Radar data processing for vehicle ego-motion estimation

Abstract

Radar data processing for vehicle ego-motion estimation. A method for determining ego-motion information of a vehicle including a radar sensor having a plurality of antenna elements is provided, the method comprising the following steps: acquiring (S410) motion spectrum data including a plurality of data elements, each data element being calculated for a corresponding Doppler bin index in a plurality of Doppler bin indices and for a corresponding spatial bin index in a plurality of spatial bin indices, each spatial bin index indicating a corresponding angle of arrival of a radar return signal at the radar sensor; and determining (S420) ego-motion information by solving a system of motion equations including a motion equation generated using the motion spectrum data, and the motion equation involving a corresponding value indicating a radial velocity, a corresponding value indicating an angular displacement, and a variable indicating a velocity of the vehicle.

Description

Technical Field

[0001] Example aspects herein relate generally to the field of radar data processing, and more particularly to methods of determining ego-motion information of a vehicle having a radar sensor using motion spectrum data derived from radar data that has been generated by the radar sensor. Background Art

[0002] Many perception algorithms employed in the automotive industry require information about the vehicle's ego-motion in order to perceive the vehicle's environment. Therefore, the requirement for a stable ego-motion estimation algorithm is an important part of advanced driver assistance systems (ADAS) and autonomous driving applications. The vehicle's ego-motion can be measured by an inertial measurement unit (IMU) or estimated using measurements acquired by other types of sensors, such as cameras or radar. However, those types of measurements often suffer from inherent errors when estimating ego-motion as well as errors that arise from using outlier measurements.

[0003] In the article "Instantaneous ego-motion estimation using multiple Doppler radars" published by Kellner D. et al. at the 2014 IEEE International Conference on Robotics and Automation (ICRA), radar detection is used to estimate the self-motion of a moving vehicle. The radar detection in this article is determined by first generating a range Doppler response map and additionally applying an angle-finding algorithm to the beam vector along the antenna dimension to obtain the angle information of the detected target. The radial velocity and angle information obtained for the detected target is then used in a regression analysis to estimate the vehicle's self-motion information. Summary of the invention

[0004] One problem with existing solutions for estimating ego-motion information using radar detections is the highly complex processing that must first perform detections in the range-Doppler domain to identify cells with high amplitude detection values ​​in the range-domain map, and then additionally perform a beamforming FFT on the beam vectors corresponding to the identified cells to obtain angle information in the three-dimensional range-Doppler angle domain. Moreover, when only a few detections are available, the estimation of ego-motion parameters cannot be performed accurately.

[0005] In view of the above situation, according to a first exemplary aspect of the present invention, a method for determining self-motion information of a vehicle (30) including a radar sensor having multiple antenna elements is designed, the method comprising the following steps: obtaining a motion spectrum data set based on radar data generated by the radar sensor, the motion spectrum data set comprising a plurality of data elements, each of the data elements being calculated for a corresponding Doppler bin index in a plurality of Doppler bin indexes and for a corresponding spatial bin index in a plurality of spatial bin indexes, each spatial bin index indicating a corresponding arrival angle of a radar echo signal relative to an axis of the radar sensor. The method further comprises the following steps: determining the self-motion information by solving a set of motion equations including a plurality of motion equations generated using the motion spectrum data set, each of the motion equations involving: a corresponding value indicating a radial velocity of a stationary object relative to the radar sensor, a corresponding value indicating an angular displacement of a stationary object relative to the axis of the radar sensor, and a variable indicating a velocity of the vehicle. The value indicating the radial velocity in the set of motion equations may be pre-calculated based on the Doppler bin index and independent of the radar data, and the value indicating the angular position in the set of motion equations may be pre-calculated based on the spatial bin index and independent of the radar data.

[0006] The individual equations of motion can have the following form:

[0007] d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x +sin(θ+θ M )v y

[0008] Where d indicates the radial velocity of the stationary object, θ indicates the angular displacement of the stationary object relative to the axis of the radar sensor, and θ M Indicates the installation angle of the radar sensor, which is the angle between the axis of the radar sensor and the axis of the vehicle coordinate system of the vehicle, l x indicates the installation position of the radar sensor along the X-axis of the vehicle coordinate system, and l y Indicates the installation position of the radar sensor along the Y axis of the vehicle coordinate system, v x indicates the X component of the vehicle's velocity, and v y indicates the Y component of the vehicle's velocity, ω indicates the vehicle's yaw rate, and the ego-motion information includes v x 、v y and the value of ω.

[0009] In the following embodiment, K radar sensors are installed at K different positions of the vehicle, each radar sensor has a plurality of antenna elements, wherein K is an integer greater than or equal to 2, and wherein the corresponding motion spectrum data set is obtained for each radar sensor in the K radar sensors based on radar data generated by the radar sensor, and the step of determining the self-motion information may include: estimating v by solving the following set of motion equations x 、v y And the value of ω:

[0010] D=H·E

[0011] in,

[0012] In the above system of motion equations, d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k,1 ,d k,2 ...,d k,F Indicates F radial velocity values ​​corresponding to F respective Doppler bin indices in a k th motion spectrum data set acquired for the k th radar sensor.

[0013] Moreover, M k (k=1,2,....,K) is an F×U dimensional matrix containing the data elements in the kth motion spectrum data set, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1,2,....,K, respectively, where M k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset, where M k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum data set,

[0014]

[0015] Among them, θ k,1 ,θ k,2 ...,θ k,U (u=1, 2, ...., U) indicates U angle values ​​corresponding to U spatial bin indices in the k-th motion spectrum dataset, respectively, and wherein, indicating a mounting angle of the kth radar sensor, the mounting angle being an angle between an axis of the kth radar sensor and an axis of a vehicle coordinate system of the vehicle, and

[0016] is the model matrix of the kth radar sensor, where indicates the installation position of the kth radar sensor along the X-axis of the vehicle coordinate system, and Indicates the installation position of the kth radar sensor along the Y-axis of the vehicle coordinate system.

[0017] In an alternative embodiment, the respective equations of motion may be given by:

[0018]

[0019] Where d indicates the radial velocity of the stationary object, θ indicates the angular displacement of the stationary object relative to the axis of the radar sensor, and θ M indicates the angle between the axis of the radar sensor and the axis of the vehicle coordinate system of the vehicle, Indicates the X component of the radar sensor's velocity, Indicates the Y component of the velocity of the radar sensor, and the ego-motion information includes a set value when the yaw rate of the vehicle is zero and Among them, v x indicates the X component of the vehicle's velocity, and v y Indicates the Y component of the vehicle's velocity.

[0020] In this alternative embodiment, when K radar sensors are installed at K different positions of the vehicle, each radar sensor has a plurality of antenna elements, wherein K is an integer greater than or equal to 1, and wherein the corresponding motion spectrum data set is obtained for each radar sensor among the K radar sensors based on radar data generated by the radar sensor, the ego-motion information can be estimated by and Values ​​and settings and To determine, where, the estimated and The steps to find the value of include solving the following system of motion equations:

[0021] D=H·E

[0022] in, Among them, d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k,1 ,d k,2 ...,d k,Findicates F radial velocity values ​​corresponding to F corresponding Doppler bin indices in a k-th motion spectrum data set acquired for the k-th radar sensor,

[0023] M k (k=1,2,....,K) is an F×U dimensional matrix containing the data elements in the kth motion spectrum data set, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1,2,....,K, respectively, where M k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset, where M k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum data set,

[0024]

[0025] And, θ k,1 ,θ k,2 ...,θ k,U , u=1,2,....,U indicates U angle values ​​corresponding to U spatial bin indices in the k-th motion spectrum data set, respectively, where, Indicates an installation angle of a kth radar sensor, the installation angle being an angle between an axis of the kth radar sensor and an axis of a vehicle coordinate system of the vehicle.

[0026] In another alternative embodiment, the respective equations of motion may be given by:

[0027] d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x

[0028] Where d indicates the radial velocity of the stationary object, θ indicates the angular displacement of the stationary object relative to the axis of the radar sensor, and θ M Indicates the angle between the axis of the radar sensor and the axis of the vehicle coordinate system of the vehicle, l x indicates the installation position of the radar sensor along the X-axis of the vehicle coordinate system, and l y indicates the installation position of the radar sensor along the Y axis of the vehicle coordinate system, ω indicates the yaw rate of the vehicle, v x is the X component of the vehicle’s velocity. The ego-motion information includes the values ​​ω and v x .

[0029] In another alternative embodiment, when K radar sensors are installed at K different positions of the vehicle, each radar sensor has multiple antenna elements, wherein K is an integer greater than or equal to 1, wherein the corresponding motion spectrum data set is obtained for each radar sensor in the K radar sensors based on radar data generated by the radar sensor, and the self-motion information can be estimated by solving the following set of motion equations to obtain values ​​ω and v x To determine:

[0030] D=H·E

[0031] in,

[0032] d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k,1 ,d k,2 ...,d k,F indicates F radial velocity values ​​corresponding to F corresponding Doppler bin indices in a k-th motion spectrum dataset generated for the k-th radar sensor,

[0033] M k (k=1,2,....,K) is an F×U dimensional matrix containing the data elements in the kth motion spectrum data set, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1,2,....,K, respectively, where M k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset, where M k Each row index of includes a data element of the corresponding Doppler bin index in the k-th motion spectrum data set,

[0034]

[0035] θ k,1 ,θ k,2 ...,θ k,U (u=1, 2, ...., U) indicates U angle values ​​corresponding to U corresponding spatial bin indices in the k-th motion spectrum dataset, respectively, where, indicating an installation angle of a kth radar sensor, the installation angle being an angle between an axis of the kth radar sensor and an axis of a vehicle coordinate system of the vehicle, and

[0036] is the model matrix of the kth radar sensor, where indicates the installation position of the kth radar sensor along the X-axis of the vehicle coordinate system, and Indicates the installation position of the kth radar sensor along the Y-axis of the vehicle coordinate system.

[0037] In the previous embodiment, the value d may be pre-calculated based on the F corresponding Doppler bin indices in the kth motion spectrum dataset and independently of the radar data. k,1 ,d k,2 ...,d k,F , and the value θ can be pre-computed based on the U corresponding spatial bin indices in the k-th motion spectrum dataset and independently of the radar data k,1 ,θ k,2 ...,θ k,U .

[0038] The motion equations can be solved using a linear least squares estimation method to determine the self-motion information. The self-motion information can be determined by the following steps: randomly selecting a plurality of Doppler bin index subsets from the Doppler bin indexes in the K motion spectrum data sets; performing the following processing for each Doppler bin index subset in the plurality of Doppler bin index subsets: k,1 ,d k,2 ...,d k,F (k=1, 2, ...., K) are set to the radial velocity values ​​corresponding to the corresponding Doppler bin indexes in the Doppler bin index subset in the k-th motion spectrum data set, and M k The rows of are set to the data elements of the corresponding Doppler bin index in the Doppler bin index subset in the kth motion spectrum data set, and the motion equation group D=H·E is solved using the linear least squares estimation method to estimate the self-motion information, and the estimation error for each Doppler bin index in the Doppler bin index subset is determined; and based on the determined estimation error, each Doppler bin index in the Doppler bin index subset is classified as one of a normal (inlier) Doppler bin index and an abnormal (outlier) Doppler bin index. The self-motion information can be determined by further performing the following processing: selecting a Doppler bin index subset from the multiple Doppler bin index subsets based on one of the following items: the number of normal Doppler bin indices in each Doppler bin index subset; and the sum of the estimation errors of all normal Doppler bin indices determined for each Doppler bin index subset; and by dividing d k,1 ,d k,2 ...,d k,F(k=1, 2, ...., K) are set to radial velocity values ​​corresponding to corresponding Doppler bin indices in the k-th motion spectrum data set classified as normal Doppler bin indices in the selected Doppler bin index subset, and M k Each row of (k=1,2,....,K) is set to the data element of the corresponding Doppler bin index classified as the normal Doppler bin index in the selected Doppler bin index subset in the kth motion spectrum data set, and the motion equation group D=H·E is solved using the linear least squares estimation method to determine the self-motion information.

