A method, system, radar and vehicle for detecting obstacles by an in-vehicle millimeter-wave radar

Abstract

The present invention provides a method, a system, a radar and an automobile for detecting obstacles by an in-vehicle millimeter-wave radar. Specifically, the present invention constructs an echo signal model of a first detection target and a second detection target, separates the echo range image obtained by converting the echo signal, obtains the range image of the first detection target, and then adopts the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target. Thus, it effectively overcomes the problem that when using an in-vehicle millimeter-wave radar to detect the first detection target, since the radar moves following the automobile, the echo signal of the second detection target cannot be eliminated by using the traditional static clutter cancellation method. The signal separation method in the present invention can effectively separate the range image of the first detection target from the original echo range image, reduce the interference of the range image of the second detection target, and obtain an accurate positioning result of the first detection target.

Description

Technical Field

[0001] The present invention belongs to the technical field of millimeter-wave radar detection, and particularly relates to a method, a system, a radar and an automobile for detecting obstacles by an in-vehicle millimeter-wave radar. Background Art

[0002] With the wide popularity of car travel, the number of traffic accidents involving pedestrians has been increasing year by year. Therefore, pedestrian safety on urban roads has become an issue that has attracted more and more attention.

[0003] Currently, there are mainly two methods to prevent pedestrian collision accidents. One is passive pedestrian protection, that is, by designing components such as bumpers and airbags of the vehicle, and reducing the harm caused by colliding with pedestrians through a physical protection model. The other is active pedestrian protection, such as pedestrian detection, collision warning, vehicle autonomous braking, etc. Existing pedestrian detection measures can only prevent pedestrians who are not blocked in front of the radar vehicle. For the first detection target of the vehicle in front of the radar vehicle, due to the strong interference of the echo of the vehicle in front, it is impossible to effectively detect the weak target of the pedestrian beside the vehicle. Therefore, the detection of the weak target of the pedestrian beside the vehicle has become one of the research difficulties in the field of in-vehicle radar anti-collision.

[0004] Many foreign research institutions have carried out research on the detection of pedestrians beside the vehicle. In 2017, scholars from the Korea Automobile Technology Research Institute used lidar-radar sensor fusion processing to achieve the detection of road pedestrian targets. This technology uses radar to detect occluded pedestrians and estimates the detection of occluded pedestrians using Doppler distribution. In 2019, scholars from Waseda University used an L-band radar to achieve non-line-of-sight pedestrian detection. In this technology, pedestrians need to carry a transponder so that the radar can receive the echo of the pedestrian target, and the distance value of the pedestrian is obtained through the echo phase information. From the above literature, it can be seen that currently, pedestrian detection mainly detects pedestrian targets through Doppler information, and auxiliary means are required to achieve pedestrian detection for the radar to detect weak pedestrian targets.

[0005] However, when the weak target of the pedestrian beside the vehicle is near the vehicle in front of the radar vehicle, due to the interference of the vehicle in front, it is difficult to accurately judge the position of the weak target of the pedestrian beside the vehicle by the above methods. Summary of the Invention

[0006] To solve the above technical problems, the present invention proposes a method, a system, a radar and an automobile for detecting obstacles by an in-vehicle millimeter-wave radar. Based on the radar echo signal model, the echo signal is separated on the echo range image, and the MVDR angle measurement and positioning method can be combined to accurately obtain the positioning result of the weak target of the pedestrian beside the vehicle.

[0007] The present invention provides a method for detecting obstacles by an in-vehicle millimeter-wave radar, including the following steps:

[0008] S1: Construct an echo signal model according to the detection scenario and detection targets of the vehicle-mounted millimeter-wave radar; the detection targets include a first detection target and a second detection target;

[0009] S2: Obtain the echo signals corresponding to the detection targets according to the echo signal model;

[0010] S3: Separate the echo signals on the echo range image to obtain the range image X of the first detection target human ;

[0011] S4: Use the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target.

[0012] In the above S1, constructing the echo signal model according to the detection scenario and detection targets of the vehicle-mounted millimeter-wave radar is specifically as follows:

[0013] S11: Define that the detection scenario of the vehicle-mounted millimeter-wave radar includes a first area and a second area; define that the first detection target is located in the first area and the second detection target is located in the second area;

[0014] S12: Obtain the path echo time delays of the first detection target and the second detection target respectively according to the detection path of the vehicle-mounted millimeter-wave radar, and construct an echo signal model according to the path echo time delays.

