A target detection method based on millimeter wave radar beamforming

Abstract

The present application belongs to the technical field of millimeter wave radar target detection, and particularly relates to a target detection method based on millimeter wave radar beam forming, comprising the following steps: obtaining millimeter wave radar original ADC data; performing FFT processing on the ADC data in the distance dimension and the Doppler dimension respectively, sampling the environmental noise, and then using a segmented threshold CA-CFAR detection algorithm to detect the distance-Doppler spectrum under each angle to obtain a distance-velocity point cloud set; merging the point cloud sets under all angles and removing the repeated parts; performing DBF processing on all angles of each point cloud in the receiving antenna dimension to obtain angle information; and finally outputting the point cloud containing the target distance, velocity and angle. The present application can solve the problem that the conventional processing method of MIMO millimeter wave radar signals cannot be applied to millimeter wave radar signal processing based on transmit beam forming, and has a good market application prospect.

Description

Technical Field

[0001] This invention belongs to the field of millimeter-wave radar target detection technology, specifically relating to a target detection method based on millimeter-wave radar beamforming. Background Technology

[0002] Millimeter-wave radar is widely used due to its compact size, low cost, all-weather operation, and advantages in ranging, altitude measurement, speed measurement, and high range resolution in the field of autonomous driving. Based on different detection ranges, millimeter-wave radar is mainly divided into four categories: ultra-short-range millimeter-wave radar (USRR), short-range radar (SRR), medium-range radar (MRR), and long-range radar (LRR), with maximum detection ranges of 15 meters, 50 meters, 150 meters, and 250 meters, respectively. The detection range of a single millimeter-wave radar system is typically around 100 meters, while cascaded radar systems can reach around 250 meters. In the field of traffic radar, the required detection range for millimeter-wave radar may reach 300 meters or even over 600 meters.

[0003] Patent application CN202211174754.X discloses a method for improving azimuth resolution of moving targets based on millimeter-wave radar. This method includes preprocessing the radar echo signal and scanning a fan-shaped detection area using beamforming technology. Then, it acquires multiple main beam signals corresponding to the same azimuth direction for azimuth processing. Finally, it stitches together the azimuth information sub-distribution maps of targets at different azimuth directions from the multiple processed methods to form a total azimuth information distribution map of the target. This invention aims to improve radar azimuth target resolution by stitching together the azimuth information sub-distribution maps of targets at different azimuth directions from the multiple processed methods, but it cannot improve the detection range.

[0004] To effectively improve detection range, one current method is to use transmitter beamforming technology. Transmit beamforming controls the phase shift of each transmitting antenna in an antenna array, causing all transmitting antennas to emit millimeter waves in a specified direction simultaneously, thereby enhancing the radiated energy and improving the detection capability of targets within the detection area. For example, the AWR2243 chip from Texas Instruments (TI) supports configuring the phase of the transmit channel, achieving a detection range greater than that of TDM-MIMO radar through customized beam pointing. Since there are certain differences between transmit beamforming data and TDM-MIMO data, TDM-MIMO target detection methods are not entirely applicable. Therefore, this invention provides a target detection method flowchart suitable for millimeter-wave radar transmit beamforming mode.

[0005] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a target detection method based on millimeter-wave radar beamforming, so as to solve the problem that traditional processing methods for MIMO millimeter-wave radar signals cannot be applied to millimeter-wave radar signal processing based on transmitted beamforming.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A target detection method based on millimeter-wave radar beamforming includes the following steps: S1. Acquire raw ADC data from the millimeter-wave radar; S2. Perform a two-dimensional fast Fourier transform on the ADC data obtained in S1 to obtain multiple distance-Doppler spectrum matrices and preprocess them. S3. Use the segmented threshold CA-CFAR algorithm to detect the range-Doppler spectrum matrix obtained in S2 at each angle to obtain the target range and velocity point cloud set at different angles, merge the point cloud set at all angles and remove duplicate parts; S4. Perform digital beamforming processing on the receiving antenna dimension data of the point cloud set obtained in S3, and obtain the target angle information through peak index.

