Radar system

Abstract

Radar system with antenna array (18) comprising multiple transmitting antennas (16) and / or receiving antennas (18), and with a control and evaluation unit (26) configured to transmit radar signals and receive radar echoes from radar targets and to detect the radar targets in a detection space having at least two spatial dimensions and at least two velocity dimensions, characterized in that the control and evaluation unit (26) is configured to calculate, for selected points in the detection space, assuming that a radar target is located at the point, for several combinations of transmitting and receiving antennas using a model, an expected measurement result for each point and to determine the detection data of the detected targets by comparing the calculated measurement results with measurement results derived from the signals actually received.

Description

[0001] The invention relates to a radar system with an antenna array having a plurality of transmitting antennas and / or a plurality of receiving antennas, and with a control and evaluation device configured to transmit radar signals and receive radar echoes from radar targets and, based on the received signals, to detect the radar targets in a detection space having at least two spatial dimensions and at least two speed dimensions.

[0002] In particular, the invention relates to a radar system for environmental detection in motor vehicles. The radar system may comprise one or more radar sensors. The antenna array is formed by the totality of the transmitting and receiving antennas of all radar sensors belonging to the system. State of the art

[0003] Radar sensors are used in motor vehicles to implement comfort functions such as adaptive cruise control and safety features such as emergency braking assistance. The key advantage of such sensors is the direct measurement of physical quantities instead of, for example, interpreting images from a video camera. The radar sensors transmit high-frequency radar beams via the antenna array and receive the beams reflected from objects. The detected objects can be stationary or moving. Using the received radar beams, the distance and direction (angle) to the object can be calculated. Furthermore, the speed of an object relative to the radar sensor can be calculated. The radar sensors typically operate in a frequency range between 76 and 81 GHz.

[0004] A well-known method for radar modulation is the so-called chirp sequence method. This method involves transmitting many fast frequency ramps, called chirps. The distance resolution (Δd) depends on the used bandwidth B, and the relative velocity resolution (Δv) depends on the measurement duration T: Δd=c / 2B Δv=c / (2f0⋅T)

[0005] Where c denotes the speed of light and f0 the center frequency of the frequency ramp.

[0006] Examples of radar systems of this type are described in DE 10 2018 202 289 A1, DE 10 2018 202 290 A1, and DE 10 2018 202 293 A1.

[0007] The performance of environmental detection can be significantly increased through the use of cooperative radar sensor networks. In this case, the data from multiple radar sensors is evaluated jointly, enabling both increased sensitivity and increased accuracy of environmental detection, as a larger antenna array and more RF channels are available than with a single sensor. In particular, the antenna array can also include virtual antennas, which are created when a receiving antenna receives signals from spatially offset transmitting antennas in time, frequency, or code division multiplexing. However, with very large virtual antenna arrays and simultaneously large bandwidths and measurement times, the individual antennas measure the target (especially in the near field) at very different distances, relative velocities, and angles, making joint evaluation of the entire virtual antenna array difficult.

[0008] In cooperative radar sensor networks, a distinction is made between systems with coherent distribution of the RF oscillator signal and those without. Systems with RF oscillator signal distribution exhibit significantly better phase noise performance, but the distribution of the oscillator signal is associated with considerable effort, so depending on the application, a solution without oscillator signal distribution may be advantageous for practical implementation. Disclosure of the invention

[0009] The object of the invention is to create a radar system with improved performance.

[0010] This object is achieved according to the invention in that the control and evaluation device is configured to calculate an expected measurement result for selected points in the detection area, assuming that a radar target is located at the point, for several combinations of transmitting and receiving antennas using a model and to determine the location data of the detected targets by comparing the calculated measurement results with measurement results derived from the actually received signals.