[0015] Further, the path echo time delay is specifically as follows:

[0016]

[0017] Among them, τ path-1 is the echo time delay of the direct path for detecting the second detection target, τ path-2 is the ground one-time reflection path time delay for detecting the first detection target, τ path-3 is the direct path for detecting the first detection target; OA is the direct path between the current vehicle-mounted millimeter-wave radar and the second detection target, OC is the direct path between the current vehicle-mounted millimeter-wave radar and the first detection target, OB and BC are the indirect paths when the first detection target is in front of the second detection target and the echo of the OC path passes through the ground point B as the intermediate reflection point; c is the electromagnetic wave propagation speed.

[0018] The echo signal model is specifically as follows:

[0019] Suppose the linear frequency modulation signal of the vehicle-mounted millimeter-wave radar is:

[0020] s(t) = A0exp(j2πf0t + jπμt 2 )u(t);

[0021]

[0022] Construct an echo signal model, and the formula is:

[0023] y1(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-2 ) + n(t);

[0024] y2(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-3 ) + n(t);

[0025] In the above formula, f0 is the carrier frequency, A0 is the amplitude of the transmitted signal, μ = B / T is the linear frequency modulation slope, B is the signal bandwidth, T is the pulse time; u(t) is the rectangular function; σ1 is the scattering coefficient of the second detection target, σ2 is the scattering coefficient of the first detection target, and n(t) represents the background noise.

[0026] Further, the specific content of S3 is as follows:

[0027] S31: Perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range image X ∈ C M×N , and de-mean the original range image X to obtain the de-mean matrix

[0028] The specific calculation process is as follows:

[0029] De-mean the original range image X ∈ C M×N . Assume that each radar signal has M cycles, and a total of N radar signals are received. Define the input vector as x n = [x 1,n L x M,n T , and the original range image matrix is X = [x1 L x N , where x m,n is the nth range value in the mth cycle. Calculate the mean vector as:

[0030]

[0031] Define the mean vector of the original range image X ∈ C M×N as μ = [μ1 L μ M T , and the de-mean operation of the input vector can be expressed as The original range image after de-meaning is expressed as

[0032] S32: Calculate the de-mean matrix ​​The covariance matrix D is obtained and eigenvalue decomposition is performed. The eigenvalues are arranged in descending order, and the eigenvectors corresponding to the top k eigenvalues are selected as the mapping matrix for dimensionality reduction, resulting in a matrix X that contains the principal component information of the second detection target. front Based on the original range image X and the matrix X containing the principal component information of the second detection target front the range image X of the first detection target is calculated as human .

[0033] The specific calculation process is as follows:

[0034] Calculate the mean-removed matrix The covariance matrix is calculated using the following formula:

[0035]

[0036] Perform eigenvalue decomposition on the above covariance matrix using the following formula:

[0037] D = UΣU -1 ;

[0038] After eigenvalue decomposition, Σ is a diagonal matrix, and U is a matrix composed of the eigenvectors of matrix D. The corresponding eigenvectors are U1, L, U M . Then, the eigenvectors are arranged in descending order according to the corresponding eigenvalues. At this time, the eigenvectors corresponding to the top k eigenvalues are selected as the mapping matrix for dimensionality reduction. The mapping matrix can be expressed as The dimensionality-reduced data matrix can be expressed as:

[0039]

[0040] This is because the range image of the first detection target signal and the range image of the second detection target signal are mixed to form the original range image. Among them, the signal intensity of the second detection target is the largest, and it can be approximately considered that the original range image X mainly contains the range image of the second detection target. Therefore, the matrix X containing the principal component information of the second detection target is obtained through dimensionality reduction. front , and the separation process of the range images of the first detection target and the second detection target is essentially completed. The range image X of the first detection target human :

[0041] X human = X - X front .

[0042] Preferably, S3 also uses non-coherent superposition and constant false alarm rate detection to test the calculated range image to obtain the range R of the first detection target human .

