DOA estimation method of vehicle-mounted millimeter wave radar based on spatio-temporal joint estimation
Through the method based on space-time joint estimation, the time domain data model and pseudo-data matrix are constructed, eigenvalue decomposition and generalized least squares solution are performed, and the problem of insufficient resolution and degree of freedom of DOA estimation and high computational complexity of vehicle-mounted millimeter wave radar in complex environments is solved, achieving a balance of high precision and real-time.
Patent Information
- Application Number
- CN202411932861.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-04-29
AI Technical Summary
The existing vehicle-mounted millimeter-wave radar has limited DOA estimation resolution and degree of freedom in complex road environments, and the calculation complexity is high, making it difficult to take into account real-time and high precision.
Using a method based on space-time joint estimation, the time domain data model received by the array is constructed, the time average vector is obtained, the pseudo-data matrix is constructed, the covariance data matrix is calculated, and the eigenvalue decomposition is performed. The equation is solved using the generalized least squares method to estimate the DOA.
It improves the DOA estimation accuracy, reduces the computational amount, and achieves a balance of high accuracy and real-time in complex environments.
Smart Images
Figure CN119902176A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of signal processing, and in particular to a DOA estimation method of a vehicle-mounted millimeter-wave radar based on time-space joint estimation. Background Art
[0002] With the development of society and the advancement of technology, the number of cars has increased dramatically, and the frequency of traffic accidents has also increased. This has prompted researchers to develop advanced driver assistance systems (ADAS) to improve the safety and convenience of vehicle driving. The millimeter-wave radar in the ADAS system has become one of the key components due to its all-weather working ability, low cost, and small size. Millimeter-wave radar can not only work normally in severe weather conditions such as rain, snow, fog, and sandstorms, but also has a lower price than lidar, with moderate cost, which is more conducive to commercial applications. In addition, millimeter-wave radar also has high ranging and speed measurement performance, with a maximum measurement distance of more than 200 meters. The 4GHz bandwidth can increase the distance resolution to within 4 centimeters, which can better meet the needs of vehicle-mounted application scenarios. However, vehicle-mounted millimeter-wave radar still has shortcomings in angular resolution, especially in complex road environments, and high-precision and high-resolution DOA (Direction of Arrival) estimation algorithms are required to improve the accuracy of target detection and recognition.
[0003] The super-resolution DOA estimation algorithm can further improve the angular resolution without changing the size of the radar array by breaking through the limitation of the Rayleigh Limit. Common super-resolution DOA estimation algorithms include methods based on subspace decomposition, iterative adaptive methods, and methods based on compressed sensing. Among them, the MUSIC (Multiple Signal Classification) algorithm is one of the most classic super-resolution DOA estimation algorithms. It analyzes the eigenvectors of the signal covariance matrix, establishes the orthogonality of the signal subspace and the noise subspace, thereby constructing the spatial spectrum and obtaining the signal wave direction through the spectrum peak search. The ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques) algorithm utilizes the rotational invariance of the signal subspace between subarrays to avoid the spectrum peak search, but it needs to perform two eigenvalue decompositions. The IAA (Iterative Adaptive Approach) algorithm gradually approaches the true directional diagram by iteratively updating the estimation of the signal subspace and the noise subspace, but the amount of calculation is large. In addition, the OMP (Orthogonal Matching Pursuit) algorithm based on compressed sensing gradually constructs a sparse representation of the signal by successively selecting the atoms most relevant to the current residual, but the resolution is low. The SBL (Sparse Bayesian Learning) algorithm can effectively handle the DOA estimation problem in non-ideal environments such as low signal-to-noise ratio and few snapshots by introducing sparse priors, but the amount of calculation is large. Although the existing technology has made some progress in the direction of arrival (DOA) estimation algorithm of millimeter wave radar, it still has the following shortcomings when dealing with complex road environments and the needs of intelligent autonomous driving: 1) Limited resolution and degrees of freedom: Sparse Planar Array (SPA): Although it can significantly reduce the number of array elements, the resolution is reduced due to the relatively small array aperture. This is prone to signal indistinguishability in complex multipath propagation environments. Sparse L-shaped Array (SLA): Although it can provide a larger array aperture, "holes" will appear in its