Method, apparatus, radar sensor, road side unit for radar sensor detecting static objects of interest

Abstract

This application provides a method for detecting static objects of interest using a radar sensor. The method may include: scanning the surrounding environment with a radar sensor to obtain echo signals, the echo signals being radar sensing data; detecting dynamic objects in the surrounding environment based on the radar sensing data; and detecting static objects of interest in the surrounding environment using a pre-trained static object detection model based on the radar sensing data; wherein the static object detection model includes radar feature data of the static objects of interest. Corresponding devices, radar sensors, roadside units, software products, storage media, etc., are also provided.

Description

Technical Field

[0001] This application relates to radar-related technologies, and more specifically, to technologies for static object detection. Background Technology

[0002] With the in-depth development of vehicle-to-everything (V2X) technology, more and more roadside units are equipped with environmental perception and information exchange capabilities. Through various sensors deployed on the roadside, traffic conditions can be captured in real time and valuable information can be processed. At the same time, they can also provide vehicles with beyond-line-of-sight information, which greatly improves traffic safety.

[0003] However, while radar sensors can currently identify objects from background noise based on their movement characteristics, they overlook objects that are stationary relative to the road surface. Therefore, static infrastructure cannot be reliably distinguished from stationary traffic participants. For example, a radar sensor scans an object and obtains an echo signal. It can detect a moving vehicle based on the movement characteristics of the vehicle carried in the echo signal. However, if the vehicle is stationary, it is difficult to identify it based on the radar sensing data.

[0004] Therefore, it is necessary to make improvements in this area to enhance the radar sensor's ability to detect stationary objects. Summary of the Invention

[0005] According to one aspect of this application, a method for detecting a static object of interest using a radar sensor is provided. The method may include: scanning a surrounding environment by a radar sensor to obtain echo signals, the echo signals being radar sensing data; detecting dynamic objects in the surrounding environment based on the radar sensing data; and detecting the static object of interest in the surrounding environment based on the radar sensing data using a pre-trained static object detection model; wherein the static object detection model includes radar feature data of the static object of interest.

[0006] In the provided method for detecting static objects of interest using a radar sensor, for example, a pre-trained static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data, including: inputting the radar sensing data into the static object detection model; and determining that the static object of interest exists in the surrounding environment when the radar sensing data includes radar feature data of the static object of interest.

[0007] In the provided method for detecting static objects of interest using a radar sensor, for example, a pre-trained static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data, including: when a suspected static object is found during the detection of dynamic objects in the surrounding environment, inputting the radar sensing data of the suspected static object into the static object detection model; the object detection model identifies whether the suspected static object is a static object of interest based on the radar sensing data of the suspected static object.

[0008] In the provided method for detecting static objects of interest using a radar sensor, for example, when a suspected static object is found during the detection of dynamic objects in the surrounding environment, the radar sensing data of the suspected static object is input into the static object detection model, including: comparing a radar sensing image formed by the radar sensing data with a background noise image of the surrounding environment; when the radar sensing image includes an object not included in the background noise image and the object is not the identified dynamic object, the object not included in the background noise image is a suspected static object; and inputting the radar sensing data of the suspected static image into the static object of interest detection model.

[0009] In the provided method for detecting static objects of interest using radar sensors, the radar feature data of the static object of interest is, for example, obtained through the following process: acquiring an image of the surrounding environment using an image acquisition unit, and determining whether a static object of interest exists in the surrounding environment; if it is determined that no static object of interest exists in the surrounding environment, causing each of the radar sensors to scan within its field of view and collecting echo signals to form first radar data; if it is determined that a static object of interest exists in the surrounding environment, causing each of the radar sensors to scan within its field of view and collecting echo signals to form second radar data, the second radar data including data of the static object of interest; comparing the second radar data with the first radar data, and extracting the data of the static object of interest therefrom to form feature data of the object of interest.

[0010] In the provided method for detecting static objects of interest using a radar sensor, for example, comparing second radar data with first radar data to extract data of the static object of interest includes: forming a background noise map based on the radar data from the first radar data; forming a field condition map based on the radar data from the second radar data; and comparing the background noise map with the field condition map to obtain data of the static object of interest based on the difference between the two.

