condition is met, update the detection area grid map.

[0130] In this embodiment, the method for updating the detection area grid map is similar to the methods in steps 201 - 205, and will not be elaborated here one by one.

[0131] The method for calibrating the detection area of the lidar provided in this embodiment, after generating the corresponding detection area grid map according to the motion area grid map to calibrate the detection area of the lidar, determines whether the detection area update condition is met; if the detection area update condition is met, the detection area grid map is updated, which can not only realize the automatic calibration of the detection area of the lidar, but also automatically update the detection area after determining that the detection area update condition is met, reducing the maintenance cost of the lidar.

[0132] Embodiment III

[0133] Figure 9 is the flowchart of the target detection method provided in Embodiment III of the present invention. As Figure 9 shown, the target detection method provided in this embodiment is based on the method for calibrating the radar detection area provided in Embodiment II of the present invention. After step 204 or step 206, it further includes the steps of obtaining the radar detection area calibrated in multiple frames of the first point cloud, and performing target detection based on the point cloud in the radar detection area to obtain the detection result. Then, the method for calibrating the radar detection area provided in this embodiment includes the following steps.

[0134] Step 301, obtain the current frame of point cloud collected by the radar.

[0135] In this embodiment, the current frame of point cloud includes the distance values of the point cloud points. Among them, after the current frame of point cloud calibrates the detection area of the radar, the radar is officially put into use to obtain the point cloud when monitoring moving targets within the detection range.

[0136] It is understandable that, similar to step 101, if the electronic device is not a radar, a communication connection is established between the electronic device and the radar in advance. After the acquisition device of the radar acquires the current frame of point cloud, the current frame of point cloud acquired by the acquisition device of the radar is obtained through communication with the radar. If the electronic device is the radar itself, after the acquisition device of the radar acquires the current frame of point cloud, the current frame of point cloud can be directly obtained from the acquisition device.

[0137] Step 302: Obtain the calibrated radar detection area in multiple frames of first point clouds.

[0138] In this embodiment, obtaining the calibrated radar detection area in multiple frames of first point clouds means obtaining the grid map of the calibrated radar detection area. If the currently stored is the detection area grid map after the first calibration, then the current detection area grid map is the detection area grid map after the first calibration. If the currently stored is the updated detection area grid map, then the current detection area grid map is the updated detection area grid map.

[0139] It is understandable that the detection area grid map after the first calibration obtained in step 302 is the determined detection area grid map after steps 201 - 204. Or the updated detection area grid map obtained in step 302 is the updated detection area grid map determined after steps 205 - 206.

[0140] Step 303: Perform target detection on the point cloud in the radar detection area to obtain a detection result.

[0141] Furthermore, in this embodiment, the distance value of the current frame of point cloud is converted into a position coordinate in the coordinate system of the detection area grid map, and then the position coordinate of the distance value of the current frame of point cloud in the coordinate system of the detection area grid map is subjected to an AND operation with the position coordinate with a value of 1 in the currently stored current detection area grid map. Thus, it can be indexed through the current detection area grid map whether the position corresponding to each point cloud point is a position within the detection area, and then rapid target detection is performed on the point cloud within the detection area to obtain a detection result.

[0142] For the calibration method of the radar detection area provided in this embodiment, after generating the corresponding detection area grid map according to the motion area grid map or after updating the detection area grid map, it further includes: obtaining the radar detection area calibrated in multiple frames of the first point cloud, and performing target detection on the point cloud in the radar detection area to obtain a detection result. Since the detection area is represented in the form of a grid map, it avoids using the road boundary line to represent the detection area, reducing the calibration difficulty of the detection area. And representing the detection area in the form of a grid map can, when identifying the point cloud data in the detection area, only determine the position coordinates of each frame of point cloud in the coordinate system of the detection area grid map, and then index through the detection area grid map whether the position corresponding to each point cloud point is within the detection area, so as to quickly identify the point cloud in the detection area, reducing the computational complexity and thus reducing the requirement for hardware resources.

[0143] Based on the same inventive concept, in one embodiment, as Figure 10 shown, the present invention also proposes a calibration method for the detection area of a multi-sensor system. The multi-sensor system includes multiple sensors, and the method includes the following steps:

[0144] Step 401, obtaining the perception information collected by each sensor at the same moment and in the same scene, and performing fusion processing on the perception information collected by each sensor to obtain a fused image.