[0043] Specifically, S4 is as follows:

[0044] In a uniform linear array model, the direction vector a(θ) of the antenna receiving signal is expressed as:

[0045]

[0046] Among them, d is the spacing between two adjacent antennas in the radar, φ is the phase difference between adjacent antennas, k is the number of antennas, θ is the incident angle of the radar signal, and λ is the wavelength of the radar signal.

[0047] After obtaining the distance of the first detection target, the minimum variance distortionless response method (MVDR) is used to calculate the azimuth angle of the target. The average output power P of the spatial domain filter MVDR is expressed as:

[0048]

[0049] Among them, R = E{x human (j)x human H (j)}, R represents the autocorrelation matrix of the input matrix X of the range profile of the first detection target human .

[0050] Then, θ in the steering vector is changed in the [-π, π] angle interval to obtain the variation curve of P MVDR (θ) and perform spectral peak search. At this time, the angle corresponding to the peak point is the azimuth angle θ of the first detection target human .

[0051] After obtaining the azimuth angle θ of the first detection target by applying the MVDR angle measurement method human , combined with the distance R of the first detection target human , the two-dimensional position coordinates of the first detection target are calculated as:

[0052]

[0053] Preferably, the present invention further provides a vehicle-mounted millimeter-wave radar obstacle detection system, which at least includes:

[0054] A radar module for detecting detection targets in the detection scene and obtaining radar reflected waves;

[0055] A detection module for detecting obstacles in front of the vehicle. The detection module at least includes an analog unit and a calculation unit;

[0056] The analog unit stores an echo signal model constructed according to the vehicle-mounted millimeter-wave radar detection scene and detection target, and is used to obtain the echo signal corresponding to the detection target;

[0057] The calculation unit is used to separate the echo signal on the echo range profile to obtain the range profile X of the first detection target human, adopt the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target;

[0058] A display unit for presenting the position of the first detection target.

[0059] Furthermore, the simulation unit further includes:

[0060] Define that the detection scenario includes a first area and a second area; define that the first detection target is located in the first area and the second detection target is located in the second area;

[0061] According to the detection path of the radar module, obtain the path echo time delays of the first detection target and the second detection target respectively, and construct an echo signal model according to the path echo time delays.

[0062] The calculation unit further includes:

[0063] Perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range profile X ∈ C M×N , de-mean the original range profile X to obtain a de-mean matrix

[0064] Calculate the covariance matrix D of the de-mean matrix and perform eigenvalue decomposition, arrange the eigenvalues in descending order, select the eigenvectors corresponding to the first k eigenvalues as the mapping matrix for dimensionality reduction, and the dimensionality reduction obtains a matrix X containing the principal component information of the second detection target front , obtain the range profile X of the first detection target human .

[0065] Preferably, the present invention also provides a radar, which is one of the vehicle-mounted millimeter-wave radars, installed anywhere on the vehicle, and the radar is communicatively connected to the vehicle-mounted control center and is used to implement a vehicle-mounted millimeter-wave radar obstacle detection method as described above.

[0066] Preferably, the present invention also provides a vehicle, including at least:

[0067] A vehicle-mounted millimeter-wave radar installed anywhere on the vehicle, and the vehicle-mounted millimeter-wave radar is used to obtain detection target information in the detection scenario and send it to a computer-readable storage medium at the vehicle-mounted MCU or SoC end;

[0068] A computer program is stored on the computer-readable storage medium;

[0069] And one or more processors, configured to execute the program in the computer-readable storage medium to implement a vehicle-mounted millimeter-wave radar obstacle detection method as described above, and present the position of the first detection target calculated by the method through a display unit.

[0070] The beneficial effects of the present invention are as follows. When detecting the first detection target using a vehicle-mounted millimeter-wave radar, since the radar moves along with the vehicle and the traditional static clutter cancellation method cannot be used to eliminate the echo signal of the second detection target, a signal separation method is adopted to separate the range image of the first detection target from the original echo range image, reducing the interference of the range image of the second detection target and obtaining an accurate positioning result of the first detection target. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 Fig. a is a schematic side view of the scene of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0072] Figure 1 Fig. b is a schematic top view of the scene of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0073] Figure 2 Fig. is a specific test diagram of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0074] Figure 3 Fig. is a schematic diagram of the original range image of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0075] Figure 4 Fig. a is a schematic diagram of the range images of the first detection target and the second detection target (front vehicle) of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0076] Figure 4 Fig. b is a schematic diagram of the range images of the first detection target and the second detection target (pedestrian) of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention.