[0011] The invention has the advantage that the effects on the received signals resulting from the different pairs of transmitting and receiving antennas or - Arrays (more precisely, their phase centers) that have different distances, relative velocities, and angular orientations to the radar target are compensated using a common model, so that the positioning data (position coordinates and relative velocity vectors of the radar targets) can be specified directly in a coordinate system fixed with respect to the antenna array. By dividing the positioning space into a (not necessarily uniform) grid of cells, the model can be used to predict, with a resolution determined by the fineness of the grid, for each set of positioning data describing a given target, how the signals received by the various virtual antennas and the measurement results derived from them (distances, relative velocities, angles) will differ from one another.For each hypothesis about a detected radar target, it is then possible to check whether the actually received signals match the predicted values. This significantly increases the reliability of object detection. This type of signal analysis can be combined with conventional analysis methods in a variety of ways, allowing a reasonable compromise between computational effort and measurement accuracy to be achieved for each task.

[0012] Advantageous embodiments of the invention are specified in the subclaims.

[0013] In one embodiment, an expected received signal is calculated for each cell of the location space, and the correlation of the measured received signals with the calculated received signals is evaluated. For example, the calculated values ​​for the expected received signals can form the parameters of a matched filter (correlation filter), which is then applied to the actually received signals to increase the signal-to-noise ratio. Optionally, the required computational effort can be reduced, for example, by working with a coarser cell grid in less relevant areas of the location space and / or by limiting the method to a low-dimensional subspace of the up to 6-dimensional location space.

[0014] In another embodiment, the antenna array is divided into several subarrays (where a subarray can be, for example, the antenna array of a single radar sensor), and the signals received in each subarray are first subjected to conventional FFT processing to determine preliminary values ​​for the positioning data (distances, radial velocities, angles). The results thus obtained can then be used to compensate for the differences (e.g., phase differences) between the signals received in the various subarrays, after which the compensated signals are summed coherently or non-coherently.

[0015] The spectra obtained from different subarrays exhibit a frequency offset, which at the points corresponding to the centers of the detection space cells is caused by the differences between the expected measured variables for distance, relative velocity, and angle. Therefore, the frequency offset can be compensated at these points. At the remaining points of the spectrum, at least approximate compensation is possible by interpolation between the grid points. The compensated signals can then be summed again coherently or non-coherently.

[0016] In yet another embodiment, conventional FFT processing is performed for individual subarrays, and conventional target detection is then performed using the resulting multidimensional spectrum, possibly with a lowered detection threshold. Slightly different positioning data will then be obtained for different subarrays, with the deviations corresponding to the prediction values ​​provided by the model. The model thus provides correction factors that enable a refined estimation of the coordinates in the positioning space.

[0017] The following examples are explained in more detail using the drawings. They show: Fig. 1 is a block diagram of a radar system according to the invention; Fig. 2 a sketch explaining a model used in the radar system; Fig. 3 - 5 diagrams to display predicted values ​​resulting from the model; and Fig. 6 - 8 Flowcharts for different evaluation methods that can be implemented in the radar system according to the invention.

[0018] In Fig. 1 schematically shows a radar system that has three radar sensors 10, 12, 14, which can be installed at different locations in a motor vehicle.

[0019] Each radar sensor has at least one transmitting antenna 16 and at least one receiving antenna 18. In practice, the number of transmitting and receiving antennas per radar sensor can be significantly larger than in the example shown here. The transmitting and receiving antennas 16, 18 of all radar sensors together form an antenna array 18, in which the antennas of each individual radar sensor 10, 12, 14 each form a subarray 20, 22, and 24, respectively.

[0020] A central control and evaluation device 26 controls the functions of the three radar sensors 10, 12, 14 and evaluates the signals received by these radar sensors in order to generate positioning data, namely location coordinates x, y and speed components v x , v y , to determine the position of detected radar targets.

[0021] Due to the central control and data evaluation, the radar sensors 10, 12, 14 form a cooperative radar system in which the results obtained from one radar sensor can be compared with the results of the other radar sensors in order to obtain more accurate and reliable positioning data.

[0022] In the example described here, radar sensors 10, 12, and 14 are FMCW radar sensors, in which the frequencies of the signals transmitted by transmitting antennas 16 are ramp-modulated. By mixing the signal received by a receiving antenna 18 with a portion of the signal transmitted at the same time, a beat signal (baseband signal) is obtained, the frequency of which depends on the radar signal's travel time from the sensor to the radar target and back. Fast Fourier transformation (FFT) is used to create a spectrum from this signal, in which each detected target appears as a peak at a specific frequency. This provides distance information about the detected objects.