[0011] According to another aspect of this application, an apparatus for detecting static objects of interest using a radar sensor is also provided. The apparatus includes: a storage module for storing program instructions; and a processing module for receiving radar sensing data, which is sensing data acquired by the radar sensor scanning the surrounding environment. The processing module is configured with a static object detection model. When the processing module executes the program instructions, it can: detect dynamic objects in the surrounding environment based on the radar sensing data; and detect static objects of interest in the surrounding environment using the static object detection model based on the radar sensing data. The static object detection model is a pre-trained model that includes radar feature data of the static objects of interest.

[0012] The provided apparatus for detecting static objects of interest using a radar sensor may optionally include a processor detecting static objects of interest in the surrounding environment through the following process when executing the instructions: inputting the radar sensing data into the static object detection model; and determining that the static object of interest exists in the surrounding environment when the radar sensing data includes radar feature data of the static object of interest.

[0013] The provided apparatus for detecting static objects of interest using a radar sensor may optionally include a processor detecting static objects of interest in the surrounding environment through the following process when executing the instructions: when a suspected static object is found during the detection of dynamic objects in the surrounding environment, the radar sensing data of the suspected static object is input into the static object detection model; the object detection model identifies whether the suspected static object is a static object of interest based on the radar sensing data of the suspected static object.

[0014] The provided apparatus for detecting static objects of interest using a radar sensor may optionally include a processor determining the suspected static object by the following process when executing the instructions: comparing a radar sensing map formed from the radar sensing data with a background noise map of the surrounding environment; and determining the object not included in the background noise map as a suspected static object when the radar sensing map includes an object not included in the background noise map and the object is not the identified dynamic object.

[0015] The provided apparatus for detecting static objects of interest using a radar sensor may optionally include a radar feature data of the static object of interest determined based on the data difference between second radar data and first radar data. The first radar data is data formed by the radar sensor scanning objects in the surrounding environment when the static object of interest is not present in the surrounding environment; the second radar data is data formed by the radar sensor scanning objects in the surrounding environment when the static object of interest is present in the surrounding environment.

[0016] According to another aspect of this application, a radar sensor is also provided, which is configured to perform any of the methods described herein.

[0017] According to other aspects of this application, a roadside unit is also provided, which includes the radar sensor described herein, or the radar sensor is configured to perform any of the methods described herein.

[0018] A program product is also provided, which includes program instructions that, when executed, enable the execution of any of the methods described herein.

[0019] It also provides a non-temporary storage cutoff, on which instructions are stored, which, when executed, can perform any of the methods described herein.

[0020] According to the examples in this application, the object of interest is one or more of vehicles, pedestrians, etc.

[0021] Based on the examples of this application, in the case of identifying static objects of interest in the surrounding environment, the type of the static objects of interest can be further identified. Attached Figure Description

[0022] This application will be more fully understood by referring to the following detailed description of specific embodiments in conjunction with the accompanying drawings, wherein the same reference numerals in the drawings refer to the same elements, wherein:

[0023] Figure 1 These are schematic illustrations of roadside units according to some embodiments of this application;

[0024] Figure 2 This is a flowchart of a method for detecting static objects of interest using a radar sensor according to some embodiments of this application;

[0025] Figure 3 This illustration shows the process of a radar sensor detecting a static object of interest according to some examples of this application;

[0026] Figure 4 This illustrates the process of a radar sensor detecting a static object of interest according to other examples of this application;

[0027] Figure 5 The flowchart illustrates the method for obtaining an object detection model from a roadside unit;

[0028] Figure 6 This is a schematic diagram of a device used by radar sensors to detect static objects of interest.

[0029] Figure 7This simplified illustration illustrates the use of radar sensors to sense dynamic and static objects of interest on a road in the prior art.

[0030] Figure 8 The process of detecting a static object of interest based on radar sensor sensing data in each example of this application is simplified and illustrated in the roadside unit according to this application, or a roadside unit including a radar sensor according to this application, or performing a method according to this application. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings. It should be noted that the described implementation methods are only a part of the implementation methods of the technical solutions of this application, and not all of them. All other implementation methods obtained by those skilled in the art based on the implementation methods described in this application document without creative effort are covered by the protection scope of this application.