differential virtual array, making the virtual array elements discontinuous, thereby reducing the system's degrees of freedom; 2) High computational complexity: MUSIC algorithm: Although it has high resolution and accuracy, one of its core steps is to estimate DOA through spectral peak search, which is very computationally intensive in two-dimensional DOA estimation. Especially in vehicle-mounted millimeter wave radar applications, the vehicle is in a high-speed moving state, which requires extremely high real-time processing capabilities, and high computational complexity becomes a major bottleneck. ESPRIT algorithm: Although it avoids spectral peak search and has relatively low computational complexity, its estimation accuracy is lower than that of the MUSIC algorithm, especially when processing multiple signal sources, the loss of accuracy is more obvious; 3) It is difficult to balance real-time and high precision: In intelligent autonomous driving and assisted driving technology, the real-time perception capability and high precision requirements of vehicle-mounted millimeter-wave radar are very high. However, existing DOA estimation algorithms can often only meet one of the requirements, and it is difficult to achieve an effective balance between the two in the same system. For example, high-precision algorithms usually have high computational complexity and are difficult to achieve real-time processing; while low-complexity algorithms are difficult to achieve the required high precision requirements. Summary of the invention
[0004] The purpose of the present invention is to overcome at least one of the problems existing in the prior art and to provide a DOA estimation method for a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation. The present invention can improve the DOA estimation accuracy and effectively reduce the amount of calculation.
[0005] As another object of the present invention, a system adapted to the method based on the aforementioned object is provided.
[0006] As another object of the present invention, a non-volatile storage medium suitable for storing a computer program implemented according to the method described is provided.
[0007] In order to achieve the above object, the present invention provides a DOA estimation method for a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation, the method comprising the following steps: Constructing a time domain data model received by the array; Acquire a time average vector based on the time domain data model; constructing a pseudo data matrix according to the time average vector, and processing the pseudo data matrix to obtain a processing result; constructing a covariance data matrix according to the processing results; Performing eigenvalue decomposition on the covariance data matrix, and constructing an equation according to the decomposition result; The equation is solved to obtain a real-valued generalized least squares solution, and the DOA is estimated based on the real-valued generalized least squares solution.
[0008] Furthermore, the constructing of the time domain data model received by the array specifically includes: the array is The nested L array consists of sensors, and its distribution pattern is as follows:
[0009]
[0010] Where d = , is the wavelength of the incident signal, if The horizontal angle is , the depression angle is The signals are incident into the nested L array, where the kth signal is represented as , and Represent deterministic complex amplitude and small frequency deviation respectively, then the subarray set on the horizontal axis The data received at time t is as follows:
[0011] in, , for noise; Subarray collection on the vertical axis The data received at time t is as follows:
[0012]
[0013] in, and Corresponding to the subarray sets and the subarray set The steering vector of the kth signal, For noise.
[0014] Further, the acquiring the time average vector based on the time domain data model includes: (1) From the subarray set The mth array element and the subarray set The data received by the nth array element is sampled T times, and the time function is obtained as follows:
[0015] Among them, x m is the signal received by the mth antenna in x(t) at time (t+τ / 2), y * n (t+τ / 2) is the conjugate of the signal received by the nth antenna in y(t) at time (t-τ / 2), , 0, ; (2) Take the subarray set The nth array element is 1, and the time average vector of n=1 is obtained, as follows:
[0016] in ; (3) Repeat step (2) n times to obtain the final time average vector , as follows: .
[0017] Further, constructing a pseudo data matrix according to the time average vector, and performing processing according to the pseudo data matrix to obtain a processing result, includes: (1) Construct a pseudo data matrix based on the time average vector , as follows:
[0018] in, , , is the pseudo sampling period, is the number of pseudo snapshots; (2) Based on the pseudo data matrix Calculate the virtual uniform plane array , as follows:
[0019] (3) For the virtual uniform planar array Perform vectorization processing to obtain the virtual uniform plane array The corresponding vector :The specific calculation method is as follows:
[0020] in, ; (4) Remove the vector Redundant part, and sort the remaining vectors to obtain the sorted vector , the vector The processing results are as follows:
[0021] in, , for A column vector of 1s.