[0008] Preferably, S1 acquires the raw ADC data from the millimeter-wave radar by including the following steps: S1.1 Calculate the phase value of the transmitting antenna for the required beam scanning angle; S1.2 Configure the millimeter-wave radar radio frequency front-end according to the phase value; S1.3. Obtain the raw ADC data in 4 dimensions, where the four dimensions are: number of sampling points N, number of transmissions M, number of scanning angles S, and number of receiving antennas K.

[0009] The formula for calculating the phase value of the transmitting antenna is:

[0010] Configure the current scanning angle. The current distance between the transmitting antenna and the reference transmitting antenna. This is the phase value vector that should be set for all transmitting antennas at the current scanning angle.

[0011] The working method of millimeter-wave radar beamforming can be described as follows: calculate the phase shift values ​​of all transmitting antennas at the current angle to obtain a phase shift vector, then all transmitting antennas simultaneously transmit a linear frequency modulated pulse signal to the current angle, repeat these two steps S times to complete one scan, repeat the scan M times, each receiving antenna will receive the beam echo signal transmitted at each angle, and the ADC data obtained after sampling the intermediate frequency signal has a dimension of [NM KS].

[0012] As a preferred option, the preprocessing in S2 specifically involves using the zero-velocity channel nulling method to achieve static clutter filtering and incoherent accumulation in the receiving antenna dimension by zeroing the data of the zero Doppler dimension and the one unit before and after it.

[0013] The 4D ADC data acquired by S1 is processed using FFT in both the range dimension (number of sampling points) and the Doppler dimension (number of transmissions) to obtain the range-Doppler spectrum matrix, i.e., the RD matrix. The number of FFT operation points is calculated based on the powers of 2 closest to N and M. For 4D millimeter-wave radar data, this step yields S×K RD matrices.

[0014] To perform subsequent CFAR detection, the RD matrix needs to be processed. Specifically, an incoherent accumulation method is used to compress the four-dimensional data matrix ([NMKS]) containing the RD matrix in the receiving antenna dimension, resulting in an S×1×N×M matrix. The final RD matrix is ​​then provided to the CFAR detector. The target detection involved in this invention targets moving target detection. Static clutter removal is required before CFAR detection. This invention achieves this through a zero-velocity channel nulling method, which zeroes the data in the zero Doppler dimension and the two cell gates before and after it.

[0015] Preferably, the detection process using the segmented threshold CA-CFAR algorithm in S3 includes the following steps: S3.1 A certain number of protection units are set on the left and right sides of the unit to be tested; S3.2 Calculate the average values ​​of the reference units on the left and right sides, and denote them as X and Y respectively; S3.3 Multiply the mean of X and Y by the dynamic threshold factor to obtain a threshold; S3.4. Compare the obtained threshold with the value of the unit to be detected. If the value of the unit to be detected is larger, it is determined that there is a target at the index.

[0016] Transmit beamforming is primarily used for long-range target detection, while millimeter-wave radar target detection relies on received echo signals. The intensity of the reflected signal from the same target varies at different distances, and the background noise distribution is not uniform. Using a fixed threshold factor may lead to detection errors. This invention uses a piecewise threshold factor. As shown in the following formula:

[0017] in, For FFT points, As the initial threshold factor, The initial threshold factor is determined by global sampling during the first CFAR detection of each frame, and the sampling method is to sample every 8 units along the distance dimension and take the average value. The segmentation principle is that the farther the target is, the smaller the given threshold factor.

[0018] Preferably, S4 obtains the target angle information by including the following steps: S4.1 Perform digital beamforming processing on the receiving antenna dimension at each angle, stitch them into a 1*S array, and find the target angle value; S4.2. The response value is obtained by multiplying the antenna dimension data by the steering vector. The calculation formula is as follows:

[0019] This is the antenna dimension data matrix under the current distance index, Doppler index, and angle index, and its size is 1×K; S4.3 Calculate the response value of the current angle, loop S times to calculate the response vector of size 1×S, and take the maximum value as the current target angle index; S4.4. Scan the angle array to obtain the final target angle value.

