High-point cloud environment identification and detection method and system based on 4D millimeter wave radar

Abstract

The invention relates to the field of vehicle-mounted radars, in particular to a high-point cloud environment recognition and detection method and system based on a 4D millimeter-wave radar, and the method comprises the steps: obtaining dynamic and static point cloud information in a front region of a vehicle through a vehicle-mounted forward 4D millimeter-wave radar; according to the 4D millimeter wave radar installation information, converting the point cloud data from a radar coordinate system to an own vehicle coordinate system taking the geometric center of the vehicle as an original point; dividing the forward detection area into three transverse detection areas and five longitudinal sub-detection areas corresponding to each transverse detection area, counting the distribution condition and characteristics of high point clouds in each longitudinal sub-area, and performing single-frame high point cloud environment judgment; through continuous detection of a set sliding time window, the recognition condition of the high-point cloud environment state in the sliding window is counted, and stable high-point cloud environment recognition is realized. The problem that an existing 4D millimeter wave radar is not sensitive to environmental information perception is solved, and dynamic calibration and algorithm switching in different environments of the radar depending on the environmental information are facilitated.

Description

Technical Field

[0001] The present invention relates to the field of vehicle-mounted radars, and in particular to a high-point cloud environment recognition and detection method and system based on 4D millimeter-wave radar. Background Art

[0002] Environmental perception is one of the important functions in autonomous driving and automatic assisted navigation driving. For vehicles, running corresponding algorithms for different environmental conditions can achieve better intelligent driving results. For sensors, obtaining environmental conditions can more effectively eliminate noise information in the corresponding environment and ensure that the perceived information is clean and accurate.

[0003] 4D millimeter wave radar has become one of the indispensable on-board sensors in current autonomous driving and automatic assisted navigation driving due to its rich and accurate four-dimensional information perception and low cost. However, the current detection of high-point cloud environment still relies on on-board cameras or high-precision maps. High-point cloud environment is identified through video detection or high-precision map and fused with 4D millimeter wave radar data. This type of method has achieved good application from the perspective of vehicles, but the radar sensor itself does not obtain environmental information, which will affect the radar effect in some scenarios.

[0004] For example, when the vehicle is driving in a tunnel, crossing under a bridge, or in a high point cloud scene with tall trees planted side by side on both sides, the radar will detect discrete, high static point clouds in addition to obtaining normal point cloud information. This type of point cloud will reduce the dynamic calibration effect of the radar, and in the tunnel, multipath noise may be generated due to reflections on both sides, affecting the radar performance. If the radar itself can detect this type of environmental information, it can use more targeted dynamic calibration algorithms and perception algorithms for this type of environment to obtain better performance. Therefore, obtaining high point cloud environmental information is of great value to the performance of 4D millimeter-wave radars, and this type of information can be achieved by the 4D millimeter-wave radar itself. Summary of the invention

[0005] The main purpose of the present invention is to solve the technical problem that the existing radar in the prior art has a weak perception of environmental information. A high point cloud environment recognition and detection method based on 4D millimeter wave radar includes the following steps:

[0006] Use the vehicle-mounted forward 4D millimeter-wave radar to obtain the forward point cloud information of the radar coordinate system of the objects in the area in front of the vehicle;

[0007] According to the installation information of the 4D millimeter wave radar, the forward point cloud information of the radar coordinate system is converted into the point cloud information of the vehicle coordinate system;

[0008] The point cloud information of the vehicle coordinate system is divided into three lateral detection areas, and each lateral detection area is divided into five longitudinal sub-areas;

[0009] Count the distribution of high point clouds and point cloud features in each sub-area, perform single-frame high point cloud environment judgment, and obtain single-frame high point cloud environment judgment results;

[0010] A sliding time window is established, and continuous high point cloud environment recognition is performed based on all single-frame high point cloud environment judgment results within the time window to obtain the high point cloud environment recognition results.

[0011] The point cloud information includes point cloud information after a large number of noise points are removed from the field of view of the 4D millimeter-wave radar. The point cloud information is located in a traditional radar coordinate system and includes horizontal angle measurement, pitch angle measurement, Doppler velocity, and relative distance information of the point cloud relative to the 4D millimeter-wave radar.