[0032] Roadside units (RSUs) are located on the roadside and can communicate with onboard units (OBUs) to perform functions such as vehicle identification. RSUs can be installed on streetlights, traffic signs, or other facilities located on or in the middle of the road that do not impede vehicle movement.

[0033] Figure 1 This is a schematic illustration of a roadside unit according to some embodiments of this application. In this example, the roadside unit 1 includes a radar sensor 10 and an image acquisition component 12. Optionally, the roadside unit may also include a processor 14. In the case where the roadside unit does not include a processor, the data acquired by the radar sensor and the image acquisition component can be transmitted to the cloud for processing. Figure 1 The illustration shows a radar sensor 10 and an image acquisition unit 12, but the roadside unit 1 may include multiple radar sensors with different fields of view and scan rates, and similarly, may include one or more image acquisition units. The radar sensor is, for example, a millimeter-wave radar sensor. The image acquisition unit is, for example, a camera.

[0034] The processor 14 is configured to control the scanning operation of the radar sensor 10. For example, it controls the frequency and duration of the radar sensor 10's scanning operation and processes the received radar echoes. In this application, radar echoes are also referred to as radar sensing data, radar sensing point data, etc. The processor 14 is also configured to control the image acquisition operation of the image acquisition unit and process the image data. In this application, the processor 14 is also at least configured to control the operation of the radar sensor 10 based on the data acquired by the image acquisition unit.

[0035] Figure 2 This is a flowchart of a method for detecting a static object of interest using a radar sensor, according to some embodiments of this application. The radar sensor is, for example, a radar sensor used by a roadside unit, which is arranged on a facility on the side of the road to scan images within its field of view to obtain radar sensing data. Accordingly, the method in this example can be... Figure 1 The roadside unit shown is used to perform this.

[0036] In step S200, the radar sensor scans the surrounding environment to obtain echo signals, i.e., acquires radar sensing data. The surrounding environment here refers to the scene being detected, such as the detectable area of ​​a roadside unit, like an intersection or road segment. A radar point cloud map can be formed based on the data sensed by the radar sensor. The radar sensing data may include radar cross-section (RCS), received signal strength (RSS1), emission angle, and range parameter data. In step S202, dynamic objects in the surrounding environment are detected based on the radar data sensed by the radar sensor. In step S204, a static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data. This static object detection model is a pre-trained model that includes radar feature data of the static objects of interest.

[0037] Figure 3 The illustration shows a process for a radar sensor to detect a static object of interest according to some examples of this application. In step S300a, radar sensing data 200 obtained in step S200 is received, and the presence of a dynamic object, i.e., a moving object such as a vehicle or pedestrian, is detected based on the radar sensing data. Furthermore, in step S300b, an object detection model detects the presence of a static object of interest based on the radar sensing data 200 obtained in step S200. For example, the static object detection model compares the received input radar sensing data with radar feature data of the static object of interest; if the received radar sensing data includes radar sensing data matching the feature data of the static object of interest, it indicates the presence of a static object of interest.

[0038] In step S302, the detection result is received, that is, the detection results generated in steps S300a and S300b are received. The detection result of step S300a may be that a dynamic object exists in the surrounding environment. If so, data related to the dynamic object, such as its distance or location, can be output. The identification result of step S300a may also be that no dynamic object is detected in the surrounding environment. The detection result of step S300b may be that no static object of interest exists in the surrounding environment. Alternatively, the detection result of step S300b may be that a static object of interest exists in the surrounding environment. In this case, information related to the static object of interest, such as its distance or location, can be output.

[0039] Figure 4 The illustration shows the process of a radar sensor detecting a static object of interest according to some other examples of this application. In step S400, radar sensing data 200 obtained in step S200 is received, and the presence of a dynamic object is detected based on the radar sensing data 200. If a suspected static object is found during the detection of dynamic objects in step S400, the process proceeds to step S402. In step S402, the radar sensing data of the suspected static object is input into a static object detection model, which identifies whether the radar sensing data of the suspected static object is a static object of interest based on the data. For example, if the static object detection model matches the received radar sensing data with the radar feature data of a static object of interest, it determines that the suspected static object is a static object of interest.