[0039] In an example embodiment, the method may further include the steps of receiving radar data comprising a corresponding plurality of data samples of each of a plurality of radar return signals received at each of the antenna elements of the radar sensor; processing the data samples to generate, for each of the antenna elements, corresponding Doppler-processed data comprising one of the following items: (i) a plurality of data values, each of the plurality of data values ​​being calculated for a corresponding range bin index in a plurality of range bin indices and for a corresponding Doppler bin index in the plurality of Doppler bin indices; and (ii) a plurality of data values, each of the plurality of data values ​​being calculated for a corresponding fast time index in a plurality of fast time indices and for a corresponding Doppler bin index in the plurality of Doppler bin indices. Moreover, a motion spectrum data set is obtained by performing the following processing on each Doppler bin index among the multiple Doppler bin indexes: selecting data values ​​of the Doppler processed data that have been calculated for the Doppler bin index; calculating a covariance matrix using the data values ​​selected for the Doppler bin index; and applying a spectral estimation algorithm that uses the covariance matrix to determine corresponding spatial spectrum values ​​of each spatial bin index among the multiple spatial bin indexes for the Doppler bin index.

[0040] In an example embodiment, each of the data values ​​of the Doppler processed data may be calculated for a corresponding range bin index in a plurality of range bin indices and for a corresponding Doppler bin index in a plurality of Doppler bin indices, and the data value selected for each Doppler bin index may define an R×I dimensional matrix including data values ​​calculated for R antenna elements and for I range bin indices of the Doppler bin index. The covariance matrix may be an R×R dimensional matrix determined from the R×I dimensional matrix using a product of an R×I dimensional matrix and a conjugate transpose of the R×I dimensional matrix.

[0041] Alternatively, in an example embodiment, each of the data values ​​of the Doppler processed data may be calculated for a corresponding fast time index in a plurality of fast time indexes and a corresponding Doppler bin index in a plurality of Doppler bin indexes, and the data value selected for each Doppler bin index may define an R×I dimensional matrix including data values ​​calculated for R antenna elements and for I fast time indexes of Doppler bin index (j). The covariance matrix may be an R×R dimensional matrix determined based on the R×I dimensional matrix using the product of the R×I dimensional matrix and the conjugate transpose of the R×I dimensional matrix.

[0042] In any of the example embodiments or variants thereof set forth above, the spectrum estimation algorithm may be a Multiple Signal Classification (MUSIC) algorithm or an Estimation of Signal Parameters via Rotational Invariance Techniques (ESPIRIT) algorithm.

[0043] Moreover, the step of generating the motion spectrum data set may further include: normalizing the spatial spectrum values ​​determined for each Doppler bin index to generate a corresponding plurality of normalized spatial spectrum values, wherein each normalized spatial spectrum value in the normalized spatial spectrum values ​​is generated for a corresponding Doppler bin index in the plurality of Doppler bin indices and for a corresponding spatial bin index in the plurality of spatial bin indices.

[0044] The step of generating a motion spectrum data set may further include: calculating, for each Doppler bin index among the multiple Doppler bin indices, data indicating the variance of the probability distribution of the angle of arrival of the radar echo signal relative to the axis of the radar sensor among the angles of arrival indicated by the corresponding spatial bin index by using the normalized spatial spectrum value generated for the spatial bin index as the probability value of the spatial bin index; and determining a subset of Doppler bin indices among the multiple Doppler bin indices, for which the calculated variance is lower than a predetermined threshold, wherein the value of each data element in the motion spectrum data is based on the corresponding normalized spatial spectrum value generated for the Doppler bin index in the Doppler bin index subset and for the spatial bin index corresponding to the data element. In this case, the value of the data element in the motion spectrum data set is generated for the Doppler bin index by processing the multiple normalized spatial spectrum values ​​generated for each Doppler bin index in the Doppler bin index subset according to the following steps: identifying a spatial bin index whose normalized spatial spectrum value is the highest among the normalized spatial spectrum values ​​generated for the multiple spatial bin indices; setting the value of the data element of the identified spatial bin index to a non-zero value; and setting the values ​​of the data elements of all other spatial bin indices to zero.

[0045] Alternatively, the value of the data element in the motion spectrum data set can be generated for the Doppler bin index by processing the multiple normalized spatial spectrum values ​​generated for each Doppler bin index in the multiple Doppler bin indexes according to the following steps: identifying a spatial bin index whose normalized spatial spectrum value is the highest among the normalized spatial spectrum values ​​generated for the multiple spatial bin indexes; setting the value of the data element of the identified spatial bin index to a non-zero value; and setting the values ​​of the data elements of all other spatial bin indexes to zero.

[0046] The step of determining the self-motion information may also include: solving the set of motion equations using a least square adjustment method, and the least square adjustment method may perform at least one iteration of an iterative process, wherein, in each iteration of the iterative process, an increment of the estimate of the self-motion information is calculated to obtain an updated estimate, and the updated estimate is based on the estimate and the increment and is used to calculate an increment in the next iteration of the iterative process, wherein the increment is calculated using the following items: (i) a first value set of a motion spectrum data set, each value in the first value set indicating a distribution of the arrival angle among the arrival angles indicated by the corresponding spatial bin index for the Doppler bin index in the motion spectrum data set. The invention further comprises the following steps: (i) a first value set for a motion spectrum data set, wherein each value in the first value set is calculated using the normalized spatial spectrum values ​​corresponding to the plurality of spatial bin indices for the Doppler bin index as the corresponding probability values ​​for the plurality of spatial bin indices, and (ii) a second value set for a motion spectrum data set, wherein each value in the second value set indicates the variance of the distribution of radial velocities among radial velocities indicated by the corresponding Doppler bin indices for the spatial bin indices in the motion spectrum data set, wherein each value in the second value set is calculated using the normalized spatial spectrum values ​​corresponding to the plurality of Doppler bin indices for the spatial bin index as the corresponding probability values ​​for the plurality of Doppler bin indices. An initial estimate of the ego-motion information used in the iterative process may be obtained using a linear least squares estimation method.

[0047] Moreover, according to a second exemplary aspect of the present invention, a computer program including computer program instructions is designed, and when the computer program instructions are executed by a computer processor, the computer processor performs the method according to the first exemplary aspect or any one of its embodiments and variants described above. For example, the computer program can be stored on a non-transitory computer-readable storage medium, or can be carried by a signal.

[0048] According to a third exemplary aspect of the present invention, a radar data processing device is designed, which is configured to determine the self-motion information of a vehicle including a radar sensor having multiple antenna elements. The radar data processing device includes a motion spectrum acquisition module, which is configured to acquire a motion spectrum data set based on radar data generated by the radar sensor, the motion spectrum data set including a plurality of data elements, each of which is calculated for a corresponding Doppler bin index in a plurality of Doppler bin indexes and a corresponding spatial bin index in a plurality of spatial bin indexes, each of which indicates a corresponding arrival angle of the radar echo signal relative to the axis of the radar sensor. The radar data processing device also includes a self-motion determination module, which is configured to determine the self-motion information by solving a group of motion equations including a plurality of motion equations generated using the motion spectrum data set, each of which involves: a corresponding value indicating the radial velocity of a stationary object relative to the radar sensor, a corresponding value indicating the angular displacement of a stationary object relative to the axis of the radar sensor, and a variable indicating the speed of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Example embodiments will now be described, by way of non-limiting example only, with reference to the following drawings. Unless otherwise indicated, like reference numerals appearing in different figures may indicate identical or functionally similar elements.

[0050] Figure 1A is a schematic diagram of a radar data processing device according to a first example embodiment.

[0051] Figure 1B is a schematic illustration of a vehicle including a radar sensor for generating radar data to be processed by a radar data processing apparatus of a first example embodiment.

[0052] Figure 2 An example implementation in programmable signal processing hardware of a radar data processing device of the first example embodiment of this document is illustrated.

[0053] Figure 3 is a flowchart illustrating a method by which a radar data processing apparatus of a first example embodiment may process radar data to generate motion spectrum data for estimating vehicle ego-motion information.

[0054] Figure 4A An example of a radar data cube representation of Doppler processed data that has been generated from radar data in a first example embodiment is illustrated.

[0055] Figure 4B An example of a two-dimensional array including spatial spectrum values ​​generated for each Doppler bin index of a plurality of Doppler bin indices is illustrated.

[0056] Figure 5 is a flow chart illustrating further processing performed by the radar data processing apparatus of the first example embodiment to obtain a motion spectrum data set.

[0057] Figure 6 An example plot of normalized spatial spectrum values ​​in the Doppler contrast angle domain generated by the radar data processing apparatus of the first example embodiment is illustrated.

[0058] Figure 7 is a flowchart illustrating a process performed by the motion spectrum acquisition module 6 of this example embodiment to estimate ego-motion parameters using a linear least squares estimation method.

[0059] Figure 8 Six different plots of motion spectrum data sets corresponding to radar data acquired by six different radar sensors mounted on a host vehicle are illustrated.

[0060] Fig. 9 The example is fitted to Figure 8 Self-motion curves for each of the six plots in , where each curve is fit using linear least squares estimation combined with an iterative RANSAC process.

[0061] Fig.10 Illustrated is a comparison of self-motion information estimated using linear least squares estimation and self-motion information obtained using other methods.

[0062] Fig.11 The comparison between the self-motion information estimated using the Gauss-Helmert model and the self-motion information obtained using other methods is illustrated.

[0063] Fig.12 A motion spectrum dataset generated from a two-dimensional array having a plurality of sub-arrays is illustrated.

[0064] Fig.13A A radar data processing device according to a second example embodiment is illustrated.

[0065] Fig. 13B is a flowchart illustrating a method of determining ego-motion information of a host vehicle by a radar data processing apparatus of a second example embodiment. DETAILED DESCRIPTION

[0066] Figure 1A 1 is a schematic diagram of a radar data processing device 10 according to a first exemplary embodiment. In this exemplary embodiment, the radar data processing device 10 may be configured to process radar data from a radar sensor 20 (eg, Figure 1BThe radar data obtained by the vehicle is processed to generate a motion spectrum data set S for estimating the self-motion information of the vehicle. Figure 1A As shown, the radar data processing device 10 includes: a motion spectrum acquisition module 6 and a self-motion information determination module 8. In this example embodiment, the radar data processing device 10 may also include a data acquisition module 2 and a Doppler processing module 4.

[0067] like Figure 1B As shown, with origin O r 、X axis r and Y axis Y r The radar coordinate system is associated with the radar sensor 20. In this example embodiment, the X-axis X r can be aligned with the boresight of the radar sensor 20, although the boresight can generally have any orientation relative to the axes of the radar coordinate system. Figure 1B As shown, a stationary object 40 within the field of view of the radar sensor 20 may form an angle θ relative to the axis of the radar sensor 20, wherein θ indicates an angular displacement of the stationary object 40 relative to the axis of the radar sensor 20. In this example, θ represents an angular displacement relative to the X-axis X of the radar sensor 20. r The azimuth angle formed.

[0068] like Figure 1B As shown, the radar sensor 20 is mounted on the vehicle 30 relative to the origin O in the vehicle coordinate system. h Limited location (l x ,l y ) x Indicates that the radar sensor 20 is along a first axis of the vehicle coordinate system (eg, the X axis X in this example). h ) installation position, and L y Instructs the radar sensor 20 to move along a second axis of the vehicle coordinate system (eg, the Y axis Y in this example). h ) installation position. The orientation of the radar sensor 20 relative to the axis of the vehicle coordinate system is determined by the installation angle θ of the radar sensor M The installation angle θ indicates the installation angle of the radar sensor 20, which is the angle between the axis of the radar sensor 20 and the axis of the vehicle coordinate system of the vehicle 30. In this example, the installation angle θ M is the sighting line (along the X-axis X) of the radar sensor 20. r ) relative to the X-axis of the vehicle coordinate system h The angle formed.

[0069] It should be understood that the coordinate system of the vehicle 30 and the coordinate system of the radar sensor 20 as well as the position and installation angle of the radar sensor 20 are not limited to Figure 1B Rather, the coordinate systems may have axes defined in any suitable manner, and radar sensor 20 may be mounted at any suitable location on host vehicle 30 at any suitable mounting angle.

[0070] Assume that the vehicle 30 is moving at a yaw rate ω and has an X-axis component v x (In this example, relative to X h defined) and the Y axis component v y (i.e., in this example, relative to Y h The radial velocity d of the stationary object 40 measured by the radar sensor 20 and the azimuth angle θ of the stationary object (as defined by Figure 1B The relationship between the illustrated axis and the axis of the radar sensor 20 may be given by the following equation:

[0071] d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x +sin(θ+θ M )v y (1)

[0072] exist Figure 1B , the radial velocity measured by radar sensor 20 for stationary object 40 indicates the velocity of vehicle 30. Specifically, as radar sensor 20 moves with vehicle 30, radar sensor 20 observes stationary object 40 moving at a velocity equal in magnitude to the velocity of radar sensor 20 but in an opposite direction.