[0022] Further, constructing a covariance data matrix according to the processing results includes: (1) The processing result, i.e., the vector Restore to Matrix , as follows:
[0023] (2) Use the two-dimensional space smoothing method to transform the matrix Constructed as The full-order covariance data matrix , as follows:
[0024] in, , , is the matrix No. arrive Line and arrive A matrix consisting of columns.
[0025] Furthermore, performing eigenvalue decomposition on the covariance data matrix and constructing an equation according to the decomposition result specifically includes: (1) Perform eigenvalue decomposition on the covariance data matrix:
[0026] is the eigenvalue decomposition operation, is a diagonal matrix consisting of the largest K eigenvalues, is the corresponding eigenvector; (2) According to the feature vector Construct an equation, the equation comprising a first equation and a second equation, wherein The first equation is as follows:
[0027] The second equation is as follows:
[0028] in, , , for The 0 vector, ,
[0029]
[0030] The selection matrix The dimension is , When q is an odd number:
[0031]
[0032] When q is an even number:
[0033] .
[0034] Further, solving the equation to obtain a real-valued generalized least squares solution, and estimating the DOA based on the real-valued generalized least squares solution specifically includes: (1) Solve the first equation and the second equation to obtain a first real-valued generalized least squares solution and a second real-valued generalized least squares solution, wherein the first real-valued generalized least squares solution is:
[0035] The second real-valued generalized least squares solution is:
[0036] (2) Restore the first real-valued generalized least squares solution and the second real-valued generalized least squares solution into matrices, as follows:
[0037]
[0038] (3). According to the matrix and matrix Estimate DOA.
[0039] Furthermore, according to the matrix and matrix Estimate DOA as follows:
[0040]
[0041]
[0042] in, and It contains the phase information of the horizontal angle and elevation angle respectively. According to the construction form of the model, the DOA can be solved to obtain the final DOA estimation value.
[0043] In order to achieve another object of the present invention, a vehicle-mounted millimeter wave radar based on time-space joint estimation is provided. A DOA estimation system, the system is based on the DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation, comprising: The first building module is used to build a time domain data model received by the array; Acquisition module: used for acquiring a time average vector based on the time domain data model; The second construction module is used to construct a pseudo data matrix according to the time average vector, and process the pseudo data matrix to obtain a processing result; The third construction module is used to construct a covariance data matrix according to the processing results; The fourth construction module is used to perform eigenvalue decomposition on the covariance data matrix and construct an equation according to the decomposition result; Estimation module: used for solving the equation, obtaining a real-valued generalized least squares solution, and estimating DOA based on the real-valued generalized least squares solution.
[0044] In order to achieve another object of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer-stored program is executed by a processor, the DOA estimation method of a vehicle-mounted millimeter-wave radar based on joint time-space estimation is implemented.
[0045] Compared with the prior art, the present invention has the following beneficial effects: The present invention constructs a pseudo data matrix with a time average vector, and the pseudo data matrix contains time information and space information, which effectively increases the number of estimated signal sources and expands the virtual aperture, thereby improving the DOA estimation accuracy; it also constructs a time domain data model of array reception, in which the array is a sparsely distributed array element, so as to achieve the expansion of the number of estimated signal sources on the basis of reducing the number of receiving array elements, and can also reduce the amount of calculation in calculating the covariance matrix, thereby reducing the amount of DOA estimation calculation; and introduces the generalized least squares method to solve the equation, thereby further improving the DOA estimation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 It is a flow chart of a DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to Embodiment 1 of the present invention; Figure 2 It is a block diagram of a DOA estimation system of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to Embodiment 2 of the present invention; Figure 3 is a nested L array diagram consisting of a subarray x and a subarray y according to Embodiment 1 of the present invention; Figure 4 is a schematic diagram of a virtual sparse planar array according to Embodiment 1 of the present invention (with For example) Figure 5 is a schematic diagram of a virtual uniform planar array according to Embodiment 1 of the present invention (with for example). DETAILED DESCRIPTION
[0047] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.