[0040] In step S404, the detection result is received, that is, the detection results generated in steps S400 and S402 are received. The detection result of step S400 may be that there is a dynamic object in the surrounding environment, or that no dynamic object is detected in the surrounding environment. The detection result of step S402 may be that there is no static object of interest in the surrounding environment; or, the detection result of step S402 may be that there is a static object of interest in the surrounding environment.

[0041] In a further example, step S400 further includes steps S4001 to S4003. In step S4001, the radar sensing map formed from the radar sensing data is compared with a background noise map of the surrounding environment. Here, the radar sensing map is, for example, a radar point cloud map formed from the radar sensing data; the background noise map of the surrounding environment is pre-formed for the surrounding environment and does not include static objects of interest. According to some examples of this application, the background noise map of the surrounding environment is formed when training a static object detection model, as will be described below. In step S4002, when the radar sensing map includes an object not included in the background noise map, and the object is not the identified dynamic object, the object not included in the background noise map can be identified as a suspected static object. In step S4003, the radar sensing data of the suspected static object is input into the static object of interest detection model for identification. It should be noted that steps S4001 to S4003 are merely an exemplary specific implementation of step S400, and not a limitation. Step S400 can also be implemented through other more specific processes. For example, the radar sensing data in step S4001 may be radar sensing data from which the identified dynamic objects have been removed, so that the resulting radar sensing map no longer includes the data of the identified dynamic objects; thus, in step S4002, if the radar sensing map includes objects not included in the background noise map, the object is regarded as a suspected static object.

[0042] according to Figure 2 , Figure 3 or Figure 4 Any example of the described method for detecting static objects of interest using a radar sensor can be obtained through Figure 1 The example is performed by a roadside unit. For instance, radar sensor 10 scans the surrounding environment to acquire radar sensing data (S200). Processor 14 detects the presence of dynamic objects in the surrounding environment based on the radar sensing data (S202), and can cause roadside unit 1 to output relevant signals, such as signals indicating the detection of a dynamic object and information such as the distance and position of the dynamic object, when a dynamic object is detected. Processor 14 detects the presence of static objects of interest in the surrounding environment based on the radar sensing data using a pre-trained object detection module configured in the processor (S204). When the object detection model detects the presence of static objects of interest in the surrounding environment based on the radar sensing data, it can be done in different ways. For example, when processor 14 detects dynamic objects from the received radar sensing data, it also detects whether there are static objects of interest within them using the object detection model, i.e., it performs a combination of... Figure 3The process described. Alternatively, when the processor 14 detects dynamic objects from the received radar sensing data, if it finds a suspected static object, it inputs the radar sensing data of the suspected static object into the object detection model, which then identifies whether the radar sensing data represents a static object of interest, i.e., performs a combination... Figure 4 The process described.

[0043] Based on the detection results, the processor 14 can output different signals. For example, in the case of detecting a dynamic object and / or a static object of interest, it outputs a signal indicating that the object has been detected and related to its distance and position. Here, the output may be transmitted to other components or output to a display screen, etc. Specifically, the information to be carried by the signal can be determined by the application scenario of the roadside unit 1. For example, when the roadside unit 1 communicates with a passing vehicle A, if it detects that a vehicle is located around vehicle A in the surrounding environment, it can transmit a signal to vehicle A indicating the presence of the static vehicle. The signal can also carry information that other roadside units need to transmit to vehicle A.

[0044] Based on some examples from this application, object detection models can, for example... Figure 1 The roadside units are used for training, and can be configured after training. Figure 1 The roadside unit enables the detection of static objects in the surrounding environment based on the echo signals from radar sensors. Figure 5 The flowchart illustrates the method for obtaining an object detection model using roadside units, where a roadside unit is, for example, a... Figure 1 The roadside unit shown is an example. It should be understood that the radar sensors and image acquisition components in roadside unit 1 are pre-calibrated, and roadside unit 1 can be equipped with multiple calibrated radar sensors or calibrated image acquisition components. Furthermore, in the example specifically described below, the image acquisition component is a camera; camera 12 will be directly used as the image acquisition component in the following description.