[0012] The three lateral detection areas include:

[0013] The horizontal detection area has the same length in the positive direction of the X axis and is at the closest detection distance L min And the maximum detection distance L max The distances between them on the Y axis are equal, but their positions are different. The Y value ranges of the left, front, and right areas are [0,2d], [-d,d], [-2d,0], respectively, where 2d is the preset width of the horizontal detection area.

[0014] The five longitudinal sub-areas include:

[0015] The horizontal detection area is divided into five longitudinal sub-areas, and the lengths of the longitudinal sub-areas are L1, L2, L3, L4 and L5 respectively, and finally a total of 15 detection sub-areas with equal width and different lengths are obtained, and each horizontal area contains five longitudinal sub-areas.

[0016] Statistics are collected on the distribution of high point clouds and point cloud features in each sub-region, and single-frame high point cloud environment judgment is performed. The single-frame high point cloud environment judgment results include:

[0017] Statistically analyze the static point cloud heights in each longitudinal sub-area, and select the confident point clouds that meet the high point cloud environment characteristics according to the thresholds of the point cloud height and signal-to-noise ratio characteristics;

[0018] The number of confident point clouds in each longitudinal sub-region is counted. If the number of confident point clouds in each longitudinal sub-region meets the threshold of the environmental characteristics, the longitudinal sub-region is marked as a high point cloud environment region.

[0019] When there are at least four high point cloud environment areas in each lateral area, and the number of qualified confident point clouds meets the set overall high point cloud number threshold, the average height and height standard deviation of all confident point clouds in each lateral area are counted. When the average height and height standard deviation indicators meet the high point cloud environment distribution characteristics, it is judged that there is a high point cloud environment in the lateral area of ​​the current frame.

[0020] The signal-to-noise ratio threshold is calculated based on the pitch angle and horizontal angle of the point cloud in the vehicle coordinate system using the radar antenna diagram of the 4D millimeter-wave radar used. The conversion formula is as follows:

[0021]

[0022] In the formula, and [SNR i ] represent the original SNR and the converted SNR of the i-th point cloud, respectively. Indicates the horizontal angle as Yaw i and the SNR reduction factor corresponding to the horizontal angle of 0, Indicates the pitch angle is pitch i And the SNR reduction coefficient when the corresponding pitch angle is 0.

[0023] Establish a sliding time window, and perform continuous high point cloud environment recognition based on all single-frame high point cloud environment judgment results within the time window to obtain high point cloud environment recognition results, including:

[0024] The sliding time window with a preset time length of T stores the recognition results of the high point cloud environment in each frame of the three horizontal regions. T , the average height of the point cloud AvgHeight T , StdHeight of the height standard deviation of the confident point cloud T And the number of confident point clouds Num T :

[0025] The high point cloud environment recognition results in the sliding time window are judged. When the nth frame Result n If it is identified as a high point cloud environment, the SUM value of the frames that meet the high point cloud environment is increased by 1. When the SUM number is greater than 0.5T, that is, at least half of the frames in the current time window have identified the high point cloud environment, it is considered that the current vehicle is in a high point cloud environment.

[0026] A counter is set. When the vehicle is in a high point cloud environment, the counter increases. When the vehicle is not in a high point cloud environment, the counter decreases. When the counter is not 0, it is always considered that the vehicle is in a high point cloud environment.

[0027] A second aspect of the present invention provides a high point cloud environment recognition and detection system based on 4D millimeter wave radar, comprising:

[0028] An information acquisition unit, used to acquire forward point cloud information of radar coordinate system of objects in the area in front of the vehicle by using the vehicle-mounted forward 4D millimeter wave radar;

[0029] A coordinate conversion unit, used for converting the forward point cloud information of the radar coordinate system into the point cloud information of the vehicle coordinate system according to the installation information of the 4D millimeter wave radar;

[0030] A region division unit is used to divide the point cloud information of the vehicle coordinate system into three lateral detection regions, and each lateral detection region is divided into five longitudinal sub-regions;

[0031] A single-frame judgment unit is used to count the distribution of high point clouds and point cloud features in each sub-area, perform single-frame high point cloud environment judgment, and obtain a single-frame high point cloud environment judgment result;

[0032] The environment recognition unit is used to establish a sliding time window, and perform continuous high point cloud environment recognition according to all single-frame high point cloud environment judgment results within the time window to obtain a high point cloud environment recognition result.