[0145] In this embodiment, after obtaining the perception information collected by each sensor, the perception information collected by each sensor at the same moment and in the same scene can be first subjected to spatio-temporal synchronization processing according to the system calibration parameters of the multi-sensor system, and then a fusion operation is performed to obtain a fused image. This fusion method can be result-level fusion (decision-level fusion) or feature-level fusion, which will not be elaborated here. Optionally, the multiple sensors in this step can be one or several of a camera, a lidar, and a millimeter-wave radar.

[0146] Step 402, obtaining the motion area grid map of the multiple frames of fused images.

[0147] Step 403, generating a corresponding detection area grid map according to the motion area grid map.

[0148] Step 404, calibrating the detection area of the multi-sensor system according to the detection area grid map.

[0149] The implementation manners of steps S402 - S404 in this embodiment are basically the same as the manners of obtaining the motion area grid map in Embodiment 1, Embodiment 2, and Embodiment 3. For specific details, reference can be made to the relevant content above, which will not be elaborated here.

[0150] The detection area calibration method of the multi-sensor system provided in this embodiment obtains the motion area grid map in the fused image; generates the corresponding detection area grid map according to the motion area grid map; and calibrates the detection area of the multi-sensor system according to the detection area grid map. Since the perception information of multiple sensors is fused first, this embodiment can achieve the calibration of the detection area in the images collected by multiple sensors simultaneously, which not only reduces the calibration difficulty of the detection area, but also improves the accuracy and efficiency of calibration. Since this embodiment can accurately and quickly calibrate the detection area, the calibration method of this embodiment can not only more accurately identify the target (using fused data with more feature dimensions), but also more quickly identify the target in the detection area (high detection area calibration efficiency).

[0151] Embodiment 4

[0152] Figure 11 is a schematic structural diagram of the calibration device for the radar detection area provided in Embodiment 4 of the present invention, as Figure 11 shown, the calibration device for the radar detection area provided in this embodiment includes: a point cloud acquisition module 41, a grid map acquisition module 42, a grid map generation module 43, and an area calibration module 44.

[0153] Among them, the point cloud acquisition module 41 is used to acquire multiple frames of first point clouds collected by the radar. The grid map acquisition module 42 is used to acquire the motion area grid map corresponding to the multiple frames of first point clouds. The grid map generation module 43 is used to generate the corresponding detection area grid map according to the motion area grid map. The area calibration module 44 is used to calibrate the radar detection area according to the detection area grid map.

[0154] The calibration device for the radar detection area provided in this embodiment can execute Figure 1 the technical solutions of the method embodiment shown, and its implementation principle and technical effects are similar, so they will not be elaborated here.

[0155] Embodiment 5

[0156] Figure 12 is a schematic structural diagram of the calibration device for the radar detection area provided in Embodiment 5 of the present invention, as Figure 12 shown, on the basis of the calibration device for the radar detection area provided in Embodiment 4, the calibration device for the radar detection area provided in this embodiment further includes: a grid map update module 51.

[0157] Optionally, the grid map acquisition module 42 is specifically used to: determine the target moving points in the first point cloud according to the background frame and the first point cloud, where the background frame is multiple frames of point clouds acquired prior to the first point cloud; generate the motion area grid map according to the target moving points in the first point cloud.

[0158] Optionally, when the grid map acquisition module 42 determines the target moving points in the first point cloud based on the background frame and the first point cloud, it specifically is used for:

[0159] Comparing the distance values of the point cloud points of each frame of point cloud corresponding to multiple target positions in the background frame to obtain the maximum distance value; using the maximum distance value corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; and determining the target moving points in the first point cloud by using the background frame matrix.

[0160] Optionally, when the grid map acquisition module 42 determines the target moving points in the first point cloud based on the background frame and the multi-frame first point cloud distance data, it specifically is used for:

[0161] Grouping the multi-frame point clouds of the background frame according to the acquisition order, where each group of point clouds includes N consecutive frames of point clouds; comparing the distance values of the point cloud points of each frame of point cloud corresponding to multiple target positions in each group to obtain the median of the maximum distances; using the median of the maximum distances corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; and determining the target moving points in the first point cloud by using the background frame matrix.

[0162] Optionally, when the grid map acquisition module 42 determines the target moving points in the first point cloud by using the background frame matrix, it specifically is used for:

[0163] Comparing the element value in the background frame matrix with the distance value of the corresponding first target point in the multi-frame first point clouds; calculating the difference between the element value in the background frame matrix and the distance value of the corresponding point cloud point in each first point cloud; and if the difference is greater than a preset distance threshold, determining the corresponding point cloud point as the target moving point.