[0048] Example 1 like Figure 1 As shown, a DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation in a preferred embodiment of an embodiment of the present invention includes: S1: construct the time domain data model received by the array; In a feasible embodiment, S1 specifically includes: The array is The nested L array consists of sensors. For the L array diagram, see Figure 3 , and its distribution law is as follows:
[0049]
[0050] In this embodiment, it is assumed that M=N, and as an example, a nested linear array is used as and :
[0051]
[0052] Where d = , is the wavelength of the incident signal, if The horizontal angle is , the depression angle is The signals are incident into the nested L array, where the kth signal is represented as , and Represent deterministic complex amplitude and small frequency deviation respectively, then the subarray set on the horizontal axis The data received at time t is as follows:
[0053] in, , is noise, specifically additive Gaussian white noise with zero mean; Subarray collection on the vertical axis The data received at time t is as follows:
[0054]
[0055] in, and Corresponding to the subarray sets and the subarray set The steering vector of the kth signal, is the noise, and Similarly, it is also additive Gaussian white noise with zero mean.
[0056] S2: Obtaining a time average vector based on the time domain data model; In a feasible embodiment, S2 specifically includes: (1) From the subarray set The mth array element and the subarray set The data received by the nth array element is sampled T times, and the time function is obtained as follows:
[0057] Among them, x m is the signal received by the mth antenna in x(t) at time (t+τ / 2), y * n (t+τ / 2) is the conjugate of the signal received by the nth antenna in y(t) at time (t-τ / 2), , 0, ; (2) Take the subarray set The nth array element is 1, and the time average vector of n=1 is obtained, as follows:
[0058] in ; (3) Repeat step (2) n times to obtain the final time average vector , as follows: .
[0059] Apparently The difference array of the generalized virtual plane array has the same form (such as Figure 4 ),for The column vector of 1 is the output data of the virtual plane array.
[0060] S3: constructing a pseudo data matrix according to the time average vector, and processing the pseudo data matrix to obtain a processing result; In a feasible embodiment, S3 specifically includes: (1) Construct a pseudo data matrix based on the time average vector , as follows:
[0061] in, , , is the pseudo sampling period, is the number of pseudo snapshots; (2) Based on the pseudo data matrix Calculate the virtual uniform plane array , as follows:
[0062] (3) For the virtual uniform planar array Perform vectorization processing to obtain the virtual uniform plane array The corresponding vector :The specific calculation method is as follows:
[0063] in, ; (4) Remove the vector Redundant part, and sort the remaining vectors to obtain the sorted vector , the vector The processing results are as follows:
[0064] in, , for A column vector of 1s. Expand to a Virtual uniform planar arrays (such as Figure 5 ), for A column vector of 1s. can be considered as coming from Receive data from a fully correlated source.
[0065] S4: constructing a covariance data matrix according to the processing results; In a feasible embodiment, S4 specifically includes: (1) The processing result, i.e., the vector Restore to Matrix , as follows:
[0066] (2) Use the two-dimensional space smoothing method to transform the matrix Constructed as The full-order covariance data matrix , as follows:
[0067] in, , , is the matrix No. arrive Line and arrive A matrix consisting of columns.
[0068] S5: performing eigenvalue decomposition on the covariance data matrix, and constructing an equation according to the decomposition result; In a feasible embodiment, S5 specifically includes: (1) Perform eigenvalue decomposition on the covariance data matrix:
[0069] is the eigenvalue decomposition operation, is a diagonal matrix consisting of the largest K eigenvalues, is the corresponding eigenvector; (2) According to the feature vector Construct an equation, the equation comprising a first equation and a second equation, wherein The first equation is as follows:
[0070] The second equation is as follows:
[0071] in, , , for The 0 vector, ,
[0072]
[0073] The selection matrix The dimension is , When q is an odd number:
[0074]
[0075] When q is even:
[0076] .
[0077] S6: Solve the equation to obtain a real-valued generalized least squares solution, and estimate the DOA based on the real-valued generalized least squares solution.