[0033] A third aspect of the present invention provides an electronic device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected through lines; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned high point cloud environment recognition and detection method based on 4D millimeter-wave radar.

[0034] The fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the above-mentioned high point cloud environment recognition and detection method based on 4D millimeter wave radar.

[0035] The present invention has the following beneficial effects:

[0036] The present invention provides a high-point cloud environment recognition and detection method based on 4D millimeter-wave radar. The method takes into account the actual state of the high-point cloud environment scene and uses the 4D millimeter-wave radar's own detection results to perform environment recognition. The environment recognition results enable the radar to optimize its performance according to the environment, thereby improving the radar's behavior chain of "recognition-calibration-perception" that only relies on itself over time. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Schematic diagram of the high point cloud environment recognition and detection method based on 4D millimeter wave radar. DETAILED DESCRIPTION

[0038] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0039] Point cloud information

[0040] Point cloud information is a data structure that expresses objects or scenes in three-dimensional space. It consists of multiple three-dimensional coordinate points (X, Y, Z). Each point represents a position in space and may contain other attribute information. A point cloud is a data set of points in space that can represent a three-dimensional shape or object. These points have no explicit connection relationship, but together describe the shape, surface, and other characteristics of the object. Each point in a point cloud is usually represented by a three-dimensional coordinate (X, Y, Z) to represent the position, and may also contain attribute information such as color (R, G, B), intensity, normal vector, texture coordinates, etc. Point clouds can be ordered or unordered. An ordered point cloud represents the topological structure of an object, while an unordered point cloud is closer to the data form of the real world.

[0041] 4D millimeter wave radar

[0042] 4D millimeter-wave radar refers to a millimeter-wave radar with the ability to detect in four dimensions: speed, distance, horizontal angle, and vertical height. Compared with traditional 3D millimeter-wave radar, 4D millimeter-wave radar adds the detection of the dimension of "height", making it more diverse in terms of the dimensions of information it acquires. At the same time, it has the advantages of higher resolution and accuracy, as well as long transmission distance and strong penetration. 4D millimeter-wave radar determines the position, speed, height and other information of the target by emitting millimeter-wave signals and receiving reflected signals. Advanced signal processing technology and algorithms are used to improve the signal-to-noise ratio and reduce the false alarm rate. 4D millimeter-wave radar can form point cloud imaging-level output and perform image recognition in a data-driven manner.

[0043] Confirm point cloud

[0044] Point Cloud is a data structure used to represent objects or scenes in three-dimensional space. It consists of a large number of points, each of which contains location information in space and possible other attributes. These points can be regarded as sampling points in three-dimensional space. There is no explicit connection between them, but they describe the shape, surface and other characteristics of the object together.

[0045] Point cloud data usually has the following characteristics:

[0046] Disorder: There is no specific order for the points in point cloud data, and each point exists independently. This feature requires the algorithm that processes point cloud data to be able to adapt to disordered data.

[0047] Rotation invariance: When point cloud data is rotated in space, the 3D object it represents remains unchanged. This property is of great significance to the robustness of point cloud processing algorithms.

[0048] High dimensionality and large scale: Point cloud data usually contains millions of points, and each point may contain multiple attributes (such as color, normal, curvature, etc.), which requires the use of large-scale computer clusters for processing.

[0049] Discreteness: Point cloud data is discrete, and the positions of data points, the intervals between them, etc. are irregularly distributed in the point cloud space.

[0050] Uneven distribution: The farther the sampling area is from the acquisition device, the lower the point density; the closer the distance is, the higher the point density. At the same time, there may be holes in the point cloud due to objects blocking each other.

[0051] High point cloud environment

[0052] By counting the number of confident point clouds in each longitudinal sub-area, according to the characteristics of radar signal detection capability, if the number of confident point clouds in each sub-area meets the threshold of environmental characteristics, the sub-area is marked as a high point cloud environment area.

[0053] Sliding time window

[0054] Sliding Time Window is a method of processing and analyzing data by dividing time series data into multiple fixed-size time windows and sliding these windows over time. The data in each window is regarded as an independent data set, which can be used to calculate statistics, perform pattern recognition or perform other data processing tasks.