[0023] Due to the Doppler effect, the frequency of the baseband signal also depends on the relative motion of the respective object. However, if the frequency ramps are sufficiently steep, this Doppler shift within a ramp is negligible. However, if multiple ramps (chirps) are transmitted and received consecutively, a measurable Doppler phase shift occurs from ramp to ramp. Therefore, by Fourier transforming the ramps, the radial relative velocity of each detected target can be determined in a two-dimensional (range Doppler) spectrum.

[0024] The angular information about the located targets is obtained by comparing the amplitudes and / or phases of the signals received by spatially offset receiving antennas 18 or transmitted by spatially offset transmitting antennas 16. In the example shown, the receiving antennas 18 of each sensor are offset only horizontally, so that only the azimuth angles of the targets can be determined. In this case, the measurable spatial coordinates and velocity components of the radar targets form a four-dimensional detection space with two spatial dimensions (horizontal plane) and two velocity dimensions. If the antennas of a subarray are also offset vertically, the elevation bars of the targets can also be determined, and the detection space has three spatial dimensions and three velocity dimensions.To simplify the presentation, only a four-dimensional location space is considered in this description.

[0025] If a subarray, such as subarrays 22 and 24, has multiple transmitting antennas 16, the difference between the signals received by the receiving antennas depends not only on the detection angle and the spatial separation of the receiving antennas, but also on the spatial separation between the receiving antennas and the transmitting antenna 16 that transmitted the signal. The azimuthal angular resolution of a radar sensor increases with the larger the aperture, i.e., the spatial extent of the array in the horizontal direction. If only one transmitting antenna is used for transmission, the aperture is determined only by the receiving antennas 18. In a MIMO (Multiple Input, Multiple Output) radar with multiple transmitting antennas 16, time, frequency, or code division multiplexing can be used to determine which transmitting antenna transmitted each received signal.In this case, the aperture of the antenna array is virtually enlarged by the multiple transmit antennas 16, because switching from one transmit antenna to another is equivalent to adding an additional (virtual) receive antenna. By applying the MIMO principle, the angular resolution can therefore be significantly increased.

[0026] In the example shown, a bistatic antenna concept is implemented for radar sensors 10, 12, 14, meaning that the transmitting antennas 16 are different from the receiving antennas 18. However, the radar system can also include radar sensors with a monostatic antenna concept, in which at least some of the antennas can operate as both transmitting and receiving antennas.

[0027] In the example shown, radar sensor 10 has a local oscillator 28, which provides the signal transmitted by transmitting antenna 16. Another oscillator 30 generates a high-frequency signal that is coherently distributed to the two other radar sensors 12, 14, so that the transmitting antennas 16 of these radar sensors transmit coherent signals, i.e., signals with a fixed and known phase relationship. This simplifies the evaluation of signals that are, for example, transmitted by a transmitting antenna 16 of radar sensor 12 and received by a receiving antenna 18 of radar sensor 14. By evaluating such cross-sensor signals, the effective aperture of antenna array 18 is considerably increased. Accordingly, a signal transmitted by a transmitting antenna 16 of radar sensor 12 and received by a receiving antenna 18 of radar sensor 10 can also be evaluated.However, the signal from the local oscillator 28 is then used to downconvert the received signal to the baseband, which is not coherent with the transmitted signal supplied by the oscillator 30, so that the signals transmitted by the radar sensor 112 or 14 may have greater phase noise.

[0028] The conventional analysis of radar signals described so far therefore primarily provides only a distance, a radial velocity, and a detection angle for each radar target. In the control and evaluation device 26, these data are converted into a Cartesian coordinate system fixed with respect to the antenna array 18, which defines the detection space with the dimensions x, y, v. x , v yHowever, the difficulty lies in the fact that the subarrays 20, 22, and 24 are relatively far apart, and therefore the distances, radial velocities, and detection angles measured by the various radar sensors differ from one another. This particularly complicates the evaluation of cross-sensor signals, which depend on the significantly different distances, radial velocities, and detection angles of the radar target to the various sensors.