[0073] Under the assumption that the yaw rate of the vehicle 30 is zero, the speed of the vehicle 30 can be taken to be equal to the speed of the radar sensor 20. In this regard, the relationship between the radial speed d of the stationary object 40 measured by the radar sensor 20 and the azimuth angle θ of the stationary object 40 can also be described by the following formula:

[0074]

[0075] in, Indicative of the X component of the velocity of the radar sensor 20, and Indicates the Y component of the velocity of the radar sensor 20. Therefore, when the yaw rate is zero or close to zero, the radar sensor velocity can be calculated by using equation (2): and is solved, and for example by setting and To determine that the vehicle speed is equal to the radar sensor speed, thereby estimating the ego-motion information of the vehicle 30 .

[0076] Furthermore, in some cases, it may be assumed that the Y-axis component v of the vehicle 30 velocity is y is zero, in this case, the above formula (1) is analyzed as follows:

[0077] d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x (3)

[0078] In this example embodiment, the radar sensor 20 may include one or more transmitting antenna elements (not shown) that are configured to send a series of radar signals to a stationary object 40 near the vehicle 30. In this example embodiment, each radar signal may be, for example, a frequency modulated signal, such as a chirp signal that changes its frequency over a predetermined bandwidth. However, the radar signal may alternatively be sent at a fixed frequency. In this example embodiment, the radar sensor 20 may also include an array of receiving antenna elements (not shown) that is configured to receive radar echo signals corresponding to the series of radar signals reflected from the stationary object 40. In this example embodiment, the receiving antenna elements form a linear, equally spaced one-dimensional array. However, the receiving antenna element array is not limited in this regard and may alternatively form a two-dimensional array having any suitable array pattern.

[0079] The radar sensor 20 may also demodulate the received radar echo signal to generate a baseband (or intermediate frequency) signal, for example, by mixing the radar echo signal with the transmitted radar signal. The obtained baseband signal is further digitized using an analog-to-digital converter to generate a plurality of time domain data samples for each of the plurality of radar echo signals received at each of the plurality of receiving antenna elements of the radar sensor 20. The radar data R including the generated time domain data samples may also be provided to the radar data processing device 10 to generate a motion spectrum data set S for estimating the ego-motion information of the vehicle 30, the details of which are described below.

[0080] Figure 2 is a schematic illustration of programmable signal processing hardware 200 that may be configured to perform the operations of the radar data processing device 10 of the example embodiment.

[0081] The programmable signal processing device 200 includes a communication interface (I / F) 210 for communicating with the radar sensor 20 to receive radar data from the radar sensor. The signal processing device 200 also includes a processor (e.g., a central processing unit CPU) 220, a working memory 230 (e.g., a random access memory), and an instruction storage section 240 storing a computer program 245, which includes computer-readable instructions that, when executed by the processor 220, cause the processor 220 to perform various functions of the radar data processing device 10 described herein. The working memory 230 stores information used by the processor 220 during the execution of the computer program 245. The instruction storage section 240 may include a ROM (e.g., in the form of an electrically erasable programmable read-only memory (EEPROM) or flash memory) preloaded with computer-readable instructions. Alternatively, the instruction storage unit 240 may include a RAM or similar type of memory, and the computer-readable instructions of the computer program 245 may be input to the instruction storage unit from a computer program product (such as a non-transitory computer-readable storage medium 250 in the form of a CD-ROM, DVD-ROM, etc., or a computer-readable signal 260 carrying computer-readable instructions). In any case, when the computer program 245 is executed by the processor 220, the processor 220 performs the method of determining the self-motion information of the vehicle 30 described below, and optionally also performs the method of generating a motion spectrum data set S for estimating the self-motion information of the vehicle as described herein. However, it should be noted that the radar data processing device 10 may alternatively be implemented in non-programmable hardware (such as an application-specific integrated circuit (ASIC)).

[0082] Figure 3 is an example Figure 1A The motion spectrum acquisition module 6 of the radar data processing device 10 can acquire a motion spectrum data set S for estimating the self-motion information of the vehicle 30. The motion spectrum data set is based on the radar data generated by the radar sensor 20 and includes a plurality of data elements, each of which is calculated for a corresponding Doppler bin index in a plurality of Doppler bin indexes and for a corresponding spatial bin index in a plurality of spatial bin indexes. Each spatial bin index indicates a corresponding arrival angle of the radar echo signal relative to the axis of the radar sensor 20. Figure 3 In the example of FIG. 1 , the motion spectrum acquisition module 6 acquires the motion spectrum data set S by generating the motion spectrum data S based on the radar data R provided by the radar sensor 20, as described in more detail below. However, the motion spectrum acquisition module 6 does not need to acquire the motion spectrum data set in this manner, and may alternatively receive a motion spectrum data set that has been generated elsewhere according to the processing described below or in other ways, for example.

[0083] exist Figure 2 In step S10, the data acquisition module 2 receives radar data R, which includes a corresponding plurality of data samples of each radar echo signal in a plurality of radar echo signals received at each antenna element of the radar sensor 20. In this example embodiment, the radar data R may be a time domain data sample generated by the radar sensor 20, as previously described. These time domain data samples may be represented as a three-dimensional array defining a radar data cube. The first dimension of the array may correspond to a "fast time" dimension, along which a plurality of data samples acquired for a single radar signal (e.g., a chirp signal) at a plurality of corresponding fast time sampling intervals may be arranged. The second dimension of the array may correspond to a "slow-time" dimension, along which data samples acquired from a plurality of different radar signals (e.g., different chirp signals) at the same fast time interval index may be arranged. The third dimension of the array may correspond to data samples acquired by a plurality of receiving antenna elements from a single radar signal (e.g., a chirp signal) and for a single fast time interval index. The received radar data R is passed to the Doppler processing module 4.

[0084] exist Figure 3 In step S20, the Doppler processing module 4 processes the data samples of the radar data R to generate corresponding Doppler processed data D for each of the receiving antenna elements. In this example embodiment, the Doppler processed data includes a plurality of data values, each of which is calculated for a corresponding range bin index in a plurality of range bin indices and for a corresponding Doppler bin index in a plurality of Doppler bin indices. As in this example, the Doppler processing module 4 may process the data samples to generate Doppler processed data D including a plurality of data values ​​by performing range FFT processing on the time domain samples in a fast time dimension, and further performing Doppler FFT processing on the output frequency domain samples of the range FFT processing in a slow time dimension. However, it should be understood that in an alternative implementation, the Doppler processing module 4 may not perform range FFT processing on the time domain data samples, but may only perform Doppler FFT to generate a plurality of data values. That is, in this alternative implementation, the Doppler processed data D instead includes a plurality of data values, each of which is calculated for a corresponding fast time index in a plurality of fast time indices and for a corresponding Doppler bin index in a plurality of Doppler bin indices.

[0085] exist Figure 3In step S30, the motion spectrum acquisition module 6 generates motion spectrum data S including a plurality of data elements, each of which is calculated for a corresponding Doppler bin index in a plurality of Doppler bin indexes and a corresponding spatial bin index in a plurality of spatial bin indexes. Each spatial bin index indicates the axis of the radar echo signal relative to the radar sensor 20 (i.e., Figure 1B Y r ). In this example embodiment, each spatial bin index may indicate an azimuth angle formed relative to the axis of the radar sensor 20. Equivalently, each spatial bin index also indicates a spatial frequency of the radar echo signal relative to the plurality of receiving antenna elements. Specifically, for an azimuth angle range of -π / 2 to π / 2, there is a one-to-one mapping between azimuth angles and spatial frequencies.

[0086] More specifically, in Figure 3 In step S30, for each Doppler bin index in the plurality of Doppler bin indexes, the motion spectrum acquisition module 6 performs the processing of steps S30-1, S30-2, and S30-3. Figure 3 In step S30-1, the motion spectrum acquisition module 6 selects the data value of the Doppler-processed data D that has been calculated for the Doppler bin index. Figure 3 In step S30-2, the motion spectrum acquisition module 6 uses the data values ​​selected for the Doppler bin to calculate the covariance matrix. Figure 3 In step S30-3, the motion spectrum acquisition module 6 applies a spectrum estimation algorithm that uses a covariance matrix to determine the corresponding spatial spectrum value of each of the plurality of spatial bin indices for the Doppler bin index. Therefore, each data element in the motion spectrum data S is based on the corresponding normalized spatial spectrum value generated for the Doppler bin index and the spatial bin index corresponding to the data element.

[0087] Figure 4A A radar data cube 400 is illustrated that represents data values ​​of Doppler processed data D that are obtained at Figure 3 In step S20, the antenna elements are generated for each of the plurality of antenna elements. Figure 4A , radar data cube 400 is an I×J×R dimensional array whose data elements are identified by corresponding indices in I range bin indices (along the range dimension of the array), J Doppler bin indices (along the Doppler dimension of the array), and R antenna element indices (along the antenna element dimension of the array). Each Doppler bin index indicates a radial velocity of a potential object relative to radar sensor 20. Also, each range bin index indicates a distance of a potential object from radar sensor 20.

[0088] In this example implementation, Figure 3The data values ​​selected for each Doppler bin index j (where j=1, 2, ..., J) in step S30-1 define an R×I dimensional matrix X j , for Doppler bin index j, the matrix includes data values ​​(in the Doppler processed data D) calculated for R antenna element indices r=1, 2, ..., R and I range bin indices i=1, 2, ..., I. Therefore, the matrix X j includes I beam vectors, each of which includes R data values ​​corresponding to R corresponding spatial bin indices. Figure 3 The covariance matrix calculated in step S30-1 may be based on the matrix X j With the matrix X j The product of the conjugate transpose of the matrix X j Determine the R×R dimensional covariance matrix S xx As a non-limiting example, the covariance matrix can be calculated as S xx =E[X j X j H ]-E[X j}E{X j H ].

[0089] However, in the case where range FFT processing is not performed to generate data values ​​(i.e., where each of the data values ​​of the Doppler processed data is calculated for a corresponding fast time index in a plurality of fast time indices and a corresponding Doppler bin index in a plurality of Doppler bin indices), Figure 3 The data values ​​selected for each Doppler bin index in step S30-1 are limited to an R×I dimensional matrix including data values ​​calculated for K antenna elements and I fast time indices for the Doppler bin index, and wherein the signal covariance matrix is ​​an R×R dimensional matrix determined from the R×I dimensional matrix using the product of an R×I dimensional matrix and a conjugate transpose of the R×I dimensional matrix.

[0090] In this example implementation, Figure 3 In step S30-3, the motion spectrum acquisition module 6 determines the spatial spectrum value of each Doppler bin by using a multiple signal classification (MUSIC) algorithm as a spectrum estimation algorithm. In this example embodiment, the motion spectrum acquisition module 6 may use the MUSIC algorithm to calculate the spatial spectrum value p of each spatial bin index in the plurality of spatial bin indexes n=1, 2..., N for each Doppler bin index in the Doppler bin index j=1, 2..., J by using the following MUSIC spatial spectrum function: jn :

[0091]

[0092] where a(θ) is the steering vector of the multiple antennas corresponding to the arrival angle θ, and E N Denotes the covariance matrix X for Doppler bin index j according to j The noise subspace derived from the eigendecomposition of . As can be observed from equation (1), there can be at most two targets for each Doppler bin, so when determining the spatial spectrum value using the MUSIC algorithm, one or two targets can be assumed.

[0093] In some embodiments, for example, the antenna elements of the radar sensor 20 may be calibrated to minimize the amplitude and phase errors between the antenna elements. In such embodiments, the calibration matrix may be first applied to the covariance matrix to generate a calibrated covariance matrix, and then the calibrated covariance matrix may be used to perform spectral estimation.

[0094] It should be noted that each spatial bin is not only mapped to an angle value in the angle domain, but can also be equivalently mapped to a spatial frequency value in the spatial frequency domain, because the spatial frequency ψ of the signal in the plane along which the receiving antenna element array is arranged depends on the arrival angle θ of the signal. Therefore, in some example embodiments, the spatial spectrum value can be determined in the spatial frequency domain, wherein each spatial bin corresponds to the spatial frequency value of the signal relative to the radar sensor 20.

[0095] Moreover, in Figure 3 The spectrum estimation algorithm used to generate the spatial spectrum value in step S30-3 is not limited to the MUSIC algorithm, and may alternatively be a signal parameter estimation via rotational invariance technique (ESPIRIT) algorithm, or any suitable spectrum estimation algorithm that can be used to estimate the spatial spectrum or arrival angle of one or more radar echo signals based on a covariance matrix obtained from the detected signal. Moreover, the spectrum estimation algorithm based on the observed covariance matrix assumes that the signal sources are uncorrelated. However, when some of the signals are correlated, spatial smoothing may be performed. More specifically, in some example embodiments, the motion spectrum acquisition module 6 may be configured to determine the covariance matrix corresponding to the multiple subarrays formed by the multiple antenna elements by performing a spatial smoothing operation, and to generate a covariance matrix based on the covariance matrix determined for the multiple subarrays, thereby determining the covariance matrix for each Doppler bin index in the multiple Doppler bin indices.