[0020] The formula for calculating the guide vector in the direction is:

[0021] in, From the current perspective, Antenna spacing, Receiver antenna array arrangement matrix.

[0022] Compared with the prior art, the present invention has the following beneficial effects: (1) The target detection method based on millimeter-wave radar beamforming of the present invention controls the phase of each transmitting antenna in the antenna array to make the beam point to a designated area, thereby enhancing the radiated energy and improving the target detection capability of these areas.

[0023] (2) The target detection method based on millimeter-wave radar beamforming of the present invention uses the CA-CFAR method with segmented threshold to perform target detection on the RD matrix, which can effectively solve the problem of false detection caused by the target signal being affected by noise signal.

[0024] (3) The target detection method based on millimeter-wave radar beamforming of the present invention merges and deduplicates the point cloud data after CFAR processing, which can reduce the amount of calculation in subsequent steps. Attached Figure Description

[0025] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the beamforming of a millimeter-wave radar. Figure 3 This is a block diagram of the CA-CFAR algorithm principle; Figure 4 It is the distance-Doppler spectrum of two adjacent frames with greater environmental noise; Figure 5 It is the distance-Doppler spectrum of two adjacent frames with less ambient noise; Figure 6 This is a diagram showing the beamforming response of a single target at various angles; Figure 7 This is a point cloud diagram in Cartesian coordinates. Detailed Implementation

[0026] The technical solution of this invention patent will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0027] See attached document Figure 1 A target detection method based on millimeter-wave radar beamforming includes the following steps: Step 1: Obtain the raw ADC data from the millimeter-wave radar, specifically: First, calculate the phase value of the transmitting antenna for the required beam scanning angle, as shown in the attached figure. Figure 2 The diagram shows a millimeter-wave radar beamforming process. Transmit beamforming refers to controlling the phase of each transmitting antenna in an antenna array to point the beam at a designated area, thereby enhancing the radiated energy and improving the target detection capability and probability in those areas. For each pointing angle, the phase values ​​of all transmitting antennas need to be calculated to change the pointing direction. In this example, the beam scanning angle is set to […]. The step angle is 1°, meaning there are a total of Each scanning angle. The formula for calculating the phase value of the transmitting antenna is:

[0028] Configure the current scanning angle. The current distance between the transmitting antenna and the reference transmitting antenna. This is the phase vector that should be set for all transmitting antennas at the current scanning angle. Imagine you want to direct the beam in a specific direction. It is necessary to ensure that the electromagnetic waves emitted by each element in the array are parallel to this direction in the far field. Assume that the first element is used as a reference, and its initial phase is set to 0. For subsequent elements, in the direction... In this process, electromagnetic waves need to travel through a series of paths, and these paths are mapped onto the phase, which becomes the phase value of the corresponding transmitting antenna. After iteratively calculating the phase vectors corresponding to all scanning angles, the RF front-end parameters are configured and written, and finally, the raw ADC data is obtained from the millimeter-wave radar RF front-end.

[0029] In this embodiment, the number of sampling points is set to N, the number of transmissions per angle is M, and the number of transmitting antennas is K. The working mode of millimeter-wave radar beamforming can be described as follows: First, the phase shift value of all transmitting antennas at the current angle is calculated. Then, all transmitting antennas work simultaneously to transmit a continuous frequency modulated wave signal to the current angle. These two steps are repeated S times to complete one scan of all angles. The above three steps are repeated M times to achieve Doppler velocity measurement. Finally, the ADC data dimension is [NMKS].

[0030] Step 2: Perform FFT processing on the 4D ADC data obtained in Step 1 in both the distance dimension (number of sampling points N) and the Doppler dimension (number of transmissions M) to obtain the range-Doppler spectrum matrix, i.e., the RD matrix.

[0031] When performing FFT operations, the number of FFT operation points is calculated based on the powers of 2 closest to N and M. For 4D millimeter-wave radar data, this step yields S×K RD matrices. Subsequent processing of the RD matrix data is then performed. It should be noted that in RD(i,j), i represents the range index number, and j represents the Doppler index number, i=0,1,...,N-1, j=0,1,...,M-1.