[0077] Figure 5 Fig. is a schematic diagram of the positioning of the first detection target of the vehicle-mounted millimeter-wave radar obstacle detection method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0078] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0079] Embodiment 1:

[0080] The application scenario of the present invention is as Figure 1As shown in the figure, where the vehicle A ahead is the second detection target, located in the direct vision area in front of the vehicle-mounted millimeter-wave radar O, that is, the second area; and the area in front of the second area is the first area. Preferably, the first area is used to detect the first detection target C. Preferably, the first detection target includes at least pedestrians and small animals, etc., but is not limited thereto. Based on this, a radar detection echo path is constructed, including at least:

[0081] The first echo path: When the vehicle-mounted millimeter-wave radar O can directly detect the first detection target, the echo path is: O→C→O, that is, after the transmitted wave reaches the first detection target, the direct reflected echo returns.

[0082] The second echo path: When the vehicle-mounted millimeter-wave radar O cannot directly detect the first detection target, it means that the first detection target is located directly in front of the second area, resulting in the vehicle-mounted millimeter-wave radar O being unable to directly reach the first detection target. At this time, the echo path is: O→B→C→B→O, that is, the vehicle-mounted millimeter-wave radar O emits a detection wave at a preset frequency, which is reflected by the ground B to the first detection target C and then returns along the original path.

[0083] Based on this, the present invention discloses a method for detecting obstacles by a vehicle-mounted millimeter-wave radar, including the following steps:

[0084] S1: Construct an echo signal model according to the detection scenario and detection target of the vehicle-mounted millimeter-wave radar; the detection target includes the first detection target and the second detection target.

[0085] The specific content of S1 is:

[0086] S11: Define that the detection scenario of the vehicle-mounted millimeter-wave radar includes a first area and a second area; define the pedestrian beside the vehicle ahead as the first detection target, and the first detection target is located in the first area; define the vehicle ahead as the second detection target, and the second detection target is located in the second area;

[0087] S12: According to the detection path of the vehicle-mounted millimeter-wave radar, respectively obtain the path echo time delays of the first detection target and the second detection target, and construct an echo signal model according to the path echo time delays.

[0088] S2: Obtain the echo signal corresponding to the detection target according to the echo signal model;

[0089] Preferably, the process of constructing the echo signal model is as follows:

[0090] Let the expression of the linear frequency modulation signal emitted by the millimeter-wave radar be:

[0091] s(t) = A0exp(j2πf0t + jπμt 2 )u(t);

[0092]

[0093] Among them, f0 represents the carrier frequency, A0 represents the amplitude of the transmitted signal, μ = B / T is the chirp slope, The signal bandwidth is expressed as B, the pulse time is expressed as T, and u(t) is the rectangular function.

[0094] At this time, the radar echo signal can be divided into two combinations. One is that the echo contains τ path-1 and τ path-2 , and the other is that it contains τ path-1 and τ path-3 . Therefore, the two echo signal models are expressed as:

[0095] y1(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-2 ) + n(t);

[0096] y2(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-3 ) + n(t);

[0097] Among them, σ1 is the scattering coefficient of the second detection target, σ2 is the scattering coefficient of the first detection target, and n(t) represents the background noise.

[0098] Furthermore, based on the above radar detection echo path, let τ path-1 be the echo delay of the direct path for detecting the second detection target, τ path-2 be the delay of the ground first - reflection path for detecting the first detection target, and τ path-3 be the direct path for detecting the first detection target; then the path echo delays of the first detection target and the second detection target, the specific expressions are:

[0099]

[0100] Among them, c is the propagation speed of electromagnetic waves; OA is the direct path between the current vehicle - mounted millimeter - wave radar and the second detection target, OC is the direct path between the current vehicle - mounted millimeter - wave radar and the first detection target, and OB and BC are the indirect paths when the echo of the OC path passes through the ground point B as the intermediate reflection point when the first detection target is in front of the second detection target.