[0096] Figure 4B is included in Figure 3 FIG. 4 is an example illustration of a two-dimensional array 420 of spatial spectrum values ​​generated for each Doppler bin index in the plurality of Doppler bin indexes in step S30. More specifically, Figure 4BThe two-dimensional array 420 in 420 includes J Doppler bin indices j=1, 2, . . . , J along a first direction of the array 420 and N spatial bin indices n=1, 2, . . . , N along a second direction of the array 420. The spatial spectrum values ​​generated for each Doppler bin index correspond to columns 430 of the array 420. Figure 4B The array 420 in includes a plurality of spatial spectrum values ​​p jn , j = 1, 2 ..., J, n = 1, 2 ..., N, where each spatial spectrum value p is calculated for the Doppler bin index j and the spatial bin index n in the array 420 jn Moreover, each data element in the motion spectrum data set S is based on the corresponding normalized spatial spectrum value p generated for the Doppler bin index j and the spatial bin index n corresponding to the data element. jn If Figure 4B As illustrated, the spatial spectrum values ​​for Doppler bin index j at column 430 are obtained by using the data values ​​in radar data cube 400 within the range of Doppler bin index j as a single “snapshot” of radar sensor measurements in the spectrum estimation algorithm.

[0097] Figure 5 is a flowchart illustrating a process that may be further performed by the motion spectrum acquisition module 6 to obtain a motion spectrum dataset S from the spatial spectrum values ​​that have been calculated for each Doppler bin index. Figure 3 After calculating the spatial spectrum value of each Doppler bin index in step S30-1, Figure 5 In step S210, the motion spectrum acquisition module 6 may further perform normalization of the spatial spectrum value determined for the Doppler bin index to generate a corresponding plurality of normalized spatial spectrum values, wherein each normalized spatial spectrum value in the normalized spatial spectrum values ​​is generated for a corresponding Doppler bin index in the plurality of Doppler bin indexes and for a corresponding spatial bin index in the plurality of spatial bin indexes. For example, the motion spectrum acquisition module 6 may perform L1 normalization on the plurality of spatial spectrum values ​​generated for each Doppler bin index so that the sum of all normalized spatial spectrum values ​​generated for the Doppler bin index is equal to a value of 1. The normalized spatial spectrum value calculated from the plurality of spatial spectrum values ​​may be represented as q jn ,j=1,...,J,n=1,...,N.

[0098] Figure 6 Example Figure 5 An example plot of the normalized spatial spectrum values ​​generated in step S210 of FIG. Figure 6In the example, the Doppler bin index covers a range of -10 m / s to 10 m / s when converted into radial velocity values, and spatial spectrum values ​​are calculated (using the MUSIC algorithm in this example embodiment) for an (azimuth) angle range of -75° to 75° relative to the axis of the radar sensor 20. Higher intensity pixel values ​​are used to display higher amplitude normalized spatial spectrum values.

[0099] like Figure 6 As shown, since there are one or more stationary targets 40 in the field of view of the radar sensor 20, for each Doppler bin index, the variation of the radial velocity with angle mapped by the peak normalized spatial spectrum value (or the high-valued normalized spatial spectrum value) approximately follows a sinusoidal trajectory, which is consistent with the relationship specified by the motion equations in equations (1), (2), and (3). Therefore, the normalized spatial spectrum value can be used to calculate the self-motion parameter v in equations (1), (2), and (3) by performing a regression analysis, for example, by fitting the function given by equations (1), (2), and (3) to the variation of the radial velocity with angle specified by the peak (or high-valued) normalized spatial spectrum value of each Doppler bin index. x 、v y and ω is estimated.

[0100] exist Figure 6 The presence of a moving target will cause isolated peaks outside the sinusoidal trajectory. For example, Figure 6 The presence of strong normalized spatial spectrum values ​​due to moving objects may result in reduced accuracy in estimating the ego-motion parameters. In particular, assuming that the angular range estimated by the spectral estimation algorithm covers a range for which equation (1) has only one unique angular solution for each radial velocity, then Figure 6 For each Doppler bin in , there should be only a single strongly normalized spatial spectrum value.

[0101] Therefore, if Figure 5 As shown in step S220, in this example embodiment, the motion spectrum acquisition module 6 further calculates, for each Doppler bin index in the plurality of Doppler bin indexes, data indicating the variance of the probability distribution of the angle of arrival of the radar echo signal relative to the axis of the radar sensor 20 among the angles of arrival indicated by the corresponding spatial bin index by using the normalized spatial spectrum value generated for the spatial bin index as the probability value of the spatial bin index. For example, for a given Doppler bin index j, a plurality of spatial bin indexes n=1, 2..., N can be used as random angle variables θ jMoreover, the normalized spatial spectrum value q calculated for each spatial bin index in the spatial bin index n=1, 2..., N for the Doppler bin index j can be jn As the probability of occurrence of the spatial bin index. Therefore, the angle variable Θ can be calculated using the following formula j At the angle θ represented by the spatial bin index n=1, 2, ..., N for Doppler bin j, respectively n The variance of the probability distribution among (n=1,2...,N):

[0102] in, is for the random angle variable Θ j However, in some embodiments, instead of using the angle variable θ j The expected value of θ can be used to calculate the variance in equation (5), and the angle variable θ can be used instead. j The mode of (corresponding to the peak normalized spatial spectrum value for Doppler bin index j).

[0103] exist Figure 5 In step S230, the motion spectrum acquisition module 6 determines a subset of Doppler bin indices among the plurality of Doppler bin indices, i.e., for the subset of Doppler bin indices, the variance calculated in step S220 is lower than a predetermined threshold. In addition, the value of each data element in the motion spectrum data S is based on the corresponding normalized spatial spectrum value generated for the Doppler bin index and the spatial bin index in the subset of Doppler bin indices corresponding to the data element. In other words, Figure 5 After normalizing the spatial spectrum values ​​in step S210 and calculating the variance for each Doppler bin index in step S220, only the Doppler bin index having a variance below a predetermined threshold is used to generate the motion spectrum data S.

[0104] like Figure 5 As shown in step S240 of Figure 5 In the steps S240-1, S240-2 and S240-3 illustrated in the example, multiple normalized spatial spectrum values ​​generated for each Doppler bin index in the Doppler bin index subset are processed to generate the value of the data element in the motion spectrum data set S for the Doppler bin index. More specifically, in step S240-1, the motion spectrum acquisition module 6 identifies the spatial bin index having the highest absolute value of the normalized spatial spectrum value among the normalized spectrum values ​​generated for the multiple spatial bin indexes. Figure 5 In step S240-2, the motion spectrum acquisition module 6 sets the value of the data element indexed by the identified spatial bin to a non-zero value. Figure 5In step S240 - 3 , the motion spectrum acquisition module 6 sets the values ​​of the data elements indexed for all other spatial bins to zero.

[0105] although Figure 5 The steps S210, S220, S230 and S240 are illustrated as being performed in sequence, but it should be understood that these steps can be performed in sequence. Figure 3 The step S220 and the step S230 are further performed to obtain the motion spectrum dataset S. That is, the motion spectrum dataset S may be the output of any of the steps S30, S210, S230 and S240 described above. In addition, in some example embodiments, the step S240 may be directly performed on the normalized spatial spectrum value generated in the step S210 to generate the motion spectrum dataset S without performing the steps S220 and S230.

[0106] After the motion spectrum data set S has been acquired, the self-motion information determination module 8 is configured to determine the self-motion information of the vehicle 30 by solving a set of motion equations including a plurality of motion equations generated using the motion spectrum data set. Each motion equation involves: a corresponding value indicating the radial velocity of the stationary object relative to the radar sensor 20, a corresponding value indicating the angular displacement of the stationary object relative to the axis of the radar sensor 20, and a variable indicating the velocity of the vehicle 30. The value indicating the radial velocity in the set of motion equations may be pre-calculated based on the Doppler bin index and independently of the radar data R. Moreover, the value indicating the angular position in the set of motion equations may be pre-calculated based on the spatial bin index and independently of the radar data R.

[0107] In this example embodiment, equation (1) may be used as the equation of motion to determine the velocity of the vehicle 30 including v x 、v y and the vehicle ego-motion information of ω.

[0108] In order to uniquely determine the ego-motion information v using equation (1) x 、v y and ω, motion spectrum data S generated from radar data acquired by at least two radar sensors installed at different positions around the vehicle 30. Each radar sensor may take the form of the radar sensor 20 described previously, and the motion spectrum acquisition module 6 may generate a corresponding motion spectrum dataset based on the radar data received at each radar sensor using any of the methods described previously for generating a motion spectrum dataset.

[0109] More specifically, for a case where K radar sensors are installed at K different locations on the vehicle 30, each radar sensor having a plurality of antenna elements, where K is an integer greater than or equal to 2, and where a corresponding motion spectrum dataset is generated for each of the K radar sensors (using a method for generating a motion spectrum dataset as described in any of the previously described embodiments / examples), the ego-motion information determination module 8 may be configured, as in the present example embodiment, to estimate v in equation (1) by solving the following set of motion equations: x 、v y And the value of ω, to determine the self-motion information:

[0110] D=H·E (6)

[0111] in,

[0112] d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ..., K, where F is an integer equal to or greater than 1. Value d k,1 ,d k,2 ...,d k,F Indicates F radial velocity values ​​corresponding to F respective Doppler bin indices in the k th motion spectrum data set S generated for the k th radar sensor 20 .

[0113] Matrix M k (k=1,2,....,K) is an F×U dimensional matrix containing data elements in the kth motion spectrum data set S, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1,2,....,K, respectively. k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset S, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset S. k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum dataset S.

[0114] The matrix T k Defined as:

[0115]

[0116] Among them, θ k,1 ,θ k,2 ...,θ k,U(u=1, 2, ...., U) indicates u angle values ​​respectively corresponding to u spatial bin indices in the k-th motion spectrum dataset. Indicates the installation angle of the kth radar sensor, which is the angle between the axis of the kth radar sensor and the axis of the vehicle coordinate system of the vehicle 20 .

[0117] The matrix L k Defined as:

[0118]

[0119] and is the model matrix of the kth radar sensor, where indicates the installation position of the kth radar sensor along the X-axis of the vehicle coordinate system, and indicates the installation position of the kth radar sensor along the Y axis of the vehicle coordinate system. The solution of (6) requires that H has at least three linearly independent rows.

[0120] It should also be noted that the system of equations (6) can be rewritten in many equivalent forms by transforming one or more variables of the system of equations into functions of other variables. As an example, if the motion spectrum dataset S includes spatial spectrum values ​​obtained at different spatial frequency bin indices, the angle value θ in the system of equations (6) can be alternatively replaced by k,1 ,...,θ k,U can be converted into corresponding spatial frequency values ​​in order to estimate the self-motion parameters in equation (4). k,1 ,...,θ k,U Transformed into the corresponding spatial frequency value ψ k,1 ,,...,,ψ k,U The relationship between can be given by ψ = bsin (θ), so the matrix T of equation (4) can be k can be rewritten equivalently as:

[0121]

[0122] The term b is a constant and can be given by, as in this example, Given, where c k is the distance between the antenna array elements of the kth radar sensor, and λ is the wavelength of the signal received at the kth radar sensor.

[0123] Solving the system of equations in (6) allows estimating the full set of ego-motion parameters v in (1) x 、v yHowever, in alternative embodiments, equation (1) may be simplified by assuming certain conditions. For example, in alternative embodiments, the yaw rate ω of the vehicle 30 may be assumed to be zero, and the previously defined equation (2) may be used to estimate the equation including v x and v y Specifically, in equation (2), (2) can be used to calculate the term and They are the X and Y components of the velocity of the radar sensor (not the velocity of the vehicle 30). Furthermore, under the condition of zero yaw rate, the and To determine the vehicle speed v x 、v y . It should also be noted that when the yaw rate is zero, the vehicle speed v x and v y It may be determined using only one motion spectrum data set S, without requiring at least two motion spectrum data sets from two radar sensors installed at two different positions on the vehicle 30 .