[0164] Optionally, the grid map acquisition module 42 is further used for:

[0165] If the difference is greater than a preset update threshold, using the distance value of the corresponding point cloud point as the element value of the background frame matrix.

[0166] Optionally, when the grid map acquisition module 42 generates a motion area grid map based on the target moving points in the first point cloud, it specifically is used for:

[0167] Initializing the motion area grid map in the motion area grid map coordinate system to obtain the initialized motion area grid map; converting the distance value of the target moving point into the corresponding target moving point position coordinate in the motion area grid map coordinate system; and generating the motion area grid map according to the target moving point coordinates and the initialized motion area grid map.

[0168] Optionally, the grid map acquisition module 42 specifically is used for:

[0169] Processing multiple frames of first point clouds using a deep learning model to obtain a motion area grid map corresponding to the multiple frames of first point clouds.

[0170] The grid map generation module 43 is specifically configured to:

[0171] Using a preset region growing algorithm model to determine the detection region and non-detection region in the motion area grid map; generating a detection region grid map according to the detection region and non-detection region.

[0172] Optionally, the grid map update module 51 is used to:

[0173] Determine whether the detection region update condition is satisfied; if the detection region update condition is satisfied, update the detection region grid map.

[0174] The calibration device for the radar detection region provided in this embodiment can execute Figures 2 - 8 The technical solutions of the method embodiments shown, and their implementation principles and technical effects are similar, so they will not be elaborated here.

[0175] Embodiment Six

[0176] Figure 13 The structural schematic diagram of the target detection device provided in Embodiment Five of the present invention is as Figure 13 shown. Based on the calibration device for the radar detection region provided in Embodiment Three or Embodiment Four of the present invention, further, it further includes: a region acquisition module 61, a target detection module 62.

[0177] Among them, the region acquisition module 61 is used to acquire the calibrated radar detection region in multiple frames of first point clouds. The target detection module 62 is used to perform target detection according to the point clouds in the radar detection region to obtain a detection result.

[0178] The calibration device for the radar detection region provided in this embodiment can execute Figure 9 The technical solutions of the method embodiments shown, and their implementation principles and technical effects are similar, so they will not be elaborated here.

[0179] Embodiment Seven

[0180] Embodiment Seven of the present invention provides an electronic device, as Figure 14 shown. The electronic device includes: a memory 71, a processor 72, and a computer program.

[0181] Among them, the computer program is stored in the memory 71 and is configured to be executed by the processor 72 to implement the calibration method for the radar detection region provided in Embodiment One or Embodiment Two of the present invention. Or it is configured to be executed by the processor 72 to implement the target detection method provided in Embodiment Three of the present invention.

[0182] For relevant descriptions, reference can be made to Figures 1 - 9 the relevant descriptions and effects corresponding to the steps, and no further elaboration will be provided here.

[0183] Among them, in this embodiment, the memory 71 and the processor 72 are connected through the bus 73.

[0184] Embodiment Eight

[0185] Embodiment Seven of the present invention provides a computer-readable storage medium, on which a computer program is stored. The computer program is executed by a processor to implement the calibration method of the radar detection area provided in Embodiment One or Embodiment Two of the present invention. Or the computer program is executed by a processor to implement the target detection method provided in Embodiment Three of the present invention.

[0186] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or modules can be in electrical, mechanical or other forms.

[0187] The modules described as separate components may or may not be physically separated. The components displayed as modules may or may not be physical modules, that is, they can be located in one place, or they can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0188] In addition, in each embodiment of the present invention, the functional modules can be integrated in a processing module, or each module can exist physically alone, or two or more modules can be integrated in one module. The above-mentioned integrated modules can be implemented in the form of hardware, or in the form of hardware plus software functional modules.

[0189] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to the processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing devices, so that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program codes can be executed entirely on the machine, partially on the machine, executed partially on the machine as an independent software package and partially on a remote machine, or executed entirely on a remote machine or server.

[0190] In the context of the present invention, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0191] Moreover, although the operations are depicted in a particular order, this should be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed to achieve the desired result. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, although several specific implementation details are included in the above discussion, these should not be construed as limitations on the scope of the present disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single implementation. Conversely, various features that are described in the context of a single implementation can also be implemented separately or in any suitable sub-combination in multiple implementations.

[0192] Although the subject matter has been described in language specific to structural features and / or methodological logical acts, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims.