[0045] In step S500, the radar sensor 10 is controlled to collect field data when there is no static object of interest in the field, forming first radar data. For example, the processor 14 determines whether there is a static object of interest in the field based on the image data captured by the camera 12. If it is determined that there is no static object of interest in the field, the processor 14 controls the radar sensor 10 to scan the field objects to obtain radar echoes. Here, for each radar sensor, the processor 14 controls the radar sensor to scan when there is no static object of interest in its field of view. The field here refers to the scene range covered by the roadside unit 1, that is, the surrounding environment of the roadside unit 1. In order to be able to detect the field (i.e., the surrounding environment) which is usually larger than the field of view, the roadside unit 1 may include multiple radar sensors, which are arranged, for example, in an array.

[0046] The camera 12 continuously captures images of the scene, and the processor 14 accordingly continuously detects whether there is a static object of interest in the scene, and controls the radar sensor 10 to scan the scene to obtain echo signals when there is no static object of interest. In this way, for the scene, several image data when there is no static object of interest can be acquired to form first image data; several radar data when there is no static object of interest can be acquired to form first radar data.

[0047] In step S502, a background noise map based on the radar data is constructed using the first radar data obtained in step S500. Taking a millimeter-wave radar sensor as an example, the echo data acquired by the radar sensor 10 is analyzed, the distance to the sensed target is calculated, and the phase difference is calculated using the echo antenna of the radar sensor 10 to form the echo difference between the left and right sides, thereby calculating the angular resolution, etc., to form a distribution image. Thus, a background noise map of the surrounding environment is constructed using radar sensing data without static object of interest data. In the example of this application, the formed background noise map is a point cloud map of the radar background.

[0048] According to some examples in this application, the amount of radar sensing data collected in step S500 should be sufficient to establish at least a radar background noise map. Therefore, when the amount of data has reached the level required to establish a background noise map, data collection in step S500 can be stopped. It should be understood that the background noise map includes background maps of the surrounding environment of the roadside unit under multiple scenarios at different time periods. For example, the roadside unit set up at intersection A may establish background maps of the surrounding environment under multiple time periods, such as 5:00-6:00 AM, 6:00-7:00 AM, etc., with an hourly time window.

[0049] In step S504, the radar sensor 10 is controlled to collect field data when there is a static object of interest at the scene, forming second radar data. For example, the processor 14 determines whether there is a static object of interest at the scene based on the image data captured by the camera 12. If a static object of interest is determined to be present, the processor 14 controls the radar sensor 10 to scan the scene to obtain radar echoes. The camera 12 can continuously capture images of the scene, and the processor 14 can correspondingly continuously detect whether there is a static object of interest at the scene, and control the radar sensor 10 to scan the scene to obtain echoes when a static object of interest is present. In this way, for the scene, a large amount of image data when there is a static object of interest can be acquired to form second image data; a large amount of radar sensing data when there is a static object of interest can be acquired to form second radar data. In this application, a static object of interest refers to a stationary object of interest. For example, if the object of interest is a vehicle, then the static object of interest is a stationary vehicle, which may be in a parked state or a stopped state. The data collection in step S504 can continue until the processor 14 determines that the static object of interest has moved based on the image data captured by the camera 12.

[0050] In step S506, the second radar data collected in step S504 forms a field situation map based on the radar data, which includes the static object of interest. The field situation map is, for example, a radar point cloud map, which includes not only the background of the field but also the point cloud of the static object of interest.

[0051] In step S508, the field situation map is compared with the background noise map, and the data of the static object of interest is obtained based on the differences between the two. For example, the different point clouds of the field situation point cloud map and the background noise point cloud map are extracted, and the point feature values ​​are calculated according to the characteristics of the point clouds, thereby obtaining the radar data of the static object of interest, and thus determining the feature data of the static object of interest under the radar echo signal.

[0052] In combination Figure 5 In the description, based on the examples in this application, acquiring radar sensing data can also consider acquiring sensing data under different natural environmental factors, such as under different weather conditions. In summary, acquiring radar sensing data under different situations and scenarios should be fully considered, thereby constructing relatively complete and comprehensive background noise maps and site maps, and thus obtaining more complete and comprehensive static feature data of the object of interest.

[0053] In combination Figure 5 In the description process, steps S502 and S506 are optional. When steps S502 and S506 are omitted, in step S508, the first radar data and the second radar data are compared to obtain data on the static object of interest based on their differences, forming characteristic data of the object of interest. Furthermore, steps S500 and S504 are not sequential; step S500 can be executed when there is no static object of interest in the surrounding environment, and step S504 can be executed when there is. Therefore, the order of execution depends on the presence of a static object of interest on site. From a certain perspective, the execution of steps S500 and S504 can be interleaved.