[0029] To better manage these complications, the control and evaluation device 26 is configured to model the differences between the measured values ​​obtained in the various radar sensors 10, 12, 14 for different points in the detection area. The model used for this purpose will be described below with reference to the Fig. 2 will be explained.

[0030] In Fig.2 are in a Cartesian coordinate system with the coordinate axes x and y the position of a radar target 32 ​​and the positions P0 and P i second antennas of the antenna array 18. For simplification, it is assumed that the antennas are arranged as in Fig.1 are arranged on a horizontal line, which here forms the y-axis, so that the antenna positions can be numbered with an index i. Furthermore, it is assumed that the optical axes of the radar sensors, and thus the directions of the main lobes of the antennas, are parallel to each other and to the x-axis. These assumptions do not always hold in practice. However, generalization to situations in which the antenna positions are also offset from each other in the x-direction and / or the antennas have different orientations to the x-axis does not pose any difficulties. The origin of the coordinate system has been placed so that it coincides with the antenna position P0. This antenna position therefore has the coordinates (0, y0) = (0,0). The antenna position P i has the coordinates (0, y i ), and the radar target 32 ​​has the coordinates (x, y). It is understood that the Fig.The xy-plane shown in Figure 2 is a subspace of the overall four-dimensional positioning space. The two velocity dimensions are not shown here.

[0031] For the antennas in positions P0 and P i These can be receiving antennas 18 or transmitting antennas 16. First, it is assumed that the transmitting and receiving antennas are monostatic. This means that the signal transmitted by the antenna at P0 is reflected by the radar target 32 ​​and received again by the same antenna. The distance d0 is measured. Accordingly, the antenna at P i for the distance to radar target 32 ​​the value d i measured.

[0032] With the antenna at P0 and neighboring antennas (not shown) belonging to the same subarray, the angle θ0 at which the radar target 32 ​​is seen from position P0 can be determined. Accordingly, an angle θ idetermine under which the same radar target 32 ​​from position P i from.

[0033] For the antenna position P i measured values ​​of d i and θ i then the following relationships apply: di=x2+(y−yi)2 θi=atany−yix

[0034] If v x and v y the (not shown) velocity components of the radar target 32 ​​relative to the antenna array 18, then at the antenna position P i for the radial velocity v r,i of the radar target the value: vr,i=vx cos θi+vy sin θi measured and for the not directly measurable tangential velocity v t,i applies: vt,i=−vx sin θi+vy cos θi

[0035] If we now assume that the radar target 32 ​​is a stationary target, then the relative velocity vector with the components v xand v y opposite to the velocity vector of the vehicle in which the radar system is installed. Using the model given by equations (1) to (4), it is possible in principle to calculate for each point in the four-dimensional positioning space how much the antenna position P i measured distance, radial velocity and angle values ​​may differ from those measured at antenna position P0.

[0036] Fig.Figure 3 is a map of a relevant section of the xy-plane, showing the difference in distance d calculated from the model for each point in the plane. The distance difference, represented by different hatchings, is given here in bins, where the bin size corresponds to the spatial distance resolution, which is inversely proportional to the bandwidth B of the frequency-modulated radar signal. Different units are used for the x-axis and the y-axis. The units in the x-direction are given in 100 m, while the units in the y-direction are given in m.

[0037] To calculate the map, the location area (in Fig. 3 the xy subspace) was divided into cells 34, and the x and y coordinates of the center of each cell 34 were inserted into equations (1) to (4) of the model. The cells 34 are in Fig.3 is shown in a highly simplified form. It can be seen that the speed differences increase with increasing distance from the straight line y = 0.

[0038] Fig. Figure 4 shows a corresponding map for the calculated differences in radial velocity vr, which is again given in bins. The bin size here corresponds to the velocity resolution, which is inversely proportional to the measurement duration T.

[0039] Fig. Figure 5 finally shows a corresponding map for the calculated differences in the azimuth angle θ. The different hatchings indicate different angular deviations in °.