[0101] When using a vehicle - mounted millimeter - wave radar to detect the first detection target, since the radar moves with the vehicle, there are no stationary targets in the environment for the radar, and the traditional static clutter cancellation method cannot be used to eliminate the echo signal of the second detection target. Considering the problem that the echo of the second detection target cannot be eliminated, resulting in difficulties in detecting the first detection target, the present invention uses a signal separation method to separate the range image of the first detection target from the original echo range image:

[0102] S3: Separate the echo signal on the echo range image to obtain the first detection target range image X human , mainly including Fourier transform and signal separation, specifically:

[0103] S31: Perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range image X ∈ C M×N , de-mean the original range image X to obtain the de-mean matrix

[0104] For the original range image X ∈ C M×N De-mean it. Assume each radar signal has M periods and a total of N radar signals are received. Define the input vector as x n = [x 1,n L x M,n T , the original range image matrix is X = [x1 L x N , where x m,n is the nth range value in the mth period;

[0105] Define the mean vector of the original range image X ∈ C M×N as μ = [μ1 L μ M T , the de-meaning operation of the input vector is expressed as The original range image after de-meaning is expressed as

[0106] S32: Calculate the covariance matrix D of the de-mean matrix , and perform eigenvalue decomposition. Arrange the eigenvalues in descending order, and select the eigenvectors corresponding to the first k eigenvalues as the mapping matrix for dimensionality reduction. After dimensionality reduction, obtain the matrix X containing the principal component information of the second detection target front , and obtain the first detection target range image X human .

[0107] Calculate the covariance matrix of the de-mean matrix , and the formula is as follows:

[0108]

[0109] Perform eigenvalue decomposition on the above covariance matrix, and the formula is as follows:

[0110] D = U∑U -1 ;

[0111] After eigenvalue decomposition, Σ is a diagonal matrix, U is the matrix composed of the eigenvectors of matrix D, and the corresponding eigenvectors are U1, L, U M ​​。Then, the eigenvectors are arranged in descending order according to the corresponding eigenvalues. At this time, the eigenvectors corresponding to the first k eigenvalues are selected as the mapping matrix for dimensionality reduction, and the mapping matrix can be expressed as Then, the data matrix after dimensionality reduction can be expressed as:

[0112]

[0113] This is because the range image of the first detection target signal and the range image of the second detection target signal are mixed to form the original range image. Among them, the intensity of the second detection target signal is the largest, and it can be approximately considered that the original range image X mainly contains the range image of the second detection target. Therefore, the matrix X of the principal component information of the second detection target is obtained by dimensionality reduction front , and the separation process of the range images of the first detection target and the second detection target is essentially completed. The range image X of the first detection target human :

[0114] X human = X - X front .

[0115] Preferably, S3 also uses non-coherent superposition and constant false alarm rate detection to test the calculated range image. The specific steps are as follows:

[0116] Use the non-coherent superposition method to accumulate the amplitudes at the target in the range image. The non-coherent superposition process for the first detection target is expressed as:

[0117]

[0118] In this formula, j represents the number of periods, i is the range cell number, h is the frame index number, and |·| represents the absolute value operation.

[0119] After the amplitude of the first detection target is accumulated, the cell-averaging constant false alarm rate detection method (CA-CFAR) is used to detect the one-dimensional range image of the first detection target after accumulation. For the i-th range cell, the detection threshold can be expressed as:

[0120]

[0121] In this formula, P f represents the false alarm probability, and N r represents the number of reference cells.

[0122] After that, the range R of the first detection target is obtained by comparing the values of all range cells in the range image X human of the first detection target with the value of the detection threshold human .

[0123] S4: Use the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target.

[0124] For a uniform linear array model, the direction vector a(θ) of the antenna received signal is expressed as:

[0125] a(θ) = [1 e -jφ L e -j(K-1)φ T

[0126] φ = 2πd sinθ / λ

[0127] where d is the spacing between two adjacent antennas in the radar, φ is the phase difference between adjacent antennas, k is the number of antennas, θ is the incident angle of the radar signal, and λ is the wavelength of the radar signal.