[0054] By repeating steps S500 to S508, a large amount of background noise data can be obtained when there is no static object of interest; a large number of on-site data including the static object of interest can be obtained when there is a static object of interest; and feature data of the static object of interest can be obtained from the difference between the two. It can be understood that the feature data of the object of interest (also called the object of interest feature data) can include multiple sets of feature data. One set of feature data represents one object of interest. In the examples of this application, the object of interest can be a vehicle, a pedestrian, etc. Thus, the feature data for a static vehicle may be different from the feature data for a stationary pedestrian; furthermore, the feature data for different types of vehicles may also be different, and so on. Therefore, the feature data of a static object of interest can include multiple sets of feature data for different objects. In this way, when a static object of interest is detected, the specific type of the static object of interest can be further identified based on the feature data.

[0055] After the static object detection model is trained and configured for use, for example, in a processor, it can also be used according to some examples of this application, for example... Figure 5 The described process is used to acquire first radar data and second radar data, thereby obtaining feature data of the object of interest, and then further training the object detection model during use.

[0056] According to this application, the recognition results of the static object recognition model during use can be further fed back to the static object recognition model so that it can be corrected.

[0057] By executing the method for detecting static objects of interest using a radar sensor according to the example of this application, performed by a radar sensor or a roadside unit including a radar sensor, not only can dynamic objects such as moving vehicles be identified based on radar sensing data, but also stationary objects such as vehicles in the surrounding environment can be detected, making the object identification function based on radar sensor sensing data more comprehensive. Thus, for roadside units and other systems that include both dynamic and static object identification, it is possible to omit some or all of the image acquisition components, which helps to save hardware costs.

[0058] According to this application, an apparatus for detecting static objects of interest using a radar sensor is also provided. Figure 6 This is a schematic diagram of the device. (For example...) Figure 6As shown, the device includes a storage module 60 and a processing module 62. The storage module 60 is, for example, a memory that stores program instructions. The processing module 62 is configured to receive radar sensing data, which is sensing data acquired by the radar sensor scanning the surrounding environment. The processing module 62 is equipped with a static object detection model. When executing the program instructions, the processing module 62: detects dynamic objects in the surrounding environment based on the radar sensing data; and the static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data. The static object detection model is a pre-trained model that includes radar feature data of the static objects of interest. In a further example, the processing module 60, when executing the program instructions, can achieve the above-described combination... Figure 3 or Figure 4 The process described.

[0059] Processing module 62 can also be configured to perform the above combination Figure 5 The process of description is used to obtain radar feature data of static objects of interest.

[0060] The processing module 62 may be a module configured in the radar sensor; or, configured in a processor that communicates with the radar sensor; or, implemented as a separate processor that can communicate with the radar sensor to control the latter.

[0061] According to examples in this application, a radar sensor is also provided, which is configured to perform the above-described combination. Figures 2 to 4 Any of the methods described herein. As an example, the static object detection model configured in this radar sensor is pre-trained and includes, for example, methods via... Figure 5 The process shown obtains static feature data of the object of interest. In this example, the radar sensor includes, for example, a memory and a processor; the memory stores the static detection model and is used to implement the reference... Figures 2 to 4 The program instructions for any of the methods described in the document. In some examples, memory also stores instructions for implementing the method. Figure 5 The program instructions shown represent the process. When the processor executes these instructions, it enables the radar sensor to detect static objects of interest.

[0062] According to another example of this application, a roadside unit is also provided. This roadside unit includes a radar sensor, such as those described above, capable of performing the aforementioned actions. Figures 2 to 4 A radar sensor that can perform any of the methods described in [the document]. This radar sensor can also perform [the following actions]. Figure 5 As shown in the diagram. As an example, the roadside unit may also include an image acquisition device and / or a processor, such as... Figure 1 The roadside unit shown.