[0040] The one in the Fig. The cards shown in 3 to 5 are based on the following parameters: m / s Bandwidth B: 3 GHz corresponding to a distance bin size of 0.05 m Measurement duration T: 20 ms corresponding to a speed bin size of 0.1 Amount of offset of the two antennas: 1 m Angular alignment of the antennas relative to the driving axis: 0° Speed ​​of the vehicle over ground: 20 m / s

[0041] Corresponding maps can be calculated for any pairing of antenna positions as required.

[0042] In a MIMO system, or generally in a bi-static system, the position of the transmitting antenna must also be taken into account when comparing the measured values ​​at two receiving antenna positions. The distance measured at a receiving antenna position, for example, di, is the arithmetic mean of the distance from the transmitting antenna to the target and the distance from the target to the receiving antenna, with these two distances being calculated according to equation (1). For the radial velocity vr,i at a receiving antenna position Pi, the azimuth angle at the transmitting antenna location and the azimuth angle at the receiving antenna location are then inserted into equation (3), and the resulting velocity values ​​are arithmetically averaged.

[0043] However, when estimating the angle, the transmission angle θ tx and the reception angle θ rx be considered separately. If the function a tx,i(θ) is the complex one-way antenna characteristic of the transmitting antenna i and the function a rx,j (θ) is the complex one-way antenna characteristic of the receiving antenna j, then one obtains the complex amplitude of the received signal for a path from the transmitting antenna i with the transmission angle θ tx to receiving antenna j with the reception angle θ rx according to the formula: a=atx,i(θtx) arx,j(θrx)

[0044] The required one-way antenna characteristics can be determined using known methods for calibrating antenna patterns.

[0045] In general, when detecting the surroundings in a driver assistance system of a motor vehicle, not all regions of the tracking area have the same relevance, so the same accuracy requirements do not apply everywhere. To reduce computational effort, a coarser cell grid can be used in certain regions of the tracking space, so that the measured value differences only need to be calculated for a smaller number of grid points. It is also possible to restrict the method to a subspace of the tracking space, i.e., a space with lower dimensions. For example, in some cases it may be sufficient to execute the method only for the xy plane, but not for the velocity space.

[0046] In some cases, it may also be sufficient to combine several transmit antennas and / or several receive antennas into a subarray, so that the calculations only need to be performed for the antenna positions that correspond to the phase centers of the subarrays.

[0047] The essential steps of an evaluation method that can be implemented in the control and evaluation device 26 are shown in Fig.6 is shown as a flow chart. In step S1, for each cell 34 of the detection space, a set of received signals is calculated in the manner described above, which would be expected if a radar target were located at the center of the respective cell. The expected measurement results BINex (cell) can be represented by BINs in the dimensions of distance, radial velocity, and angle, or by BIN offsets between signals that are to be expected at different antenna positions. These expected BIN offsets are used in step S2 to create a matched filter with which the signals a actually measured at the respective antenna positions are filtered in order to increase the signal-to-noise ratio.

[0048] In an optional step S3, the number of cells considered and thus the data volume is reduced, for example, through mathematical operations such as maximum calculation or averaging. For example, for each position cell in the xy plane, there is a "column" of cells in both velocity dimensions. However, many of these velocity cells can often be discarded, leaving only a few velocity cells with the highest performance in the position cell. It is also conceivable to perform the evaluation using the method described here only for stationary targets, thus reducing the dimensionality of the detection space to the two spatial dimensions.

[0049] In step S4, object detection is then performed by determining an occupancy probability for each remaining cell in the detection space. In the simplest case, a binary occupancy map is distorted by threshold comparison, in which each cell contains either the value one (object present) or the value zero (empty). In an optional step S5, the detected velocities are used to predict how the occupancy map will change until the next measurement time, and / or to aggregate the occupancy map over time.

[0050] Fig. Figure 5 illustrates another possible evaluation method that requires less computational effort. In a step S10, for each of several subarrays, for example, subarrays 20, 22, 24 in Fig.1, the received signals a are processed in the dimensions of distance, Doppler, and angle using a conventional 3D FFT. Using the model given by equations (1) to (4), the expected differences between the 3D spectra calculated in step 10 are then calculated in step S11. In a subsequent step S12, the bin offsets obtained from the model are used to adjust the spectra to eliminate the differences between the measured quantities before the spectra are then summed incoherently.