[0124] More specifically, for a case where K radar sensors are installed at K different locations of the vehicle 30, each radar sensor having a plurality of antenna elements, where K is an integer greater than or equal to 1, and where a corresponding motion spectrum dataset is generated for each of the K radar sensors (using a method for generating a motion spectrum dataset as described in any of the previously described embodiments / examples), the ego-motion information determination module 8 may be configured to estimate the ego-motion information by: and and set the ego-motion parameters to and To determine the self-motion information. In particular, estimate and The steps to find the value of include solving the following system of motion equations:

[0125] D=H·E (7)

[0126] in,

[0127] d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k,1 ,d k,2 ...,d k,FIndicates F radial velocity values ​​corresponding to F respective Doppler bin indices in the k-th motion spectrum data set S acquired for the k-th radar sensor 20 .

[0128] Matrix M k (k=1, 2, ...., K) is an F×U dimensional matrix containing the data elements in the kth motion spectrum data set S, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1, 2, ...., K, respectively. k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset S, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset S. k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum dataset S.

[0129] T k is defined as:

[0130]

[0131] Among them, θ k,1 ,θ k,2 ...,θ k,U (u=1, 2, ...., U) indicates U angle values ​​corresponding to U spatial bin indices in the k-th motion spectrum dataset S, respectively. Indicates the installation angle of the kth radar sensor, which is the angle between the axis of the kth radar sensor and the axis of the vehicle coordinate system of the vehicle 30 .

[0132] In some embodiments, it may be assumed that the Y-axis component of the vehicle velocity v y is zero, in which case the previously defined equation (3) can be used to obtain the equation including the values ​​ω and v x Self-motion information.

[0133] More specifically, for a case where K radar sensors are installed at K different positions of the vehicle 30, each of which has a plurality of antenna elements, where K is an integer greater than or equal to 1, and where a corresponding motion spectrum dataset S is generated for each of the K radar sensors (using a method for generating a motion spectrum dataset S as described in any of the previously described embodiments / examples), the ego-motion information determination module 8 may be configured to estimate the values ​​ω and v by solving the following set of motion equations: x , to determine the self-motion information:

[0134] D=H·E (8)

[0135] in,

[0136] d k =[d k,1 ,d k,2 ...,d k,F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k,1 ,d k,2 ...,d k,F Indicates F radial velocity values ​​corresponding to F respective Doppler bin indices in the k th motion spectrum data set S generated for the k th radar sensor 20 .

[0137] Matrix M k (k=1, 2, ...., K) is an F×U dimensional matrix containing the data elements in the kth motion spectrum data set S, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1, 2, ...., K, respectively. k Each row index of M corresponds to the corresponding Doppler bin index in the k-th motion spectrum dataset, and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset. k Each row of includes a data element of a corresponding Doppler bin index in the kth motion spectrum dataset.

[0138] The matrix T k Defined as:

[0139]

[0140] Among them, θ k,1 ,θ k,2 ...,θ k,U (u=1, 2, ...., U) indicates U angle values ​​corresponding to U corresponding spatial bin indices in the k-th motion spectrum dataset, respectively, where, Indicates the installation angle of the kth radar sensor, which is the angle between the axis of the kth radar sensor and the axis of the vehicle coordinate system of the vehicle 30 .

[0141] is the model matrix of the kth radar sensor, where indicates the installation position of the kth radar sensor along the X-axis of the vehicle coordinate system, and Indicates the installation position of the kth radar sensor along the Y-axis of the vehicle coordinate system.

[0142] In addition, it should be understood that the estimation of the ego-motion parameters can be equivalently performed by taking the transposition of both sides of the equations (6), (7) and (8).

[0143] In this example implementation, in order to generate a corresponding motion spectrum dataset for each radar sensor among the K radar sensors, for the radar sensor, Figure 3 The spatial spectrum value generated at step S30 is executed Figure 5 Therefore, the corresponding motion spectrum data set generated for each of the K radar sensors at step S240 includes a non-zero value (e.g., 1) at a single spatial bin index for each Doppler bin index, wherein the data elements of the remaining spatial bin indexes for each Doppler bin index are set to zero values ​​(e.g., as shown in FIG. 1 ). Figure 5 As described in step S240-3 of ). As in the present example, the non-zero value may be the value 1, but may alternatively be any non-zero value.

[0144] In this example embodiment, the value d may be pre-calculated based on the F corresponding Doppler bin indices in the kth motion spectrum dataset and independently of the radar data received at the kth radar sensor. k,1 ,d k,2 ...,d k,F Moreover, the value θ may be pre-computed based on the U corresponding spatial bin indices in the k-th motion spectrum dataset and independently of the radar data received at the k-th radar sensor. k,1 ,θ k,2 ...,θ k,U .

[0145] Specifically, in the present example embodiment, by using a predetermined conversion factor based on the Doppler resolution of the kth radar sensor 20, the value d may be derived from the corresponding Doppler bin index. k,1 ,d k,2 ...,d k,F Similarly, in this example embodiment, the angle value θ may be derived from the corresponding spatial bin index by using a predetermined conversion factor based on the angular resolution of the kth radar sensor 20 . k,1 ,θ k,2 ...,θ k,U In this example embodiment, before the radar data of each of the K radar sensors is acquired and processed by the radar data processing device 10, the radial velocity value d is pre-calculated and stored in the memory. k,1 ,d k,2 ...,d k,F 、Angle value θ k,1 ,θ k,2 ...,θ k,U , and the values ​​of the trigonometric terms in equations (6), (7), or (8). In other words, d k,1 ,d k,2...,d k,F The value of θ k,1 ,θ k,2 ...,θ k,U The values ​​of , as well as the values ​​of the triangular terms in the system of equations (6), (7) or (8) need only be calculated once, because their values ​​are known in advance due to the bin discretization.

[0146] To estimate the ego-motion parameters, the system of motion equations can be solved in equations (6), (7) or (8) using a linear least squares estimation method to determine the ego-motion information.

[0147] For example, in this embodiment, to estimate the ego-motion parameter v in equation group (6), x 、v y and the value of ω, the self-motion information determination module 8 can be set to use a linear least squares estimation algorithm. For example, the formula for ordinary least squares estimation can be used To calculate the self-motion parameters initial estimate of .

[0148] However, the linear least squares estimation is not limited to the above formula, and other formulas of the linear least squares estimation may be used.

[0149] In some example embodiments, in order to improve the robustness of the linear least squares estimation, the self-motion information determination module 8 may perform the linear least squares estimation by executing an iterative algorithm that classifies the Doppler bin indices in the K motion spectrum data sets S as normal and abnormal, and performing the estimation based only on the data elements of the normal Doppler bin indices. The iterative algorithm may, for example, be based on applying the Random Sample Consensus (RANSAC) method or the M-estimator Sample Consensus (MSAC) method to the data elements of the Doppler bin indices in the K motion spectrum data sets S in order to select a subset of Doppler bin indices that produces the best estimate of the self-motion parameters. However, other robust estimation methods such as M-estimators (e.g., Huber K-estimators) may also be used.

[0150] Figure 7 The example shows that the self-motion information determination module 8 can be used to estimate the self-motion parameter v using the linear least squares estimation method. x 、v y and the iterative method of ω.

[0151] exist Figure 7In step S310, the self-motion information determination module 8 randomly selects a plurality of Doppler bin index subsets from the Doppler bin indexes in the K motion spectrum data sets S.

[0152] exist Figure 7 In step S320, for each Doppler bin index subset in the plurality of Doppler bin index subsets, the ego-motion information determination module 8 performs the processing of steps S320-1, S320-2, and S320-3.

[0153] Specifically, in step S320-1, the self-motion information determination module 8 estimates the self-motion information by the following operation: based on the Doppler bin index settings D and H in the Doppler bin index subset, the motion equation group D=H·E listed in (6) is solved using the linear least squares estimation method. More specifically, the self-motion information determination module 8 estimates the self-motion information by the following operation: d k,1 ,d k,2 ...,d k,F (k=1, 2, ..., K) is set to the radial velocity value corresponding to the corresponding Doppler bin index in the Doppler bin index subset in the k-th motion spectrum data set, and M in equation group (6) is also set to k Each row of (included in the matrix H) is set to the data element of the corresponding Doppler bin index in the k-th motion spectrum data set that is in the Doppler bin index subset.

[0154] exist Figure 7 In step S320-2, the ego-motion information determination module 8 determines the estimation error of each Doppler bin index in the Doppler bin index subset. Figure 7 In step S320 - 3 , the ego-motion information determination module 8 classifies each Doppler bin index in the Doppler bin index subset as a normal Doppler bin index or an abnormal Doppler bin index based on the determined estimation error of the Doppler bin index.

[0155] In this example of using equation group (6) to determine the self-motion information, for step S320-2, the self-motion information determination module 8 can be based on the radial velocity value corresponding to the Doppler bin index and the value based on v x 、v y The estimated error of each Doppler bin index in the Doppler bin index subset is determined by the difference between the corresponding fitted values ​​of the regression model determined by the estimated value of ω and v. x 、v yand the estimated value of ω are substituted back into the set of motion equations (6) to determine the fitted radial velocity value for each Doppler bin index in the Doppler bin index subset. Therefore, the estimated error (i.e., error residual) for each Doppler bin index in the Doppler bin index subset can be given by the square root of the difference between the radial velocity value corresponding to the Doppler bin index and the fitted radial velocity value corresponding to the Doppler bin index. However, different metrics can be used to calculate the estimated error for each Doppler bin index.

[0156] In this example, for step S320-3, each Doppler bin index in the Doppler bin index subset can be classified as a normal Doppler bin index or an abnormal Doppler bin index based on a threshold value. A Doppler bin index with an estimated error below the threshold value can be determined as a normal Doppler bin index. Moreover, a Doppler bin index with an estimated error above the threshold value can be determined as an abnormal Doppler bin index.

[0157] exist Figure 7 In step S330, the self-motion information determination module 8 selects a Doppler bin index subset from the plurality of Doppler bin index subsets based on: (i) the total number of normal Doppler bin indexes of each Doppler bin index subset, or (ii) the sum of the estimation errors of all normal Doppler bin indexes determined for each Doppler bin index subset. For example, in the case of using the RANSAC algorithm, the self-motion information determination module 8 may select a Doppler bin index subset from the plurality of Doppler bin index subsets based on: Figure 7 In step S330 , a Doppler bin index subset having the most normal Doppler bin index is selected based on the determination of step S320 - 3 .

[0158] In some example embodiments, when the MSAC algorithm is used, the self-motion information determination module 8 may alternatively Figure 7In step S330, the self-motion information determination module 8 selects a Doppler bin index subset that has the lowest value in terms of the sum of estimated errors calculated for all normal Doppler bin indices in the subset. In other words, as part of step S330 of the present embodiment, the self-motion information determination module 8 may calculate, for each Doppler bin index subset, the sum of estimated errors of all normal Doppler bin indices in the Doppler bin index subset. In step S330, the Doppler bin index subset that generates the lowest value in the sum of estimated errors may be selected as the subset. In some embodiments using the MSAC algorithm, the self-motion information determination module 8 may be configured to select the Doppler bin index subset from the multiple Doppler bin index subsets by further considering the abnormal Doppler bin indices in each subset. Specifically, a constant error value may be assigned to each abnormal Doppler bin index. Then, a cumulative error may be determined for each Doppler bin index subset, the cumulative error being the sum of the estimated errors of all normal Doppler bin indices and the total constant error value of all abnormal Doppler bin indices (the total constant error value being calculated by multiplying the constant error value by the number of abnormal Doppler bin indices). In step S330, the Doppler bin index subset having the lowest cumulative error may be selected as the subset.

[0159] exist Figure 7 In step S340, the self-motion information determination module 8 determines the self-motion information by solving the motion equation group D=H·E as listed in (6) using the linear least squares estimation method through the following operation: D and H are set based on the Doppler bin index in the selected Doppler bin index subset. Specifically, the self-motion information determination module 8 determines the self-motion information by solving the motion equation group D=H·E as listed in (6) using the linear least squares estimation method through the following operation: d k,1 ,d k,2 ...,d k,F (k=1, 2, ...., K) are set to radial velocity values ​​corresponding to corresponding Doppler bin indices in the k-th motion spectrum data set classified as normal Doppler bin indices in the selected Doppler bin index subset, and M k Each row of (k=1, 2, ...., K) is set to a data element of a corresponding Doppler bin index in the k-th motion spectrum data set classified as a normal Doppler bin index in the selected Doppler bin index subset.

[0160] It should be understood that although in this example Figure 7Steps S320-1 and S340 are described as being performed to solve the system of equations listed in (6), but these two steps may also be performed to solve the system of equations listed in (7) or (8) when only some of the self-motion parameters are required.

[0161] The above linear optimization process assumes that the motion spectrum data set S generated for each radar sensor provides accurate measurement results without errors. However, this may not be true for noisy real radar measurements. In some example embodiments, the estimation of ego-motion parameters using the set of motion equations (6) (or using the set of equations (7) or (8)) can be improved by further considering the uncertainties in the observations, which are the changes in the angle and radial velocity in the estimation of the ego-motion information.