[0063] Figure 7This simplified illustration illustrates how radar sensors in the prior art detect dynamic and static objects of interest on a road. In this example, the object of interest is, for instance, a vehicle. Figure 7 As shown, there are two vehicles 701 and 702 traveling on road 70, and two stationary vehicles 711 and 712. The roadside unit's camera 90a captures images of the road 70 scene, obtaining image data including the road 70, the two traveling vehicles 701 and 702, and the stationary vehicles 711 and 712. The roadside unit's radar sensor 80a scans the road 70 scene, obtaining echo signals of the scanned objects, i.e., radar sensing data 800. For convenience, the figure presents a virtual image 801 constructed based on the radar sensing data of the radar sensor 80a, similar to an image 901 captured by the camera 90a. This image shows the two traveling vehicles 701” and 702”. However, in the prior art, radar sensing data ignores the static characteristics of vehicles, thus ignoring the stationary vehicles 711” and 712”. Vehicles 711” and 712” are indicated by dashed lines in the figure, representing the situation where stationary vehicles cannot be sensed under the prior art.

[0064] Figure 8 The simplified illustration illustrates the process of detecting a static object of interest based on radar sensor sensing data in each example of this application, including a roadside unit according to this application, or a roadside unit including a radar sensor according to this application, or a roadside unit performing the method according to this application. For example... Figure 8 As shown, there are two vehicles 701 and 702 traveling on road 70, and two stationary vehicles 711 and 712. Radar sensor 80b scans the scene on road 70 and obtains the echo signals of the scanned objects, i.e., radar sensing data 800. After identifying the traveling vehicles 701 and 702 based on the sensing data 800 of radar sensor 80b, the sensing data of the traveling vehicles 701 and 702 is removed from the sensing data of radar sensor 80b. The point cloud map 803 formed by the remaining sensing data 8001 is compared with the background noise map 101 of the scene, thereby determining the suspected static objects. The radar sensing data of the suspected static objects is sent to the static object of interest detection model, which compares these data with the feature data of the static object of interest to determine the presence of stationary vehicles 711 and 712. For ease of viewing, the figure presents a virtual image 801' constructed from radar sensing data of radar sensor 80a according to this process, where moving vehicles 701 and 702 are indicated as 701” and 702, and stationary vehicles 711 and 712 are indicated as stationary vehicles 711” and 712.

[0065] Therefore, based on radar sensor data, not only were dynamic objects identified, namely moving vehicles 701 and 702, but also stationary vehicles 711 and 712 were identified. For ease of comparison, Figure 8The image also illustrates a virtual image 803 that identifies both stationary and moving vehicles 711 and 712. In the virtual image 803, vehicles 711 and 712 are represented by solid lines to indicate that they are identified in the same way as moving vehicles 701 and 702.

[0066] In the various examples of methods in this application, the data used to indicate each step is not intended to indicate the order in which the steps are performed, but is only used to distinguish the steps.

[0067] The technical features in the various embodiments of this application can be combined with each other to form new implementation methods without departing from the spirit of this application and without conflicting with each other. Although specific embodiments of this application have been shown and described in detail to illustrate the principles of this application, it should be understood that this application can be implemented in other ways without departing from such principles.

Claims

1. A method for detecting static objects of interest using a radar sensor, characterized in that, The method includes: The surrounding environment is scanned by a radar sensor to obtain echo signals, which are radar sensing data. Detect dynamic objects in the surrounding environment based on the radar sensing data; A pre-trained static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data; The static object detection model includes radar feature data of the static object of interest.

2. The method according to claim 1, characterized in that, The static object detection model, pre-trained, detects static objects of interest in the surrounding environment based on the radar sensing data, including: The radar sensing data is input into the static object detection model; When the static object detection model includes radar feature data of the static object of interest in the radar sensing data, it determines that the static object of interest exists in the surrounding environment.

3. The method according to claim 1, characterized in that, The static object detection model, pre-trained, detects static objects of interest in the surrounding environment based on the radar sensing data, including: When a suspected static object is detected during the detection of dynamic objects in the surrounding environment, the radar sensing data of the suspected static object is input into the static object detection model. The object detection model identifies whether the suspected static object is a static object of interest based on the radar sensing data of the suspected static object.