[0128] After obtaining the distance of the first detection target, the minimum variance distortionless response method (MVDR) is used to calculate the azimuth angle of the target. The average output power P(θ) of the spatial filter is expressed as:

[0129]

[0130] where R = E{x human (j)x human H (j)}, and R represents the autocorrelation matrix of the first detection target range profile input matrix X human of.

[0131] Then, θ in the direction vector is changed in the [-π, π] angle interval to obtain the P MVDR (θ) change curve and perform spectral peak search. At this time, the angle corresponding to the peak point is the azimuth angle θ of the first detection target human .

[0132] Apply the MVDR angle measurement method to obtain the azimuth angle θ of the first detection target human After that, combined with the first detection target distance R human , calculate the two-dimensional position coordinates of the first detection target as:

[0133]

[0134] Preferably, the present invention also provides a vehicle-mounted millimeter-wave radar obstacle detection system, which mainly includes:

[0135] A radar module for detecting detection targets in a detection scene and obtaining radar reflected waves;

[0136] There is also a detection module for detecting obstacles in front of the vehicle. The detection module at least includes an analog unit and a calculation unit;

[0137] ​The simulation unit stores an echo signal model constructed based on the detection scenario and detection target of the vehicle-mounted millimeter-wave radar, and is used to obtain the echo signal corresponding to the detection target. Specifically: the detection scenario is defined to include a first area and a second area; a first detection target is defined to be located in the first area, and a second detection target is defined to be located in the second area; according to the detection path of the radar module, the path echo time delays of the first detection target and the second detection target are respectively obtained, and an echo signal model is constructed based on the path echo time delays.

[0138] The calculation unit is used to separate the echo signal on the echo range profile to obtain the range profile X of the first detection target human , and adopt the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target. Specifically: perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range profile X∈C M×N , de-mean the original range profile X to obtain the de-mean matrix X; calculate the covariance matrix D of the de-mean matrix X, and perform eigenvalue decomposition, arrange the eigenvalues in descending order, and select the eigenvectors corresponding to the first k eigenvalues as the mapping matrix for dimensionality reduction, and the dimensionality reduction obtains a matrix X containing the principal component information of the second detection target front , and calculate the range profile X of the first detection target according to the original range profile X and the matrix X of the principal component information of the second detection target front . human .

[0139] It further includes a display unit for presenting the position of the first detection target.

[0140] Preferably, the present invention further provides a radar, which is one of the vehicle-mounted millimeter-wave radars, is installed anywhere on the vehicle, is communicatively connected to the vehicle-mounted control center, and is used to implement a vehicle-mounted millimeter-wave radar obstacle detection method as described above.

[0141] Preferably, the present invention further provides a vehicle, which at least includes:

[0142] A vehicle-mounted millimeter-wave radar installed anywhere on the vehicle, and the vehicle-mounted millimeter-wave radar is used to obtain detection target information in the detection scenario and send it to a computer-readable storage medium at the vehicle-mounted MCU or SoC terminal.

[0143] Specifically, the vehicle of the present invention includes at least a vehicle body. The millimeter-wave radar can be arranged in front of and / or behind the vehicle body, or at other suitable positions. One or more millimeter-wave radars can be arranged on the vehicle body. Based on the millimeter-wave radar, the foregoing obstacle detection method can be implemented, so as to accurately and effectively determine the detected moving target, improve the safety and robustness of the entire ADAS, be able to more effectively protect against lateral shuttle obstacles (such as crossing pedestrians), and effectively avoid accidents caused by lateral moving objects during driving, thereby enhancing the user experience.

[0144] A computer program is stored on the computer-readable storage medium, and one or more processors are used to execute the program in the computer-readable storage medium to implement a vehicle-mounted millimeter-wave radar obstacle detection method as described above, and present the first detection target position calculated by the method through a display unit.

[0145] Embodiment 2:

[0146] Detect the first detection target based on the vehicle-mounted millimeter-wave radar obstacle detection method. The actual measurement scenario is as Figure 2 shown. The scenario includes a moving radar, a stationary vehicle in front, and a pedestrian target, where the pedestrian target moves around the stationary vehicle in front (not shown in the figure).