[0162] Therefore, in some example embodiments, the ego-motion information determination module 8 may be configured to determine the ego-motion information by further using the least squares adjustment method (which is v in the present example of solving the equation system (6) x 、v y and ω). The least squares adjustment method performs at least one iteration of an iterative process, wherein an increment of the estimate of the self-motion information is calculated in each iteration of the iterative process to obtain an updated estimate, the updated estimate being based on the estimate and the increment and being used to calculate the increment in the next iteration of the iterative process. Moreover, the increment is calculated based on a first value set of each motion spectrum data set in the K motion spectrum data sets and a second value set of each motion spectrum data set in the K motion spectrum data sets. Each value in the first value set indicates a variance of a distribution of an angle of arrival among angles of arrival indicated by a corresponding spatial bin index for a Doppler bin index in the motion spectrum data S, wherein each value in the first value set is calculated using a normalized spatial spectrum value corresponding to the multiple spatial bin indexes for the Doppler bin index as a corresponding probability value for the multiple spatial bin indexes. Moreover, each value in the second value set indicates a variance of a distribution of radial velocities among radial velocities indicated by a corresponding Doppler bin index for a spatial bin index in the motion spectrum data S, wherein each value in the second value set is calculated using normalized spatial spectrum values ​​corresponding to the plurality of Doppler bin indexes for the spatial bin index as corresponding probability values ​​for the plurality of Doppler bin indexes. As in the present embodiment, an initial estimate of the self-motion information used in the iterative process can be obtained using a linear least squares estimation method.

[0163] As in this example, the least squares adjustment method can be used with the Gauss-Helmert model, which is described in "Convex Optimization for Inequality Constrained Adjustment Problems", PhD thesis of LRRoese-Koerner, Institut für und Geoinformation derRheinischen Friedrich-Wilhelms- This is described in more detail in Bonn (2015).

[0164] More specifically, in the Gauss-Helmert model, we denote the F observations represented by the observation vector l, and denote the variance-covariance matrix of the observations as Q ll , the constraints given by the F nonlinear relationships between the observations l and the model parameter vector x can be written as:

[0165] g(l,x)=0 (9)

[0166] The nonlinear function can be solved by linearizing function (9) around the initial value and then iteratively solving the function. Specifically, function (9) can be linearized using a Taylor series expansion around a selected point (l0, x0) to express the nonlinear function in an approximate linear form. More specifically, it can be written as

[0167]

[0168]

[0169] where e represents the residual vector associated with the observations, generated using the linearization of (9) using the Taylor series expansion around the Taylor point (l0, x0)

[0170]

[0171] For the least squares adjustment in the nonlinear Gauss-Helmert model, the goal is to make the weighted least squares object function e T Q u -1 e is minimized. This minimization problem can be solved using the method of Lagrange multipliers to obtain the following normal equation system (13), which is solved to calculate the increment Δx of the parameter vector x, and the vector k of F Lagrange multipliers:

[0172]

[0173] After solving for Δx, we can use equation (10) to get an estimate for this iteration Furthermore, the residual vector e of this iteration can be calculated as

[0174] e=Q u B T z (14)

[0175] After calculating the residual vector e using equation (14), the estimated value can be further calculated using equation (11) This estimate can then be and is used as the starting point (l0, x0) for the next iteration of the estimation. More specifically, in the next iteration, the system of equations (13) is solved using the new values ​​of A, B, and w (the new values ​​are obtained using the new starting point (l0, x0)), and it is possible to calculate and New estimated value of .

[0176] In this example, for simplicity, the value of U in equation (6) is taken as F, so that the matrix M in equation (6) is k (k=1, 2, ..., K) is an F×F dimensional matrix. However, it should be understood that the value of U is not limited thereto and can be selected as any suitable value. Each equation in the system of equations (6) can be written in the form given by (9) so that each equation in the system of equations (6) is re-expressed as a function of the following form:

[0177] g(θ,d,ω,v x ,v y )=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x +sin(θ+θ M )v y -d=0 (15)

[0178] In the system of equations (6), F functions of the form expressed in (15) can be derived for each value of k = 1, 2, ...., K, where F can be chosen to be the same for all values ​​of k or can alternatively be chosen independently for each value of k. For each value of k = 1, 2, ...., K, the function of the form expressed in (15) can be derived for each value of k = 1, 2, ...., K using the function around the corresponding initial point The Taylor series expansion of is used to linearize each of the F functions. Moreover, as previously explained, the weighted least squares function minimization using the method of Lagrange multipliers can then be performed to derive the normal system of equations of the form shown in equation (13), which can be given as follows:

[0179]

[0180] in,

[0181] k=1,2,....,K,

[0182] Among them, w k (k=1,2,....,K) is a vector whose elements are obtained by replacing the initial point It is determined by substituting each of k=1, 2, ...., K into each of the corresponding F functions. and are the values ​​of d and θ in each equation of the equation group (6). The initial values ​​ω0, and The estimate provided by the linear least squares estimation method described previously is selected. However, the initial point is not limited to this, and any appropriate method can be used to select ω0, and The value of .

[0183] For each value of k = 1, 2, ...., K, the matrix A k By taking the initial point Substitute the self-motion parameters ω, v x 、v y The Jacobian matrix formed by the partial derivatives of the F functions (for the value of k).

[0184]

[0185] For each value of k = 1, 2, ...., K, the matrix B k By taking the initial point The Jacobian matrix formed by substituting the partial derivatives of F functions (for values ​​of k) with respect to d and θ (f = 1, 2, ..., F). Matrix B k It can be expressed in the following form:

[0186]

[0187] where, for f=1,2,...,F,

[0188]

[0189]

[0190] For each value of k=1,2,....,K, Represents the observation results The uncertainty in , and can be given by,

[0191]

[0192] in, It is the observation result The variance-covariance matrix of , which can be calculated based on the kth motion spectrum data set derived from the radar data measured by the kth radar sensor. More specifically, is the radial velocity measurement result The covariance matrix (f = 1, 2, ..., F) of each radial velocity measurement result includes the variance (or uncertainty) of each radial velocity measurement result and the correlation between different radial velocity measurement results. corresponds to the corresponding Doppler bin index in the kth motion spectrum data set, and the variance of each radial velocity measurement result can be based on Figure 5 Specifically, the variance value of each Doppler bin (of the kth motion spectrum data set) corresponding to the radial velocity measurement result can be determined using equation (5) described previously. The variance value determined for each radial velocity measurement result can be formed as diagonal elements of .

[0193] The angle measurement result The covariance matrix of , which may be calculated based on the kth motion spectrum data set derived from the radar data measured by the kth radar sensor. More specifically, Including the variance of each angle measurement result and the correlation between the angle measurement results. Equation (5) can be similarly adapted to determine the variance of each spatial bin corresponding to the angle measurement result in the k-th motion spectrum data set. Specifically, The diagonal elements of may be calculated by determining the variance of the distribution of radial velocities among the radial velocities indicated by the corresponding Doppler bin indexes for each spatial bin index in the k-th motion spectrum dataset. The variance is calculated by using the normalized spatial spectrum values ​​corresponding to the plurality of Doppler bin indices for the spatial bin index as the corresponding probability values ​​for the plurality of Doppler bin indices. and Characterizes the covariance between the radial velocity measurements and the angle measurements of the kth motion spectrum data set.

[0194] In solving equation (16), z, Δω, Δv x , Δv y When the value of x , Δv y Respectively with the starting point ω0, Add to get the estimate for the current iteration and Furthermore, the residual vector e can be calculated using equation (14). The observation result can also be calculated using the obtained residual vector e and equation (11) and An estimate of (for use in the next iteration of the estimation).

[0195] Although the foregoing example of the least squares adjustment method is based on the set of equations (6), it should be understood that the same method can be applied to the previously defined set of equations (7) or (8) when only some of the ego-motion parameters need to be calculated.

[0196] Although the present example uses a least squares adjustment method (using a Gauss-Helmert model) to perform the estimation of the ego-motion parameters, in alternative embodiments, a total least squares (TLS) method may alternatively be used.

[0197] Figure 8 Six different plots of motion spectrum data S corresponding to radar data acquired by six different radar sensors mounted on a host vehicle are illustrated, each radar sensor being mounted at a different location on the vehicle and scanning a different field of view around the vehicle. In each plot, the variation of the radial velocity with azimuth angle given by the high intensity values ​​in the plot is consistent with the motion equation (1).

[0198] Fig. 9 The solid white line illustrates the use of linear least squares estimation combined with iterative RANSAC processing (as in Figure 7 described) and fitted to Figure 8 The self-motion curves of each of the six motion spectrum data S plots in FIG. Moreover, for comparison, Fig. 9 Each of the graphs in FIG. 1 also shows the self-motion curve measured by the motion sensor on the vehicle as a dotted line. Fig. 9 As shown, the self-motion curve obtained using least squares estimation for the motion spectrum data S substantially matches the self-motion curve measured by the motion sensor.

[0199] Fig.10 The comparison between the self-motion information estimated using linear least squares estimation and the self-motion information obtained via other methods is illustrated. More specifically, Fig.10The velocity and yaw rate estimated based on linear least squares estimation combined with iterative MSAC processing are illustrated (in Fig.10 ) with the speed and yaw rate obtained using GPS (“GPS filter speed / yaw rate”), the speed and yaw rate obtained using the host vehicle motion sensor (“host vehicle speed / yaw rate”), and the speed and yaw rate obtained using the instantaneous radar motion (“IRM speed / yaw rate”).

[0200] Fig.11 The comparison of self-motion information estimated using the Gauss-Helmert model and self-motion information obtained via other methods is illustrated. More specifically, Fig.11 The velocity and yaw rate estimated using the aforementioned nonlinear Gauss-Helmert model are illustrated (in Fig.10 ) with the speed and yaw rate obtained using GPS (“GPS filter speed / yaw rate”), the speed and yaw rate obtained using the host vehicle motion sensor (“host vehicle speed / yaw rate”), and the speed and yaw rate obtained using the instantaneous radar motion (“IRM speed / yaw rate”).

[0201] Fig.12 The example shows motion spectrum data generated from a two-dimensional array having a plurality of linear sub-arrays. Specifically, since the elevation angle is not considered in the generation of the motion spectrum data, the motion spectrum data can also be calculated using each individual sub-array of the two-dimensional array. Fig.12 As shown, in this scenario, for each Doppler bin, there is an ambiguity in the angle measurement due to the presence of two strong peaks corresponding to the two respective curves. In order to determine the vehicle ego-motion from the motion spectrum data, only one of the two curves should be selected and used to estimate the ego-motion parameters using the above method. Moreover, the ambiguity can be resolved by analyzing known motion behavior or measuring motion from another sensor in order to select the correct curve.

[0202] although Figure 1A The self-motion information determination module 8 in the motion spectrum acquisition module 6 uses the above reference Figure 3 The motion spectrum data set S is generated by the process described above, but the motion spectrum acquisition module can instead receive the motion spectrum data set from an external source. In this way, the device 10 does not need to include Figure 1A The data acquisition module 2 and the Doppler processing module 4 are illustrated in the embodiment, and may have a variation of the motion spectrum acquisition module 6. Fig.13AA device 10 ′ according to the second example embodiment is illustrated, which includes a modified motion spectrum acquisition module 6 ′ and the self-motion information determination module 8 of the first example embodiment, but does not include the data acquisition module 2 and the Doppler processing module 4 of the first example embodiment.

[0203] Fig. 13B A method of determining ego-motion information of a vehicle 30 by the radar data processing device 10 ′ of the second example embodiment is summarized.

[0204] exist Fig. 13B In step S410, Fig.13A The motion spectrum acquisition module 6' acquires a motion spectrum data set based on radar data generated by the radar sensor 20. The motion spectrum data includes a plurality of data elements, each of which is calculated for a corresponding Doppler bin index j in a plurality of Doppler bin indexes and for a corresponding spatial bin index n in a plurality of spatial bin indexes. Each spatial bin index n indicates a corresponding arrival angle of the radar echo signal relative to the axis of the radar sensor 20.

[0205] exist Fig. 13B In step S420, Fig.13A The ego-motion information determination module 8 determines the ego-motion information by solving a set of motion equations including a plurality of motion equations generated using the motion spectrum data set. Each motion equation involves a corresponding value indicating a radial velocity of the stationary object relative to the radar sensor 20, a corresponding value indicating an angular displacement of the stationary object relative to the axis of the radar sensor 20, and a variable indicating the velocity of the vehicle 30.