4. The method according to claim 3, characterized in that, During the detection of dynamic objects in the surrounding environment, suspected static objects were identified, including: The radar sensing map formed from the radar sensing data is compared with the background noise map of the surrounding environment. If the radar sensing map includes objects not included in the background noise map, and the objects not included in the background noise map are not identified dynamic objects, then the objects not included in the background noise map are considered as suspected static objects.

5. The method according to any one of claims 1 to 4, characterized in that, The radar feature data of the static object of interest is obtained through the following process: The image acquisition component acquires images of the surrounding environment to determine whether there is a static object of interest in the surrounding environment. If it is determined that there is no static object of interest in the surrounding environment, each of the radar sensors scans within its field of view and collects echo signals to form first radar data. If it is determined that there is a static object of interest in the surrounding environment, each of the radar sensors scans within its field of view and collects echo signals to form second radar data, which includes data of the static object of interest. The second radar data is compared with the first radar data, and the data of the static object of interest is extracted from them to form the feature data of the object of interest.

6. The method according to claim 5, characterized in that, The second radar data is compared with the first radar data to extract the data of the static object of interest, including: A background noise map based on radar data is generated from the first radar data; A field situation map based on radar data is generated from the second radar data; The background noise map is compared with the scene condition map to obtain the data of the static object of interest based on the difference between the two.

7. The method according to any one of claims 1 to 6, characterized in that, The objects of interest are one or both of vehicles and pedestrians.

8. An apparatus for detecting static objects of interest using a radar sensor, characterized in that, The device includes: The storage module is used to store program instructions; The processing module is used to receive radar sensing data, which is the sensing data obtained by the radar sensor scanning the surrounding environment. The processing module is configured with a static object detection model. When the processing module executes the program instructions: Detect dynamic objects in the surrounding environment based on the radar sensing data; The static object detection model detects static objects of interest in the surrounding environment based on the radar sensing data; The static object detection model is a pre-trained model that includes radar feature data of the static object of interest.

9. The apparatus according to claim 8, characterized in that, When executing the instructions, the processor detects static objects of interest in the surrounding environment through the following process: The radar sensing data is input into the static object detection model; When the static object detection model includes radar feature data of the static object of interest in the radar sensing data, it determines that the static object of interest exists in the surrounding environment.

10. The apparatus according to claim 8, characterized in that, When executing the instructions, the processor detects static objects of interest in the surrounding environment through the following process: When a suspected static object is detected during the detection of dynamic objects in the surrounding environment, the radar sensing data of the suspected static object is input into the static object detection model. The object detection model identifies whether the suspected static object is a static object of interest based on the radar sensing data of the suspected static object.

11. The apparatus according to claim 10, characterized in that, The processor determines the suspected static object through the following process when executing the instruction: The radar sensing map formed from the radar sensing data is compared with the background noise map of the surrounding environment. When the radar sensing map includes an object not included in the background noise map and that object is not an identified dynamic object, the object not included in the background noise map is determined to be a suspected static object.

12. The apparatus according to any one of claims 8 to 11, characterized in that, When the processor executes the instruction: In the case of identifying static objects of interest in the surrounding environment, the type of the static objects of interest can be further identified.

13. The apparatus according to claim 8, characterized in that, The static radar feature data of the object of interest is determined based on the data difference between the second radar data and the first radar data, wherein, The first radar data is data formed by the echo signals obtained by the radar sensor scanning objects in the surrounding environment when the static object of interest is not present in the surrounding environment; The second radar data is data formed by the radar sensor scanning the objects in the surrounding environment when the static object of interest is present.

14. The apparatus according to claim 8, characterized in that, The static radar feature data of the object of interest is obtained by comparing the background noise map formed by the second radar data and the field situation map formed by the first radar data.

15. The apparatus according to any one of claims 8 to 14, characterized in that, The objects of interest are vehicles, pedestrians, or both.

16. A radar sensor, characterized in that, The radar sensor is configured to perform the method according to any one of claims 1 to 7.

17. A roadside unit, characterized in that, The roadside unit includes a radar sensor configured to perform the method according to any one of claims 1 to 7.

18. A program product comprising program instructions, characterized in that, The program instructions, when executed, can implement the method according to any one of claims 1 to 7.

19. A non-temporary storage medium storing program instructions thereon, characterized in that, The program instructions, when executed, can implement the method according to any one of claims 1 to 7.

Images