[0055] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of a high point cloud environment recognition and detection method based on a 4D millimeter wave radar in an embodiment of the present invention includes:

[0056] The embodiment of the present invention provides a high point cloud environment recognition and detection method based on 4D millimeter wave radar, such as Figure 1 As shown, the method comprises the following steps:

[0057] S1: Use the on-board forward 4D millimeter wave radar to sense the point cloud information of objects in the area in front of the vehicle, including dynamic point clouds of moving targets and static point clouds of stationary targets. The point cloud data is a single frame of data acquired in real time. If a frame is lost, the point cloud data of the previous frame is used.

[0058] S2: According to the installation information of the 4D millimeter wave radar, the forward point cloud information detected in the radar coordinate system is converted to the vehicle coordinate system by using coordinate transformation. In practical applications, the installation position of the radar can be fixed in advance by fixing the installation position, including the installation information of the (x, y, z) coordinates and (yaw, pitch) horizontal angle and pitch angle relative to the geometric center of the vehicle.

[0059] The vehicle coordinate system takes the geometric center of the vehicle as its origin, the positive direction of the X-axis points to the vehicle's forward direction, the positive direction of the Y-axis is perpendicular to the vehicle's forward direction and points to the left side of the vehicle, and the positive direction of the Z-axis is perpendicular to the ground and points to the sky.

[0060] S3: Divide the lateral detection area into three areas to detect the point cloud status information in the left side, front and right side of the vehicle respectively; divide the lateral detection area into five longitudinal sub-areas, analyze the distribution of high point clouds in the corresponding lateral areas and the characteristics of longitudinal distribution, and make single-frame high point cloud environment judgment.

[0061] Specifically, "analyzing the distribution of high point clouds in the corresponding horizontal area and the characteristics of the vertical distribution to make a single-frame high point cloud environment judgment" includes the following steps:

[0062] S31: Divide the detection area into three horizontal detection areas and their corresponding five vertical detection sub-areas.

[0063] Specifically, the X value range of the detection area is [L min , L max ], the Y value range is [-2d, 2d], and the X value ranges of the left, middle and right sides of the three horizontal detection areas are all [L min , L max ], and the Y value ranges are [0, 2d], [-d, d] and [-2d, 0] respectively.

[0064] For the vertical detection sub-area, the Y value range is the same as the corresponding horizontal detection area, with a width of 2d, and the X value range depends on the position of the vertical sub-area. min , L max] is divided into five longitudinal sub-regions, the lengths of the longitudinal sub-regions are L1, L2, L3, L4 and L5 respectively, and finally a total of 15 detection sub-regions with equal widths but different lengths are obtained, and each transverse region contains five longitudinal sub-regions.

[0065] Taking the middle detection area as an example, the Y value ranges of the five vertical detection sub-areas are all [-d, d], and the X value ranges are [L min , L min +L1]、[L min +L1,L min +L1+L2]、[L min +L1+L2,L min +L1+L2+L3]、[L min +L1+L2+L3,L min +L1+L2+L3+L4]、[L min +L1+L2+L3+L4,L max ].

[0066] S32: Analyze the distribution of high point clouds in the corresponding horizontal area and the characteristics of the vertical distribution, and make a single-frame high point cloud environment judgment.

[0067] The static point cloud heights in each longitudinal sub-area in each lateral area are statistically analyzed, and the confident point clouds that meet the high point cloud environment characteristics are screened out according to the thresholds of the point cloud height and signal-to-noise ratio characteristics. Since the radar signal strength is different at different pitch angles and horizontal angles, the signal-to-noise ratio threshold is converted to a certain proportion based on the pitch angle and horizontal angle of the point cloud in the vehicle coordinate system using the radar antenna diagram of the 4D millimeter-wave radar used. The conversion formula is as follows:

[0068]

[0069] In the formula, and [SNR i ] represent the original SNR and the converted SNR of the i-th point cloud, respectively. Indicates the horizontal angle as Yaw i and the SNR reduction factor corresponding to the horizontal angle of 0, Indicates the pitch angle is pitch i And the SNR reduction coefficient when the corresponding pitch angle is 0.

[0070] The number of confident point clouds in each longitudinal sub-area is counted. According to the characteristics of radar signal detection capability, if the number of confident point clouds in each sub-area meets the threshold of environmental characteristics, the sub-area is marked as a high point cloud environment area.