[0206] The example aspects described herein circumvent specific computer technology-based limitations associated with the estimation of ego-motion information of a moving vehicle that may hinder the implementation of various advanced driver assistance systems (ADAS) and autonomous driving applications. In particular, ego-motion measured by an inertial measurement unit (IMU) or estimated using measurements obtained by another type of sensor (such as a camera or radar) is often subject to inherent errors. According to the example aspects described herein, a motion spectrum data set for estimating vehicle ego-motion information is obtained, and the ego-motion information of the vehicle is determined by solving a system of motion equations including multiple motion equations generated using the motion spectrum data set. The values ​​indicating radial velocity and angular position in the system of motion equations are pre-calculated based on bin indices in the motion spectrum data set and independently of radar data. As a result, the complexity of determining ego-motion information is reduced because many terms in the system of motion equations are pre-calculated and may only need to be calculated once. Moreover, the use of the motion spectrum data set allows the determination of ego-motion information without having to first process the radar data for detection in the Doppler angle domain and then apply an angle measurement algorithm on a separate beam vector corresponding to the detection. Using a motion spectrum dataset to estimate ego-motion information can be advantageous because the quality of the estimate does not depend on the number of individual radar detections. Moreover, based on the aforementioned computer-based capabilities of the example aspects described herein, the example aspects described herein improve computers and computer processing / functionality, and also improve at least the field of ego-motion determination and various advanced driver assistance systems (ADAS) and autonomous driving applications.

[0207] In the foregoing description, example aspects are described with reference to a number of example embodiments. Therefore, this description should be considered illustrative rather than restrictive. Similarly, the diagrams illustrated in the accompanying drawings highlighting the features and advantages of the example embodiments are presented for illustrative purposes only. The architecture of the example embodiments is sufficiently flexible and configurable that it can be utilized (and navigated) in a manner other than that shown in the accompanying drawings.

[0208] In an example embodiment, the software implementation of the examples presented herein may be provided as a computer program or software (such as one or more programs having instructions or instruction sequences), which is included or stored in an article such as a machine-accessible or machine-readable medium, an instruction storage unit, or a computer-readable storage device, each of which may be non-transitory. Programs or instructions on a non-transitory machine-accessible medium, a machine-readable medium, an instruction storage unit, or a computer-readable storage device may be used to program a computer system or other electronic device. Machine or computer-readable media, instruction storage units, and storage devices may include, but are not limited to, floppy disks, optical disks, and magneto-optical disks, or other types of media / machine-readable media / instruction storage units / storage devices suitable for storing or transmitting electronic instructions. The techniques described herein are not limited to any particular software configuration. They may find applicability in any computing or processing environment. As used herein, the terms "computer-readable", "machine accessible medium", "machine-readable medium", "instruction storage" and "computer-readable storage device" shall include any medium that can store, encode or transmit instructions or sequences of instructions for execution by a machine, computer or computer processor and cause the machine / computer / computer processor to perform any of the methods described herein. Furthermore, it is common in the art to refer to software in one form or another (e.g., program, procedure, process, application, module, unit, logic, etc.) as taking an action or causing a result. Such expressions are merely a shorthand way of stating that execution of the software by a processing system causes the processor to perform an action to generate a result.

[0209] Some embodiments may also be implemented by the preparation of application specific integrated circuits, field programmable gate arrays, or by interconnecting an appropriate network of conventional component circuits.

[0210] Some embodiments include computer program products. A computer program product may be a storage medium, instruction storage unit, or storage device storing instructions that may be used to control or cause a computer or computer processor to perform any of the processes in the example embodiments described herein. The storage medium / instruction storage unit / storage device may include, for example but not limited to, an optical disk, ROM, RAM, EPROM, EEPROM, DRAM, VRAM, flash memory, flash memory card, magnetic card, optical card, nanosystem, molecular-level memory integrated circuit, RAID, remote data storage device / archive / warehousing, and / or any other type of device suitable for storing instructions and / or data.

[0211] Some implementations stored on a computer-readable medium, instruction store, or storage device include software for both controlling the system's hardware and enabling the system or microprocessor to interact with a human user or other mechanism using the results of the example implementations described herein. Such software may include, but is not limited to, device drivers, operating systems, and user applications. Ultimately, such computer-readable media or storage devices also include software for executing the example aspects herein, as described above.

[0212] Included in the programming and / or software of the system are software modules for implementing the processes described herein. In some example embodiments herein, the modules include software, but in other example embodiments herein, the modules include hardware, or a combination of hardware and software.

[0213] Although various example embodiments have been described above, it should be understood that they are presented as examples rather than limitations. It will be apparent to those skilled in the relevant art that various changes in form and detail may be made therein. Therefore, the present disclosure should not be limited to any of the example embodiments described above, but should only be limited according to the attached claims and their equivalents.

[0214] In addition, the purpose of the abstract is to enable patent offices and the public, especially scientists, engineers, and practitioners in the field who are not familiar with patent or legal terms or phraseology, to quickly determine the nature and essence of the technical disclosure of the present application from a cursory inspection. The abstract is not intended to limit the scope of the example embodiments presented herein in any way. It is also understood that the processes recited in the claims need not be performed in the order presented.

[0215] Although this specification contains many specific implementation details, these details should not be interpreted as limiting, but as a description of features specific to the particular implementation described herein. Certain features described in this specification in the context of a separate implementation can also be implemented in combination in a single implementation. On the contrary, the various features described in the context of a single implementation can also be implemented in multiple implementations individually or in any appropriate sub-combination. In addition, although the features are described above as working in certain combinations, and even claimed to be protected from the beginning, one or more features from the claimed combination can be removed from the combination in some cases, and the claimed combination can be directed to a sub-combination or a variation of the sub-combination.

[0216] In some circumstances, multitasking and parallel processing may be advantageous. In addition, the separation of various components in the above-described embodiments should not be understood as requiring such separation in all embodiments, but it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

[0217] Now that some illustrative embodiments and implementations have been described, it is apparent that the foregoing embodiments, which have been presented by way of example, are illustrative and not restrictive. In particular, although many of the examples presented herein relate to specific combinations of devices or software elements, these elements may be combined in other ways to achieve the same purpose. Actions, elements, and features discussed only in conjunction with one embodiment are not intended to be excluded from similar roles in other one or more embodiments.

[0218] The devices and computer programs described herein may be embodied in other specific forms without departing from their characteristics. The foregoing embodiments are illustrative rather than limiting of the described systems and methods. Thus, the scope of the devices and computer programs described herein is indicated by the appended claims rather than the foregoing description, and all changes within the meaning and scope of the equivalents of the claims are included herein.

Claims

1. A method for determining self-motion information of a vehicle (30), the vehicle (30) comprising a radar sensor (20) having a plurality of antenna elements, the method comprising the following steps: receiving (S10) radar data (R), the radar data comprising a corresponding plurality of data samples of each of a plurality of radar echo signals received at each of the antenna elements of the radar sensor (20); The data samples are processed (S20) to generate, for each of the antenna elements, corresponding Doppler processed data (D) comprising one of: a plurality of data values, each data value in the plurality of data values ​​being calculated for a corresponding range bin index (i) in the plurality of range bin indices and for a corresponding Doppler bin index (j) in the plurality of Doppler bin indices; as well as a plurality of data values, each data value of the plurality of data values ​​being calculated for a corresponding fast time index of the plurality of fast time indices and for a corresponding Doppler bin index (j) of the plurality of Doppler bin indices; Acquiring a motion spectrum data set based on radar data generated by the radar sensor (20), the motion spectrum data set comprising a plurality of data elements, each of the data elements being calculated for a corresponding Doppler bin index (j) among a plurality of Doppler bin indexes and for a corresponding spatial bin index (n) among a plurality of spatial bin indexes, each spatial bin index (n) indicating a corresponding arrival angle of a radar return signal relative to an axis of the radar sensor (20), The motion spectrum dataset (S) is obtained (S30) by performing the following processing on each Doppler bin index (j) in the plurality of Doppler bin indices: Selecting (S30-1) a data value of the Doppler processed data that has been calculated for the Doppler bin index (j); calculating (S30-2) a covariance matrix using said data values ​​selected for that Doppler bin index (j); and applying (S30-3) a spectral estimation algorithm, the spectral estimation algorithm using the covariance matrix to determine, for the Doppler bin index (j), a corresponding spatial spectrum value for each of the plurality of spatial bin indices; and The ego-motion information is determined by solving a system of motion equations including a plurality of motion equations generated using the motion spectrum data set, each motion equation involving a respective value indicating a radial velocity of a stationary object (40) relative to the radar sensor (20), a respective value indicating an angular displacement of the stationary object (40) relative to an axis of the radar sensor (20), and a variable indicating a velocity of the vehicle (30).

2. The method according to claim 1, wherein: The individual equations of motion have the following form: d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x +sin(θ+θ M )v y Wherein, d indicates the radial velocity of the stationary object (40), θ indicates the angular displacement of the stationary object (40) relative to the axis of the radar sensor (20), θ M indicating an installation angle of the radar sensor (20), the installation angle being an angle between an axis of the radar sensor (20) and an axis of a vehicle coordinate system of the vehicle (30), l x indicating the installation position of the radar sensor (20) along the X-axis of the vehicle coordinate system, and y indicating the installation position of the radar sensor (20) along the Y axis of the vehicle coordinate system, v x indicates the X component of the velocity of the vehicle (30), and v y indicative of a Y component of the speed of the vehicle (30), ω indicates the yaw rate of the vehicle (30), and The self-motion information includes v x 、v y and the value of ω.

3. The method according to claim 2, wherein: K radar sensors (20) are installed at K different positions on the vehicle (30), each radar sensor having a plurality of antenna elements, wherein K is an integer greater than or equal to 2, and wherein a corresponding motion spectrum data set (S) is obtained for each of the K radar sensors, each corresponding motion spectrum data set (S) is based on radar data generated by the radar sensor (20), and wherein the step of determining the self-motion information comprises: estimating v by solving the following set of motion equations x 、v y And the value of ω: D=H·E in, d k =[d k，1 , d k，2 …, d k，F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k，1 , d k，2 …, d k，F indicating F radial velocity values ​​corresponding to F corresponding Doppler bin indices in a kth motion spectrum data set (S) acquired for a kth radar sensor (20), Among them, M k is an F×U dimensional matrix containing the data elements in the k-th motion spectrum data set (S), k=1, 2, ..., K, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1, 2, ..., K, respectively, where M k Each row index of M corresponds to a corresponding Doppler bin index in the k-th motion spectrum dataset (S), and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset (S), where M k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum data set (S), in, Among them, θ k，1 ,θ k，2 …, θ k，U indicates U angle values ​​corresponding to U spatial bin indices in the k-th motion spectrum dataset (S), u=1, 2, ..., U, respectively, and wherein, indicating an installation angle of a kth radar sensor (20), the installation angle being an angle between an axis of the kth radar sensor (20) and an axis of a vehicle coordinate system of the vehicle (30), in, is the model matrix of the kth radar sensor (20), k=1, 2, ..., K, where indicating the installation position of the kth radar sensor (20) along the X-axis of the vehicle coordinate system, and Indicates the installation position of the kth radar sensor (20) along the Y axis of the vehicle coordinate system.

4. The method according to claim 1, wherein: The individual equations of motion are given by: Wherein, d indicates the radial velocity of the stationary object (40), θ indicates the angular displacement of the stationary object (40) relative to the axis of the radar sensor (20), θ M indicating the angle between the axis of the radar sensor (20) and the axis of the vehicle coordinate system of the vehicle (30), indicative of an X component of the velocity of the radar sensor (20), and indicative of a Y component of the velocity of the radar sensor (20), The ego-motion information includes a setting value when the yaw rate of the vehicle (30) is zero and Among them, v x indicates the X component of the velocity of the vehicle (30), and v y a Y component indicative of the velocity of the vehicle (30), wherein K radar sensors (20) are installed at K different positions on the vehicle (30), each radar sensor having a plurality of antenna elements, wherein K is an integer greater than or equal to 1, and wherein a corresponding motion spectrum data set (S) is obtained for each of the K radar sensors, each corresponding motion spectrum data set being based on radar data generated by the radar sensor (20), and wherein the step of determining the self-motion information comprises: estimating and , and set and Among them, it is estimated and The steps to find the value of include solving the following system of motion equations: D=H·E in, d k =[d k，1 , d k，2 …, d k，F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k，1 , d k，2 …, d k，F Indicates the radial velocity values ​​corresponding to the corresponding Doppler bin indices in the k-th motion spectrum data set (S) acquired for the k-th radar sensor (20), Among them, M k is an F×U dimensional matrix containing the data elements in the k-th motion spectrum data set (S), k=1, 2, ..., K, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1, 2, ..., K, respectively, where M k Each row index of M corresponds to a corresponding Doppler bin index in the k-th motion spectrum dataset (S), and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset (S), where M k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum data set (S), in, Among them, θ k，1 ,θ k，2 …, θ k，U Indicates U angle values ​​corresponding to U spatial bin indices in the k-th motion spectrum data set, u=1, 2, ..., U, where, Indicating an installation angle of a kth radar sensor, the installation angle being an angle between an axis of the kth radar sensor (20) and an axis of a vehicle coordinate system of the vehicle (30).