[0032] Step 3: Use CFAR to perform target detection on the S×K RD matrices obtained in Step 2. Before detection, channel compression needs to be performed on the S×K RD matrices in the receiving antenna dimension. Channel compression usually has coherent accumulation and non-coherent accumulation methods. Non-coherent accumulation is to square the data of each channel and accumulate them in the channel dimension. Coherent accumulation is to first construct a steering vector based on the array arrangement at a certain angle, multiply this steering vector with the data of each channel to obtain a complex matrix, and finally take the square of the matrix.

[0033] Coherent accumulation can improve the signal-to-noise ratio of a target, but it involves a large amount of computation. Considering practical implementation, this embodiment uses an incoherent accumulation method, ultimately obtaining S×1 RD matrices. This embodiment targets scenarios involving both stationary and moving targets detected by radar, requiring static clutter removal before CA-CFAR detection. This invention achieves this by using a zero-velocity channel nulling method, which zeroes the data in the zero Doppler dimension and the data in the cells before and after it; after CA-CFAR detection, the point cloud sets from all angles are merged and duplicates are removed.

[0034] Target detection is performed on S RD matrices using the CA-CFAR method. Figure 3 This is a schematic diagram of CA-CFAR. In this embodiment, the CA-CFAR parameters include the range dimension and the Doppler dimension; specifically, the range dimension has 4 reference cells and 8 guard cells; the Doppler dimension has 4 reference cells and 2 guard cells.

[0035] The detection process of CA-CFAR can be described as follows: First, in the unit to be detected... A certain number of protection units need to be set on the left and right sides to prevent noise from neighboring units from affecting the unit under test. Calculate the average values ​​X and Y of the reference units on both sides of the detection unit, and then compare the average values ​​of X and Y with the dynamic threshold factor. Multiply, and finally add to the unit to be detected. Compare, if If the value is large, it is determined that a target exists at that index.

[0036] Threshold factor It is dynamically adjusted. Since the main application of transmitted beamforming is for long-range target detection, millimeter-wave radar target detection is based on echo signals. The intensity of the reflected signal from the same target varies at different distances, and the background noise is not necessarily uniform. Using a fixed threshold factor will lead to detection errors.

[0037] See attached document Figure 4 and attached Figure 5 This represents the RD matrix changes over two consecutive frames in the experiment of this embodiment. (See attached image) Figure 4In environments with significant noise, using the same threshold factor can easily mask the noise. This embodiment uses a segmented threshold factor, as shown in the following formula:

[0038] in, The distance dimension is the number of FFT points. As the initial threshold factor, The initial threshold factor, representing a scaling factor, is determined through global sampling during the first CA-CFAR detection of each frame. Specifically, sampling is performed at 8-unit intervals along the distance dimension, and the average value is taken. The threshold factor is segmented according to the principle that the farther the target distance, the smaller the given threshold factor.

[0039] It should be noted that this embodiment first performs detection in the range dimension, and then filters out the Doppler dimension data where the target index is located for CA-CFAR detection, rather than performing a complete detection. In the transmitter beamforming mode, the coverage areas of adjacent beams may have overlapping parts, and the RD matrix at different angles may have point cloud duplication after CA-CFAR detection in both the range and Doppler dimensions. This embodiment removes duplicate point clouds by taking the union of the point clouds at each angle, reducing the computational load of subsequent angle estimation.

[0040] Step 4: Perform receiver-side digital beamforming (DBF) on the receiving antenna dimension at each angle, stitch them into a 1*S array, and find the target angle value. DBF constructs guide vectors for each angle within the field of view, multiplies these guide vectors by the echo signal of the array to obtain the energy value at each angle, and the maximum value obtained is the direction of the actual target echo. The formula for calculating the directional guidance vector is as follows:

[0041] in From the current perspective, Antenna spacing, The receiving antenna array arrangement matrix is ​​then used, and the response value is obtained by multiplying the antenna dimension data by the steering vector, as shown in the following formula:

[0042] This is the antenna dimension data matrix with a size of 1×K, based on the current distance index, Doppler index, and angle index. The response value for the current angle can be calculated using two formulas, and the estimated angle response vector of size 1×S is calculated iteratively S times.