[0147] Use a frequency-modulated continuous-wave millimeter-wave radar to detect the first detection target and the second detection target. The radar frequency is 77 GHz and the bandwidth is 500 MHz. Here, it is assumed that the radar position is the center origin (0, 0) m and the height of the radar system is 0.35 m. At the initial stage of the experiment, the vehicle in front remains stationary and the radar is initially 15 m away from the vehicle in front. The remote control car moves back and forth within a range of 5 - 15 m from the vehicle in front, and the pedestrian beside the car moves around the stationary vehicle in front.

[0148] According to the processing steps of the present invention: First, obtain the original target range image through Fourier transform operation, as Figure 3 shown. The original target range image includes the range images of the vehicle in front and the pedestrian beside the car. Since the pedestrian beside the car moves around the vehicle in front, the two range images are fused, and it is difficult to distinguish the range image of the pedestrian beside the car from the original range image.

[0149] After that, separate the original range image to obtain the range images of the vehicle in front and the pedestrian beside the car, as Figure 4 shown, highlighting the trajectory of the pedestrian beside the car. Then, perform non-coherent summation and CFAR detection on the range image of the pedestrian beside the car to obtain the range value of the pedestrian beside the car.

[0150] Use the MVDR angle measurement algorithm to calculate the angle of the pedestrian beside the car, and finally obtain its positioning point from the position coordinates of the pedestrian beside the car.

[0151] The positioning result is as Figure 5As shown. Through comprehensive analysis, it is known that the position and movement of the pedestrian beside the vehicle are consistent with the movement law of the radar platform, and an accurate positioning result of the pedestrian beside the vehicle is obtained.

[0152] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely exemplary and are not intended to limit the scope of the present invention thereto. Those of ordinary skill in the art can make various changes and modifications therein without departing from the scope and spirit of the present invention. All such changes and modifications are intended to be included within the scope of the present invention as claimed in the appended claims.

[0153] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0154] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0155] Each component embodiment of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that a microprocessor or a digital signal processor (DSP) can be used in practice to implement some or all of the functions of some modules according to the embodiments of the present invention. The present invention can also be implemented as a device program (for example, a computer program and a computer program product) for executing a part or all of the methods described herein. Such a program implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0156] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0157] The above-described embodiments merely represent one implementation mode of the present invention, and the description thereof is relatively specific and detailed. However, it should not be construed as a limitation on the scope of the patent for the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent for the present invention shall be subject to the appended claims.

Claims

1. A method for detecting obstacles by an in-vehicle millimeter-wave radar, characterized in that, It includes the following steps: S1: Construct an echo signal model according to the detection scenario and detection targets of the vehicle-mounted millimeter-wave radar; the detection targets include a first detection target and a second detection target; S2: Obtain the echo signals corresponding to the detection targets according to the echo signal model; S3: Separate the echo signal on the echo range image to obtain the first detected target range image X human ; S4: Adopt the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target and determine the position of the first detection target; In S1, constructing the echo signal model according to the detection scenario and detection targets of the vehicle-mounted millimeter-wave radar is specifically as follows: S11: Define that the detection scenario of the vehicle-mounted millimeter-wave radar includes a first area and a second area; define that the first detection target is located in the first area and the second detection target is located in the second area; S12: Obtain the path echo time delays of the first detection target and the second detection target respectively according to the detection path of the vehicle-mounted millimeter-wave radar, and construct an echo signal model according to the path echo time delays; The path echo time delay is specifically: Among them, τ path-1 is the echo time delay of the direct path of the second detection target, τ path-2 is the time delay of the ground first reflection path of the first detection target, τ path-3 is the direct path of the first detection target; OA is the direct path between the current vehicle-mounted millimeter-wave radar and the second detection target, OC is the direct path between the current vehicle-mounted millimeter-wave radar and the first detection target, OB and BC are the indirect paths when the echo of the OC path passes through the ground point B as an intermediate reflection point when the first detection target is in front of the second detection target; c is the electromagnetic wave propagation speed; S3 is specifically: S31: Perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range image X ∈ C M×N , and de-mean the original range image X to obtain a de-mean matrix S32: Calculate the mean-removed matrix to obtain its covariance matrix D, perform eigenvalue decomposition, arrange the eigenvalues in descending order, select the eigenvectors corresponding to the top k eigenvalues as the mapping matrix for dimensionality reduction, and obtain the matrix X containing the principal component information of the second detection target after dimensionality reduction front , and calculate the distance image X of the first detection target based on the original distance image X and the matrix X containing the principal component information of the second detection target front ; human ; Among them, it is set that each radar signal has M cycles, and a total of N radar signals are received.