[0078] In a feasible embodiment, S6 specifically includes: (1) Solve the first equation and the second equation to obtain a first real-valued generalized least squares solution and a second real-valued generalized least squares solution, wherein the first real-valued generalized least squares solution is:
[0079] The second real-valued generalized least squares solution is:
[0080] (2) Restore the first real-valued generalized least squares solution and the second real-valued generalized least squares solution into matrices, as follows:
[0081]
[0082] (3). According to the matrix and matrix Estimate DOA as follows:
[0083]
[0084]
[0085] in, and It contains the phase information of the horizontal angle and elevation angle respectively. According to the construction form of the model, the DOA can be solved to obtain the final DOA estimation value.
[0086] This embodiment constructs a pseudo data matrix with a time-averaged vector, which contains time information and spatial information, effectively increasing the number of estimated signal sources and expanding the virtual aperture, thereby improving the DOA estimation accuracy. This embodiment also constructs a time domain data model of array reception, in which the array is a sparsely distributed array element, thereby reducing the number of receiving array elements and expanding the number of estimated signal sources, and can also reduce the amount of calculation in calculating the covariance matrix, thereby reducing the amount of DOA estimation calculations. The generalized least squares method is introduced to solve the equation, thereby further improving the DOA estimation accuracy.
[0087] Example 2 like Figure 2 As shown, a DOA estimation system of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to an embodiment of the present invention includes: The first building module is used to build a time domain data model received by the array; Acquisition module: used for acquiring a time average vector based on the time domain data model; The second construction module is used to construct a pseudo data matrix according to the time average vector, and process the pseudo data matrix to obtain a processing result; The third construction module is used to construct a covariance data matrix according to the processing results; The fourth construction module is used to perform eigenvalue decomposition on the covariance data matrix and construct an equation according to the decomposition result; Estimation module: used for solving the equation, obtaining a real-valued generalized least squares solution, and estimating DOA based on the real-valued generalized least squares solution.
[0088] The system proposed in this embodiment is based on a DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation in Embodiment 1. Therefore, the optional items proposed in Embodiment 1 are still applicable in this embodiment and will not be described in detail.
[0089] Example 3 An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer-stored program is executed by a processor, the DOA estimation method of a vehicle-mounted millimeter-wave radar based on joint spatiotemporal estimation is implemented.
[0090] In summary, the embodiments of the present invention provide a DOA estimation method, system and storage medium for a vehicle-mounted millimeter-wave radar based on joint time-space estimation. The method constructs a pseudo data matrix with a time-averaged vector, which contains time information and space information, effectively increases the number of estimated signal sources, and expands the virtual aperture, thereby improving the DOA estimation accuracy. The method also constructs a time domain data model received by the array, in which the array is a sparsely distributed array element, thereby reducing the number of receiving array elements and expanding the number of estimated signal sources. The method can also reduce the amount of calculation in calculating the covariance matrix, thereby reducing the amount of DOA estimation calculation. The generalized least squares method is introduced to solve the equation, thereby further improving the accuracy of DOA estimation.
[0091] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention. These improvements and substitutions should also be regarded as the scope of protection of the present invention.
Claims
1. A DOA estimation method for vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation, characterized in that: The method comprises the following steps: Constructing a time domain data model received by the array; Acquire a time average vector based on the time domain data model; constructing a pseudo data matrix according to the time average vector, and processing the pseudo data matrix to obtain a processing result; constructing a covariance data matrix according to the processing results; Performing eigenvalue decomposition on the covariance data matrix, and constructing an equation according to the decomposition result; The equation is solved to obtain a real-valued generalized least squares solution, and the DOA is estimated based on the real-valued generalized least squares solution.
2. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 1, characterized in that: The constructing of a time domain data model received by the array specifically includes: the array is The nested L array consists of sensors, and its distribution pattern is as follows: Where d = , is the wavelength of the incident signal, if The horizontal angle is , the depression angle is The signals are incident into the nested L array, where the kth signal is represented as , and Represent deterministic complex amplitude and small frequency deviation respectively, then the subarray set on the horizontal axis The data received at time t is as follows: in, , for noise; Subarray collection on the vertical axis The data received at time t is as follows: in, and Corresponding to the subarray sets and the subarray set The steering vector of the kth signal, For noise.
3. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 1, characterized in that: The acquiring the time average vector based on the time domain data model comprises: (1) From the subarray set The mth array element and the subarray set The data received by the nth array element is sampled T times, and the time function is obtained as follows: Among them, x m is the signal received by the mth antenna in x(t) at time (t+τ / 2), y * n (t+τ / 2) is the conjugate of the signal received by the nth antenna in y(t) at time (t-τ / 2), , 0, ; (2) Take the subarray set The nth array element is 1, and the time average vector of n=1 is obtained, as follows: in ; (3) Repeat step (2) n times to obtain the final time average vector , as follows: 。 4. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 1, characterized in that: The constructing a pseudo data matrix according to the time average vector, and performing processing according to the pseudo data matrix to obtain a processing result, includes: (1) Construct a pseudo data matrix based on the time average vector , as follows: in, , , is the pseudo sampling period, is the number of pseudo snapshots; (2) Based on the pseudo data matrix Calculate the virtual uniform plane array , as follows: (3) For the virtual uniform planar array Perform vectorization processing to obtain the virtual uniform plane array The corresponding vector :The specific calculation method is as follows: in, ; (4) Remove the vector Redundant part, and sort the remaining vectors to obtain the sorted vector , the vector The processing results are as follows: in, , for A column vector of 1.
5. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 1, characterized in that: Constructing a covariance data matrix according to the processing results includes: (1) The processing result, i.e., the vector Restore to Matrix , as follows: (2) Use the two-dimensional space smoothing method to transform the matrix Constructed as The full-order covariance data matrix , as follows: in, , , is the matrix No. arrive Line and arrive A matrix consisting of columns.
6. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 5, characterized in that: The performing eigenvalue decomposition on the covariance data matrix and constructing an equation according to the decomposition result specifically includes: (1) Perform eigenvalue decomposition on the covariance data matrix: is the eigenvalue decomposition operation, is a diagonal matrix consisting of the largest K eigenvalues, is the corresponding eigenvector; (2) According to the feature vector Construct an equation, the equation comprising a first equation and a second equation, wherein The first equation is as follows: The second equation is as follows: in, , , for The 0 vector, , The selection matrix The dimension is , When q is an odd number: When q is even: 。 7. The DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation according to claim 6, characterized in that: Solving the equation to obtain a real-valued generalized least squares solution, and estimating the DOA based on the real-valued generalized least squares solution, specifically includes: (1) Solve the first equation and the second equation to obtain a first real-valued generalized least squares solution and a second real-valued generalized least squares solution, wherein the first real-valued generalized least squares solution is: The second real-valued generalized least squares solution is: (2) Restore the first real-valued generalized least squares solution and the second real-valued generalized least squares solution into matrices, as follows: (3). According to the matrix and matrix Estimate DOA.
8. The DOA estimation method of the vehicle-mounted millimeter-wave radar based on the spatiotemporal joint estimation according to claim 7 The method is characterized in that The matrix and matrix Estimate DOA as follows: in, and It contains the phase information of the horizontal angle and elevation angle respectively. According to the construction form of the model, the DOA can be solved to obtain the final DOA estimation value.
9. A DOA estimation system for vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation, the system is based on A DOA estimation method for a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation as claimed in any one of claims 1 to 8 Methods include: The first building module is used to build a time domain data model received by the array; Acquisition module: used for acquiring a time average vector based on the time domain data model; The second construction module is used to construct a pseudo data matrix according to the time average vector, and process the pseudo data matrix to obtain a processing result; The third construction module is used to construct a covariance data matrix according to the processing results; The fourth construction module is used to perform eigenvalue decomposition on the covariance data matrix and construct an equation according to the decomposition result; Estimation module: used for solving the equation, obtaining a real-valued generalized least squares solution, and estimating DOA based on the real-valued generalized least squares solution.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: Said When the computer stored program is executed by a processor, the DOA estimation method of a vehicle-mounted millimeter-wave radar based on spatiotemporal joint estimation as described in any one of claims 1 to 8 is implemented.