[0043] See attached document Figure 6The values ​​shown are the experimental results in this embodiment. The maximum value is then taken as the current target's angle index. After the above four steps, the angle, distance, and velocity information of the target point cloud in polar coordinates can be obtained. Figure 7 The diagram shows the conversion of the point cloud into a Cartesian coordinate system.

[0044] The target detection method based on millimeter-wave radar beamforming of the present invention effectively improves the detection range and solves the problem that the target detection method of TDM-MIMO cannot be fully adapted due to the certain difference between the data of transmitted beamforming and TDM-MIMO data. It is applicable to target detection in the transmitted beamforming mode of millimeter-wave radar.

[0045] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A target detection method based on millimeter-wave radar beamforming, characterized in that, Includes the following steps: S1. Acquire raw ADC data from the millimeter-wave radar; S2. Perform a two-dimensional fast Fourier transform on the ADC data obtained in S1 to obtain multiple distance-Doppler spectrum matrices and preprocess them. S3. Use the segmented threshold CA-CFAR algorithm to detect the matrix obtained in S2 to obtain the target distance and velocity point cloud sets under different angles and take the union of each set; S4. Perform digital beamforming processing on the receiving antenna dimension data of the point cloud set obtained in S3, and obtain the target angle information through peak index. S1 acquires raw ADC data from millimeter-wave radar through the following steps: S1.1 Calculate the phase value of the transmitting antenna for the required beam scanning angle; S1.2 Configure the millimeter-wave radar radio frequency front-end according to the phase value; S1.

3. Obtain the raw ADC data in 4 dimensions, where the four dimensions are: number of sampling points N, number of transmissions M, number of scanning angles S, and number of receiving antennas K.

2. The target detection method based on millimeter-wave radar beamforming according to claim 1, characterized in that, The preprocessing in S2 specifically involves using the zero-velocity channel zeroing method to achieve static clutter filtering and incoherent accumulation in the receiving antenna dimension by zeroing the data in the zero Doppler dimension and one cell before and after it.

3. The target detection method based on millimeter-wave radar beamforming according to claim 1, characterized in that, The detection process using the segmented threshold CA-CFAR algorithm in S3 includes the following steps: S3.1 Multiple protection units are set on the left and right sides of the unit to be tested; S3.2 Calculate the average values ​​of the reference units on the left and right sides, and denote them as X and Y respectively; S3.3 Multiply the mean of X and Y by the dynamic piecewise threshold factor to obtain a threshold; S3.

4. Compare the obtained threshold with the value of the unit to be detected. If the value of the unit to be detected is larger, it is determined that there is a target at the index.

4. The target detection method based on millimeter-wave radar beamforming according to claim 3, characterized in that, The segmented threshold factor The expression is: in, For FFT points, As the initial threshold factor, The initial threshold factor is determined by global sampling during the first CFAR detection of each frame, and the sampling method is to sample every 8 units along the distance dimension and take the average value. The segmentation principle is that the farther the target is, the smaller the given threshold factor.

5. The target detection method based on millimeter-wave radar beamforming according to claim 1, characterized in that, S4 acquires target angle information through the following steps: S4.1 Perform digital beamforming processing on the receiving antenna dimension at each angle, stitch them into a 1*S array, and find the target angle value; S4.

2. The response value is obtained by multiplying the antenna dimension data by the steering vector. The calculation formula is as follows: This is the antenna dimension data matrix under the current distance index, Doppler index, and angle index, and its size is 1×K; S4.3 Calculate the response value of the current angle, loop S times to calculate the response vector of size 1×S, and take the maximum value as the current target angle index; S4.

4. Scan the angle array to obtain the final target angle value.

6. The target detection method based on millimeter-wave radar beamforming according to claim 5, characterized in that, The formula for calculating the guide vector in the direction is: in, From the current perspective, Antenna spacing, Receiver antenna array arrangement matrix.

Images