[0051] Alternatively, in step 13, the three 3D spectra can be mapped to the expected measured values ​​d j , θ j , v r,j for each cell before the spectra are then summed coherently or non-coherently. That is, for each bin position in the dimensions d j , θ j , v r,jof the spectrum, a check is made to determine whether the corresponding point in the location space lies on a grid point, i.e., whether it is the center of a cell. If the point lies between two grid points, the deviations of the measured values ​​calculated for these two grid points are interpolated to correct the spectra accordingly.

[0052] The step S12 or S13 is then followed by an optional reduction step S14, a detection step S15 and an optional prediction step S16, which correspond to the steps S3 to S5 in Fig. 6 correspond.

[0053] If the spectra are only summed non-coherently in step S12 or step S13, radar sensors can also be included in the evaluation which do not receive a distributed RF oscillator signal but, like the radar sensor 10 in Fig.1, generate a local oscillator signal. However, due to the increased phase noise, it may be useful to combine only the signals obtained from monostatic phases.

[0054] Another variant of the evaluation procedure is in Fig. 8. Similar to step S10 in Fig. 7 are in Fig.8 In a step S20, the signals a obtained in the individual subarrays are subjected to 3D-FFT processing. This is immediately followed by a detection step S21, in which object detection takes place in the conventional manner for each of the spectra obtained in step S20, if necessary with a lowered detection threshold so that even weaker signals can still be evaluated. In a subsequent correction step S22, the model given by equations (1) to (4) is then used to combine the complex spectral data of the subarrays for each detected target and, using correction data from the model, to make a refined estimate of the coordinates of the radar targets in the detection space. The refined estimates for the detected targets are then output in step S23 and / or used in step S24 to update an occupancy map. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] DE 10 2018 202 289 A1

[0006] DE 10 2018 202 290 A1

[0006] DE 10 2018 202 293 A1

[0006]

Claims

[1] Radar system with an antenna array (18) having a plurality of transmitting antennas (16) and / or receiving antennas (18), and with a control and evaluation device (26) which is configured to transmit radar signals and to receive radar echoes from radar targets (32) and to detect the radar targets in a detection space having at least two spatial dimensions and at least two speed dimensions on the basis of the received signals, characterized in that the control and evaluation device (26) is configured to calculate an expected measurement result for selected points in the detection space, assuming that a radar target is located at the point, for a plurality of combinations of transmitting and receiving antennas on the basis of a model, and to determine the location data of the detected targets by comparing the calculated measurement results with measurement results derived from the actually received signals. [2] Radar system according to claim 1, wherein the detection space or at least a subspace thereof is divided into a grid of cells and the selected points are the centers of the cells. [3] Radar system according to claim 2, wherein the size of the cells is different for different regions in the detection space. [4] Radar system according to claim 2 or 3, in which a correlation between the actually received signals and the signals resulting from the model is examined in order to compare the measurement results. [5] Radar system according to claim 4, wherein the evaluation of the received signals is carried out using a matched filter approach. [6] Radar system according to one of claims 3 to 5, in which an occupancy probability is determined for each cell. [7] Method according to one of the preceding claims, in which for a plurality of subarrays (20, 22, 24) of the antenna array (18) an at least three-dimensional spectrum with the dimensions distance, relative speed and angle is calculated, the spectral values in the spectra for the individual subarrays are corrected point by point using the model and the corrected spectra are summed. [8] Radar system according to claim 7, in which the calculation of the expected measurement results is carried out for individual grid points in the detection space and interpolation is carried out between the grid points when correcting the spectra. [9] A radar system according to claim 7 or 8, wherein spectra are summed coherently. [10] Radar system according to claim 1 or 2, in which a target is first detected in a spectrum of the received signals and then groups of related targets are identified using the model and the spectral data are fused. [11] Radar system according to one of the preceding claims, comprising a plurality of radar sensors (10, 12, 14) forming a cooperative radar system. [12] Radar system according to claim 11, wherein a high-frequency signal of an oscillator (30) is coherently distributed to a plurality of radar sensors (12, 14).

Images