[0071] When there are at least four high point cloud environment areas in each lateral area, and the number of qualified confident point clouds meets the set overall high point cloud number threshold, the average height and height standard deviation of all confident point clouds in each lateral area are statistically distributed. When the average height and height standard deviation indicators meet the high point cloud environment distribution characteristics, it is judged that there is a high point cloud environment in the lateral area of ​​the current frame.

[0072] S4: Establish a sliding time window with a time length of T, perform continuous high-point cloud environment recognition, count the environment recognition status within the time window, and obtain a stable high-point cloud environment recognition result.

[0073] The sliding time window with a preset time length of T stores the recognition results of the high point cloud environment in each frame of the three horizontal regions. T , the average height of the point cloud AvgHeight T , StdHeight of the height standard deviation of the confident point cloud T And the number of confident point clouds Num T .

[0074] The high point cloud environment recognition results in the sliding time window are judged. When the nth frame Result n If it is identified as a high point cloud environment, the SUM value of the frames that meet the high point cloud environment is increased by 1. When the SUM number is greater than 0.5T, that is, at least half of the frames in the current time window have identified the high point cloud environment, it is considered that the current vehicle is in a high point cloud environment.

[0075] A counter is added to ensure the recognition of the high point cloud environment in the curve scene. When the vehicle is in a high point cloud environment, the counter increases. When the vehicle is not in a high point cloud environment, the counter decreases. When the counter is not 0, it is always considered that the vehicle is in a high point cloud environment.

[0076] By counting the number of confident point clouds that meet the height and signal-to-noise ratio thresholds in each longitudinal sub-area, and combining the detection capability evaluation of the 4D millimeter-wave radar, the high point cloud environment sub-area that meets the conditions can be accurately identified. This method ensures that only the true high point cloud environment will be identified, avoiding false positives and missed positives.

[0077] The signal-to-noise ratio threshold is dynamically adjusted according to the pitch angle and horizontal angle of the point cloud in the ego-vehicle coordinate system, which fully considers the influence of the radar antenna diagram, makes the threshold setting more reasonable, and improves the recognition accuracy.

[0078] Using a sliding time window to store and analyze the high-point cloud environment detection results of multiple time frames can achieve continuous recognition of the high-point cloud environment. This method effectively filters out occasional noise and false alarms and improves the stability of recognition.

[0079] A counting mechanism is introduced to enhance the recognition of high point cloud environments in curved scenes. When a vehicle enters a curve, the change in radar viewing angle may cause the high point cloud environment to disappear temporarily or be falsely reported. Through the counting mechanism, as long as the counter is non-zero, the vehicle is always considered to be in a high point cloud environment, thereby improving the recognition accuracy in curved scenes.

[0080] In summary, the present invention proposes a high point cloud environment recognition and detection method based on 4D millimeter-wave radar, which identifies high point cloud environments such as tunnels and overhead bridges by relying only on the 4D millimeter-wave radar itself, thereby meeting the environmental prior requirements of dynamic calibration, signal processing and target perception.

[0081] Compared with traditional radar, the 4D millimeter wave radar of the present invention can provide richer three-dimensional spatial information (including distance, direction, height and speed), thereby realizing high-precision and high-resolution perception of objects in front. This helps to accurately identify and distinguish different types of obstacles, such as vehicles, pedestrians, obstacles, etc.

[0082] By converting the point cloud information in the radar coordinate system to the vehicle coordinate system, the present invention can more effectively integrate it with other sensor data of the vehicle (such as cameras, lidar, etc.), thereby improving the accuracy and robustness of overall environmental perception. This coordinate unification also facilitates subsequent path planning and decision control.

[0083] The present invention divides the sensing area into three horizontal detection areas and five vertical sub-areas, which helps to analyze the point cloud status in different areas in a targeted manner, reduce the amount of data processing, and improve processing efficiency. At the same time, by statistically analyzing the distribution and characteristics of high point clouds in each area, the high point cloud status of the current environment can be more accurately judged, providing a basis for subsequent decision-making.

[0084] The present invention can effectively filter out occasional noise and false alarms by establishing a sliding time window for continuous high-point cloud environment recognition, thereby improving the stability and reliability of environment recognition. This continuous monitoring mechanism is particularly important for dynamically changing traffic environments, and can ensure the safe driving of vehicles in complex environments.