2. The vehicle-mounted millimeter-wave radar obstacle detection method according to claim 1, wherein, The echo signal model is specifically: Let the linear frequency modulation signal of the vehicle-mounted millimeter-wave radar be: s(t) = A0exp(j2πf0t + jπμt 2 )u(t); Construct an echo signal model, and the formula is: y1(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-2 ) + n(t); y2(t) = σ1s(t - τ path-1 ) + σ2s(t - τ path-3 ) + n(t); In the above formula, f0 is the carrier frequency, A0 is the amplitude of the transmitted signal, μ = B / T is the linear frequency modulation slope, B is the signal bandwidth, T is the pulse time; u(t) is the rectangular function; σ1 is the scattering coefficient of the second detection target, σ2 is the scattering coefficient of the first detection target, and n(t) represents the background noise.

3. A vehicle-mounted millimeter-wave radar obstacle detection system for implementing the vehicle-mounted millimeter-wave radar obstacle detection method according to any one of claims 1-2, characterized in that, The system at least includes: A radar module for detecting detection targets in the detection scenario and obtaining radar reflected waves; A detection module for detecting obstacles in front of the vehicle, and the detection module at least includes an analog unit and a calculation unit; The analog unit stores an echo signal model constructed according to the detection scenario and detection targets of the vehicle-mounted millimeter-wave radar, and is used to obtain the echo signals corresponding to the detection targets; The calculation unit is configured to separate the echo signal on the echo range image to obtain the first detection target range image X human , and adopt the MVDR angle measurement method to obtain the two-dimensional position coordinates of the first detection target, and determine the position of the first detection target; A display unit for presenting the position of the first detection target.

4. An in-vehicle millimeter-wave radar obstacle detection system according to claim 3, characterized in that, The analog unit further includes: Define that the detection scenario includes a first area and a second area; define that the first detection target is located in the first area and the second detection target is located in the second area; Obtain the path echo time delays of the first detection target and the second detection target respectively according to the detection path of the radar module, and construct an echo signal model according to the path echo time delays.

5. The vehicle-mounted millimeter-wave radar obstacle detection system according to claim 4, characterized in that, The calculation unit further includes: Perform fast-time dimension Fourier transform processing on the echo signal to obtain the original range image \(X\in\mathbb{C}\). M×N , and de-mean the original range image \(X\) to obtain the de-mean matrix Calculate the mean-removed matrix to obtain the covariance matrix D, perform eigenvalue decomposition, arrange the eigenvalues in descending order, select the eigenvectors corresponding to the top k eigenvalues as the mapping matrix for dimensionality reduction, and obtain the matrix X containing the principal component information of the second detection target after dimensionality reduction front , and calculate the range image X of the first detection target based on the original range image X and the matrix X containing the principal component information of the second detection target front ; human ; Among them, it is set that each radar signal has M cycles, and a total of N radar signals are received.

6. A radar, which is one of the vehicle-mounted millimeter-wave radars and is installed anywhere on the vehicle, is characterized in that, The radar is communicatively connected to the vehicle-mounted control center and is used to implement a method for detecting obstacles by a vehicle-mounted millimeter-wave radar as described in any one of claims 1-2.

7. An automobile, characterized in that, It at least includes: A radar installed at any location of the vehicle, the radar is one of the vehicle-mounted millimeter-wave radars, and the vehicle-mounted millimeter-wave radar is used to obtain detection target information in the detection scenario and send it to a computer-readable storage medium at the vehicle-mounted MCU or SoC end; A computer program is stored on the computer-readable storage medium; And one or more processors controlled by the vehicle-mounted MCU or SoC end, and the processors are used to execute the program in the computer-readable storage medium to implement a method for detecting obstacles by a vehicle-mounted millimeter-wave radar as described in any one of claims 1-2, and present the position of the first detection target calculated by the method through the display unit.

Images