5. The method according to claim 1, wherein: The individual equations of motion are given by: d=(-l y cos(θ+θ M )+l x sin(θ+θ M ))ω+cos(θ+θ M )v x Wherein, d indicates the radial velocity of the stationary object (40), θ indicates the angular displacement of the stationary object (40) relative to the axis of the radar sensor (20), θ M indicating the angle between the axis of the radar sensor (20) and the axis of the vehicle coordinate system of the vehicle (30), l x indicating the installation position of the radar sensor (20) along the X-axis of the vehicle coordinate system, and y indicating the installation position of the radar sensor (20) along the Y axis of the vehicle coordinate system, ω indicates the yaw rate of the vehicle (30), v x is the X component of the velocity of the vehicle (30), The self-motion information includes ω and v x The value of wherein K radar sensors (20) are installed at K different positions on the vehicle (30), each radar sensor having a plurality of antenna elements, wherein K is an integer greater than or equal to 1, wherein a corresponding motion spectrum data set (S) is obtained for each of the K radar sensors, each corresponding motion spectrum data set being based on radar data generated by the radar sensor (20), and wherein the step of determining the ego-motion information comprises: estimating ω and v by solving the following set of motion equations: x Values: D=H·E in, d k =[d k，1 , d k，2 ..., d k，F ] T , k = 1, 2, ...., K, where F is an integer equal to or greater than 1, and d k，1 , d k，2 …, d k，F indicating F radial velocity values ​​corresponding to F corresponding Doppler bin indices in a kth motion spectrum data set (S) acquired for a kth radar sensor (20), Among them, M k is an F×U dimensional matrix containing the data elements in the k-th motion spectrum data set (S), k=1, 2, ..., K, where U is an integer greater than or equal to one, and the values ​​of F and U are selected for each of k=1, 2, ..., K, respectively, where M k Each row index of M corresponds to a corresponding Doppler bin index in the k-th motion spectrum dataset (S), and M k Each column index of M corresponds to the corresponding spatial bin index in the k-th motion spectrum dataset (S), where M k Each row of includes a data element of a corresponding Doppler bin index in the k-th motion spectrum data set (S), in, Among them, θ k，1 ,θ k，2 …, θ k，U Indicates U angle values ​​corresponding to U corresponding spatial bin indices in the k-th motion spectrum data set (S), u=1, 2, ..., U, where, indicating an installation angle of a kth radar sensor (20), the installation angle being an angle between an axis of the kth radar sensor (20) and an axis of a vehicle coordinate system of the vehicle (30), in, is the model matrix of the kth radar sensor (20), k=1, 2, ..., K, where indicating the installation position of the kth radar sensor (20) along the X-axis of the vehicle coordinate system, and Indicates the installation position of the kth radar sensor (20) along the Y axis of the vehicle coordinate system.

6. The method according to any one of claims 3 to 5, wherein: The system of motion equations is solved using a linear least squares estimation method to determine the ego-motion information.

7. The method according to claim 6, wherein: The self-motion information is determined by the following steps: Randomly select a plurality of Doppler bin index subsets from the Doppler bin indices in the K motion spectrum datasets (S); The following processing is performed for each Doppler bin index subset of the plurality of Doppler bin index subsets: By dividing the d where k = 1, 2, ..., K k，1 , d k，2 …, d k，F is set to the radial velocity value corresponding to the corresponding Doppler bin index in the k-th motion spectrum data set (S) in the Doppler bin index subset, and by k The rows of are set to the data elements of the corresponding Doppler bin index in the Doppler bin index subset in the k-th motion spectrum data set (S), and the motion equation set D=H·E is solved using the linear least squares estimation method to estimate (S320-1) the self-motion information, determining (S320-2) an estimated error of each Doppler bin index in the Doppler bin index subset; as well as Based on the determined estimation error, classifying (S320-3) each Doppler bin index (i) in the Doppler bin index subset as one of a normal Doppler bin index (i) and an abnormal Doppler bin index (i); A subset of Doppler bin indices is selected from the plurality of subsets of Doppler bin indices based on one of: the number of normal Doppler bin indices for each Doppler bin index subset, and the sum of said estimated errors of all normal Doppler bin indices determined for each subset of Doppler bin indices, and By dividing the d where k = 1, 2, ..., K k，1 , d k， 2…, d k，F = is set to the radial velocity value corresponding to the corresponding Doppler bin index classified as the normal Doppler bin index in the selected Doppler bin index subset in the kth motion spectrum data set (S), and also by M where k=1, 2, ...., K k The individual rows of are set to data elements of the corresponding Doppler bin index classified as the normal Doppler bin index in the selected Doppler bin index subset in the kth motion spectrum data set (S), and the motion equation group D=H·E is solved using the linear least squares estimation method to determine the self-motion information.

8. The method according to claim 1, wherein: Each of the data values ​​of the Doppler processed data (D) is calculated for a corresponding range bin index (i) of a plurality of range bin indices and for a corresponding Doppler bin index (i) of a plurality of Doppler bin indices, and wherein the data value selected for each Doppler bin index (i) defines an R data set including data values ​​calculated for R antenna elements and for I range bin indices for the Doppler bin index (i). × I-dimensional matrix, and wherein the covariance matrix is ​​an R×R dimensional matrix determined from the R×I dimensional matrix using the product of the R×I dimensional matrix and the conjugate transpose of the R×I dimensional matrix, or Each of the data values ​​of the Doppler processed data (D) is calculated for a corresponding fast time index of a plurality of fast time indices and a corresponding Doppler bin index (i) of a plurality of Doppler bin indices, and wherein the data value selected for each Doppler bin index (i) defines an R data set comprising data values ​​calculated for R antenna elements and for I fast time indices of the Doppler bin index (i). × I-dimensional matrix, and wherein the covariance matrix is ​​an R×R dimensional matrix determined from the R×I dimensional matrix using the product of the R×I dimensional matrix and the conjugate transpose of the R×I dimensional matrix.

9. The method according to claim 1 or 8, wherein: The step of generating (S30) the motion spectrum data set (S) further comprises: normalizing (S210) the spatial spectrum values ​​determined for each Doppler bin index (i) to generate a corresponding plurality of normalized spatial spectrum values, each of the normalized spatial spectrum values ​​being generated for a corresponding Doppler bin index (i) in the plurality of Doppler bin indices and for a corresponding spatial bin index (n) in the plurality of spatial bin indices.

10. The method according to claim 9, wherein: The step of generating the motion spectrum data set (S) further comprises: For each Doppler bin index (i) among the plurality of Doppler bin indices, by using the normalized spatial spectrum value generated for the spatial bin index as a probability value for the spatial bin index, calculating (S220) data indicating a variance of a probability distribution of an angle of arrival of the radar return signal relative to the axis of the radar sensor (20) among the angles of arrival indicated by the corresponding spatial bin index; and determining (S230) a subset of Doppler bin indices from the plurality of Doppler bin indices for which the calculated variance is below a predetermined threshold, The value of each data element in the motion spectrum data set (S) is based on the corresponding normalized spatial spectrum value generated for the Doppler bin index (i) in the Doppler bin index subset and for the spatial bin index (n) corresponding to the data element.

11. The method according to claim 10, wherein: The value of the data element in the motion spectrum data set (S) is generated for the Doppler bin index (i) by processing the plurality of normalized spatial spectrum values ​​generated for each Doppler bin index (i) in the Doppler bin index subset according to the following steps: identifying (S240-1) a spatial bin index (n) for which the normalized spatial spectrum value is the highest among the normalized spatial spectrum values ​​generated for the plurality of spatial bin indices; Setting (S240-2) the value of the data element of the identified spatial bin index (n) to a non-zero value; as well as The values ​​of the data elements of all other spatial bin indexes are set (S240-3) to zero.

12. The method according to claim 9, wherein: The value of the data element in the motion spectrum data set (S) is generated for each Doppler bin index (j) in the plurality of Doppler bin indices by processing the plurality of normalized spatial spectrum values ​​generated for the respective Doppler bin index (j) according to the following steps: identifying a spatial bin index (n) for which the normalized spatial spectral value is highest among the normalized spatial spectral values ​​generated for the plurality of spatial bin indices; Setting the value of the data element of the identified spatial bin index (n) to a non-zero value; as well as Set the values ​​of the data elements of all other spatial bin indices to zero.

13. The method according to claim 9, wherein: The step of determining the self-motion information comprises: solving the set of motion equations using a least squares adjustment method, wherein the least squares adjustment method performs at least one iteration of an iterative process, wherein in each iteration of the iterative process, an increment of the estimated value of the self-motion information is calculated to obtain an updated estimated value, wherein the updated estimated value is based on the estimated value and the increment and is used to calculate an increment in a next iteration of the iterative process, Wherein, the increment is calculated using the following terms: a first value set of the motion spectrum data set (S), each value in the first value set indicating a variance of a distribution of the angles of arrival among the angles of arrival indicated by the corresponding spatial bin index for the Doppler bin index (j) in the motion spectrum data set (S), wherein each value in the first value set is calculated using the normalized spatial spectrum values ​​corresponding to the plurality of spatial bin indices for the Doppler bin index (j) as corresponding probability values ​​for the plurality of spatial bin indices, and A second value set of the motion spectrum data set (S), each value in the second value set indicating a variance of a distribution of the radial velocities among radial velocities indicated by the corresponding Doppler bin index for the spatial bin index (n) in the motion spectrum data set (S), wherein each value in the second value set is calculated using the normalized spatial spectrum values ​​corresponding to the plurality of Doppler bin indices for the spatial bin index (n) as corresponding probability values ​​for the plurality of Doppler bin indices.

14. A radar data processing device, the radar data processing device being configured to determine ego-motion information of a vehicle (30) comprising a radar sensor (20) having a plurality of antenna elements, the radar data processing device (10) comprising: a data acquisition module (2), the data acquisition module being configured to receive radar data (R), the radar data comprising a corresponding plurality of data samples of each of a plurality of radar echo signals received at each of the antenna elements of the radar sensor (20); A Doppler processing module (4) is arranged to process the data samples to generate corresponding Doppler processed data (D) for each of the antenna elements, comprising one of the following: a plurality of data values, each data value in the plurality of data values ​​being calculated for a corresponding range bin index (i) in the plurality of range bin indices and for a corresponding Doppler bin index (j) in the plurality of Doppler bin indices; as well as a plurality of data values, each data value of the plurality of data values ​​being calculated for a corresponding fast time index of the plurality of fast time indices and for a corresponding Doppler bin index (j) of the plurality of Doppler bin indices; A motion spectrum acquisition module (6), the motion spectrum acquisition module being configured to acquire a motion spectrum data set (S) based on radar data generated by the radar sensor (20), the motion spectrum data set comprising a plurality of data elements, each of the data elements being calculated for a corresponding Doppler bin index (j) in a plurality of Doppler bin indexes and for a corresponding spatial bin index (n) in a plurality of spatial bin indexes, each spatial bin index (n) indicating a corresponding arrival angle of a radar echo signal relative to an axis of the radar sensor (20), The motion spectrum acquisition module (6) is configured to acquire the motion spectrum data set (S) by performing the following processing on each Doppler bin index (j) in the plurality of Doppler bin indexes: Selecting (S30-1) a data value of the Doppler processed data that has been calculated for the Doppler bin index (j); calculating (S30-2) a covariance matrix using said data values ​​selected for that Doppler bin index (j); and applying (S30-3) a spectral estimation algorithm, the spectral estimation algorithm using the covariance matrix to determine, for the Doppler bin index (j), a corresponding spatial spectrum value for each of the plurality of spatial bin indices; and A self-motion information determination module (8) is configured to determine the self-motion information by solving a set of motion equations including a plurality of motion equations generated using the motion spectrum data set (S), each motion equation involving a corresponding value indicating a radial velocity of a stationary object (40) relative to the radar sensor (20), a corresponding value indicating an angular displacement of the stationary object (40) relative to an axis of the radar sensor (20), and a variable indicating a velocity of the vehicle (30).

Images