[0085] Strong adaptability and scalability:

[0086] The solution is based on 4D millimeter-wave radar, which is not affected by external conditions such as light and weather, and has strong environmental adaptability. At the same time, by adjusting parameters such as area division and time window size, it can flexibly adapt to different driving scenarios and needs. In addition, the solution can also be integrated with other sensors and algorithms to further improve the performance and functions of the perception system.

[0087] The above describes the high point cloud environment recognition and detection method based on 4D millimeter wave radar in an embodiment of the present invention. The following describes the high point cloud environment recognition and detection device based on 4D millimeter wave radar in an embodiment of the present invention. The embodiment of the high point cloud environment recognition and detection device based on 4D millimeter wave radar in an embodiment of the present invention includes:

[0088] An information acquisition unit, used to acquire forward point cloud information of radar coordinate system of objects in the area in front of the vehicle by using the vehicle-mounted forward 4D millimeter wave radar;

[0089] A coordinate conversion unit, used for converting the forward point cloud information of the radar coordinate system into the point cloud information of the vehicle coordinate system according to the installation information of the 4D millimeter wave radar;

[0090] A region division unit is used to divide the point cloud information of the vehicle coordinate system into three lateral detection regions, and each lateral detection region is divided into five longitudinal sub-regions;

[0091] A single-frame judgment unit is used to count the distribution of high point clouds and point cloud features in each sub-area, perform single-frame high point cloud environment judgment, and obtain a single-frame high point cloud environment judgment result;

[0092] The environment recognition unit is used to establish a sliding time window, and perform continuous high point cloud environment recognition according to all single-frame high point cloud environment judgment results within the time window to obtain a high point cloud environment recognition result.

[0093] The structure of an electronic device provided by an embodiment of the present invention may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) (for example, one or more processors) and memories, and one or more storage media for storing applications or data (for example, one or more mass storage devices). Among them, the memories and storage media may be short-term storage or permanent storage. The program stored in the storage medium may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the electronic device. Furthermore, the processor may be configured to communicate with the storage medium and execute a series of instruction operations in the storage medium on the electronic device.

[0094] The electronic device may also include one or more power supplies, one or more wired or wireless network interfaces, one or more input and output interfaces, and / or one or more operating systems, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that the electronic device structure does not constitute a limitation on the electronic device, and may include more or less components than the aforementioned, or combine certain components, or arrange the components differently.

[0095] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of a high point cloud environment recognition and detection method based on a 4D millimeter-wave radar.

[0096] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0097] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0098] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high point cloud environment recognition and detection method based on 4D millimeter wave radar, characterized in that: The following steps are involved: Use the vehicle-mounted forward 4D millimeter-wave radar to obtain the forward point cloud information of the radar coordinate system of the objects in the area in front of the vehicle; According to the installation information of the 4D millimeter wave radar, the forward point cloud information of the radar coordinate system is converted into the point cloud information of the vehicle coordinate system; The point cloud information of the vehicle coordinate system is divided into three lateral detection areas, and each lateral detection area is divided into five longitudinal sub-areas; Count the distribution of high point clouds and point cloud features in each sub-area, perform single-frame high point cloud environment judgment, and obtain single-frame high point cloud environment judgment results; A sliding time window is established, and continuous high point cloud environment recognition is performed based on all single-frame high point cloud environment judgment results within the time window to obtain the high point cloud environment recognition results.

2. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 1 is characterized in that: The point cloud information includes point cloud information after a large number of noise points are removed from the field of view of the 4D millimeter-wave radar. The point cloud information is located in a traditional radar coordinate system and includes horizontal angle measurement, pitch angle measurement, Doppler velocity, and relative distance information of the point cloud relative to the 4D millimeter-wave radar.

3. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 1 is characterized in that: The three lateral detection areas include: The horizontal detection area has the same length in the positive direction of the X axis and is at the closest detection distance L min And the maximum detection distance L max The distances between them on the Y axis are equal, but their positions are different. The Y value ranges of the left, front, and right areas are [0,2d], [-d,d], [-2d,0], respectively, where 2d is the preset width of the horizontal detection area. The five longitudinal sub-areas include: The horizontal detection area is divided into five longitudinal sub-areas, and the lengths of the longitudinal sub-areas are L1, L2, L3, L4 and L5 respectively, and finally a total of 15 detection sub-areas with equal width and different lengths are obtained, and each horizontal area contains five longitudinal sub-areas.

4. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 1 is characterized in that: The statistics of the high point cloud distribution and point cloud features in each sub-area, and the single-frame high point cloud environment judgment are performed to obtain the single-frame high point cloud environment judgment result, specifically including: Statistically analyze the static point cloud heights in each longitudinal sub-area, and select the confident point clouds that meet the high point cloud environment characteristics according to the point cloud height and signal-to-noise ratio threshold; The number of confident point clouds in each longitudinal sub-region is counted. If the number of confident point clouds in each longitudinal sub-region meets the threshold of the environmental characteristics, the longitudinal sub-region is marked as a high point cloud environment region. When there are at least four high point cloud environment areas in each horizontal area, and the number of qualified confident point clouds meets the set overall high point cloud number threshold, the average height and height standard deviation of all confident point clouds in each horizontal area are counted. When the average height and height standard deviation indicators meet the high point cloud environment distribution characteristics, it is judged that there is a high point cloud environment in the horizontal area of ​​the current frame.

5. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 4 is characterized in that: The signal-to-noise ratio threshold is calculated based on the pitch angle and horizontal angle of the point cloud in the vehicle coordinate system using the radar antenna diagram of the 4D millimeter-wave radar used. The conversion formula is as follows: In the formula, and [SNR i ] represent the original SNR and the converted SNR of the i-th point cloud, respectively. Indicates the horizontal angle as Yaw i and the SNR reduction factor corresponding to the horizontal angle of 0, Indicates the pitch angle is pitch i And the SNR reduction coefficient when the corresponding pitch angle is 0.

6. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 1 is characterized in that: The step of establishing a sliding time window and performing continuous high point cloud environment recognition according to all single-frame high point cloud environment judgment results within the time window to obtain a high point cloud environment recognition result specifically includes: The sliding time window with a preset time length of T stores the recognition results of the high point cloud environment in each frame of the three horizontal regions. T , the average height of the point cloud AvgHeight T , StdHeight of the height standard deviation of the confident point cloud T And the number of confident point clouds Num T ; The high point cloud environment recognition results in the sliding time window are judged. When the nth frame Result n If it is identified as a high point cloud environment, the SUM value of the frames that meet the high point cloud environment is increased by 1. When the SUM number is greater than 0.5T, that is, at least half of the frames in the current time window have identified the high point cloud environment, it is considered that the current vehicle is in a high point cloud environment.

7. The high point cloud environment recognition and detection method based on 4D millimeter wave radar according to claim 6 is characterized in that: The continuous high point cloud environment recognition also includes: A counter is set. When the vehicle is in a high point cloud environment, the counter increases. When the vehicle is not in a high point cloud environment, the counter decreases. When the counter is not 0, it is always considered that the vehicle is in a high point cloud environment.

8. A high point cloud environment recognition and detection system based on 4D millimeter wave radar, characterized in that: The high point cloud environment recognition and detection system comprises: An information acquisition unit, used to acquire forward point cloud information of radar coordinate system of objects in the area in front of the vehicle by using the vehicle-mounted forward 4D millimeter wave radar; A coordinate conversion unit, used for converting the forward point cloud information of the radar coordinate system into the point cloud information of the vehicle coordinate system according to the installation information of the 4D millimeter wave radar; A region division unit is used to divide the point cloud information of the vehicle coordinate system into three lateral detection regions, and each lateral detection region is divided into five longitudinal sub-regions; A single-frame judgment unit is used to count the distribution of high point clouds and point cloud features in each sub-area, perform single-frame high point cloud environment judgment, and obtain a single-frame high point cloud environment judgment result; The environment recognition unit is used to establish a sliding time window, and perform continuous high point cloud environment recognition according to all single-frame high point cloud environment judgment results within the time window to obtain a high point cloud environment recognition result.

9. An electronic device, comprising a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the electronic device executes each step of the high point cloud environment recognition and detection method based on 4D millimeter wave radar as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by the processor, each step of the high point cloud environment recognition and detection method based on 4D millimeter wave radar as described in any one of claims 1 to 7 is implemented.

Images