Claims

1. A calibration method for a radar detection area, characterized in that including: obtaining multiple frames of first point clouds collected by the radar; obtaining a motion area grid map corresponding to the multiple frames of first point clouds; generating a corresponding detection area grid map according to the motion area grid map; calibrating the radar detection area according to the detection area grid map to obtain the motion area grid map corresponding to the multiple frames of first point clouds, including: determining target motion points in the first point cloud according to a background frame and the first point cloud, where the background frame is multiple frames of point clouds obtained prior to the first point cloud; generating the motion area grid map according to the target motion points in the first point cloud; wherein, determining the target motion points in the first point cloud according to the background frame and the first point cloud includes: comparing the distance values of the point cloud points of each frame of point clouds corresponding to multiple target positions in the background frame to obtain the maximum distance value; using the maximum distance value corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; determining the target motion points in the first point cloud by using the background frame matrix.

2. The method according to claim 1, characterized in that, determining the target motion points in the first point cloud according to the background frame and the distance data of the multiple frames of first point clouds, including: grouping the multiple frames of point clouds of the background frame according to the acquisition order, and each group of point clouds includes consecutive N frames of point clouds; comparing the distance values of the point cloud points of each frame of point clouds corresponding to multiple target positions group by group to obtain the median of the maximum distances; using the median of the maximum distances corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; determining the target motion points in the first point cloud by using the background frame matrix.

3. The method according to claim 1, characterized in that, determining the target motion points in the first point cloud by using the background frame matrix includes: comparing the element value in the background frame matrix with the distance value of the first target point corresponding to the multiple frames of first point clouds; calculating the difference between the element value in the background frame matrix and the distance value of the corresponding point cloud point in each of the first point clouds; if the difference is greater than a preset distance threshold, determining the corresponding point cloud point as a target motion point.

4. The method according to claim 3, characterized in that, The method further includes: if the difference is greater than a preset update threshold, using the distance value of the corresponding point cloud point as the element value of the background frame matrix.

5. The method according to claim 4, characterized in that generating the motion area grid map according to the target motion points in the first point cloud, including: initializing the motion area grid map in the coordinate system of the motion area grid map to obtain an initialized motion area grid map; converting the distance value of the target motion point into the position coordinate of the corresponding target motion point in the coordinate system of the motion area grid map; generating the motion area grid map according to the target motion point coordinates and the initialized motion area grid map.

6. The method according to claim 1, wherein obtaining the motion area grid map corresponding to the multiple frames of first point clouds, including: processing the multiple frames of first point clouds by using a deep learning model to obtain the motion area grid map corresponding to the multiple frames of first point clouds.

7. The method according to claim 1, characterized in that, The generating the corresponding detection area grid map according to the motion area grid map includes: using a preset region growing algorithm model to determine the detection area and non-detection area in the motion area grid map; generating the detection area grid map according to the detection area and the non-detection area.

8. The method according to claim 1, wherein The method further includes: judging whether the detection area update condition is satisfied; If the detection area update condition is satisfied, the detection area grid map is updated.

9. A target detection method, characterized in that, The method includes: Using the calibration method of the radar detection area according to any one of claims 1-8, obtaining the calibrated radar detection area in multiple frames of the first point cloud; Performing target detection based on the point cloud in the radar detection area to obtain a detection result.

10. A calibration device for a radar detection area, characterized in that, Including: A point cloud acquisition module for acquiring multiple frames of the first point cloud collected by the radar; A grid map acquisition module for acquiring the motion area grid map corresponding to the multiple frames of the first point cloud; A grid map generation module for generating a corresponding detection area grid map according to the motion area grid map; An area calibration module for calibrating the radar detection area according to the detection area grid map; Wherein, obtaining the motion area grid map corresponding to the multiple frames of the first point cloud includes: Determining the target motion points in the first point cloud according to the background frame and the first point cloud, wherein the background frame is multiple frames of point clouds acquired prior to the first point cloud; Generating the motion area grid map according to the target motion points in the first point cloud; Wherein, determining the target motion points in the first point cloud according to the background frame and the first point cloud includes: Comparing the distance values of the point cloud points of each frame of point cloud corresponding to multiple target positions in the background frame to obtain the maximum distance value; Taking the maximum distance value corresponding to each target position as the element value of the background frame matrix to generate the background frame matrix; Using the background frame matrix to determine the target motion points in the first point cloud.

11. An electronic device, characterized in that, Including: A memory, a processor, and a computer program; Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is executed by the processor to implement the method according to any one of claims 1-9.

Images