How to calibrate a radar sensor
The radar sensor calibration method addresses angular inaccuracies by correcting amplitude and phase deviations through statistical evaluation of deviations from control vectors, enhancing accuracy and reliability in multi-target estimation and object formation.
Patent Information
- Application Number
- JP2024535468
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-12-16
- Filing Date
- 2022-11-09
- Publication Date
- 2024-11-28
- Estimated Expiration
- 2042-11-09
AI Technical Summary
Radar sensors in automobile driving support systems suffer from angular errors and decreased correlation values due to aging, temperature effects, and misalignment, leading to inaccuracies in azimuth and elevation measurements, which affect multi-target angle estimation and object formation.
A method for calibrating radar sensors by determining and storing an antenna diagram before use, calculating deviations from control vectors using received signals, and statistically evaluating these deviations to correct amplitude and phase errors, thereby improving angular accuracy and reducing distortive agnosia.
The calibration method enhances the accuracy of radar measurements by correcting angular errors and improving multi-target angle estimation and object formation, even in the presence of distortive agnosia and temperature variations.
Smart Images

Figure 2024544270000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a method for calibrating a radar sensor. [Background technology]
[0002] In automotive driver assistance systems, e.g. automatic distance control systems and collision warning systems, radar sensors are often used to detect the traffic environment. In addition to distance and relative speed, the azimuth angle of the located object is usually important, since for example the azimuth angle allows lane matching when locating the preceding vehicle. The elevation angle of the located object may also be important, since it allows decisions about the relevance of the target, e.g. whether it is surmountable, drivable or an obstacle that may cause a collision.
[0003] The azimuth and elevation angles of a target can be determined from the amplitude and / or phase difference of the transmitting and / or receiving antennas of an antenna array. During angle estimation, the received signal is compared with an antenna diagram according to previously measured angles. If a single target is located, or if several targets that are clearly distinguishable from each other based on their distance and relative speed are located, the estimated angle is taken as the position of the best match (correlation) between the received signal and the antenna diagram. In the general case of multiple target estimation, special estimation algorithms are known that provide estimates of the orientation angles of all the targets involved.
[0004] Conventionally, the antenna diagram of each radar sensor is typically measured in the factory before the radar sensor is put into service. To be able to store and analyze it in the control system, the measurement data is converted into a format that is predefined for this purpose. Normalization takes place here. The antenna diagram can also be defined analytically. In this case, the relative phase is given by 2π sin(d RX,TX / λ), d RX,TXis the distance between the considered transmitter and receiver combinations in the virtual array. Such analytical antenna diagrams are specified purely by calculation.
[0005] Ageing effects, temperature effects and hidden installation of the radar sensor behind a bumper or the vehicle manufacturer's emblem can lead to deviations between the measured antenna diagram and the amplitude and phase differences between the transmitting and / or receiving antennas that actually occur. Such deviations can in principle also be caused by radar sensor misalignment (e.g. elevation misalignment: several targets have elevation angles that deviate significantly from the azimuth calibration) or by incomplete calibration (low number of azimuth and / or elevation calibration measurements). These deviations can lead to angular errors and poor correlation values.
[0006] The correlation value is used, for example, to detect the overlap of several targets in the measurement cell, to activate the multi-target angle estimation algorithm, to detect distortive agnosia, i.e. the disturbance of the angle measurement capability due to coatings on the radar sensor (ice, snow, slush, etc.), to serve as a quality criterion for the reliability of the estimation values and / or to serve as a criterion for object formation (tracking). Thus, a decrease in the correlation value due to the above-mentioned effects increases on the one hand the malfunction of the multi-target angle estimation algorithm (ghost targets with large angle errors of several degrees) and on the other hand the false detection of distortive agnosia. A decrease in the correlation value may also disturb object formation.
[0007] From DE 10 2014208899 A1, a method is known in which correction of amplitude and / or phase differences is performed in a MIMO radar sensor (multiple-input-multiple-output, i.e. multiple transmit and multiple receive antennas) using SIMO angle estimation (single-input-multiple-output, i.e. one transmit antenna and multiple receive antennas) or MISO angle estimation (multiple-input-single-output, i.e. multiple transmit and one receive antenna). Summary of the Invention
[0008] A method for calibrating a radar sensor is proposed, in which an antenna diagram of the radar sensor is determined and stored in a manner known per se before the radar sensor is put into service. The antenna diagram assigns a control vector to each of a number of angles or angle combinations consisting of azimuth and elevation pairs. Only a few coefficients can then be stored, from which the control vector can be reconstructed. It is also possible to store the complete antenna calibration curve, but this is not essential.
[0009] After the radar sensor is put into service, radar measurements are performed on one or more targets. At this time, targets suitable for calibration can be selected. For example, only targets with a signal-to-noise ratio above a threshold can be considered during the measurement. The received signals acquired during the radar measurement are stored in a measurement vector for each target.
[0010] The deviation of the measurement vector from the control vector of the antenna diagram is then calculated for each target. For this purpose, preferably the scalar product
[0011]
number
[0012] , the angle
[0013]
number
[0014] Hermitian conjugate control vector of
[0015]
number
[0016] and the measurement vector
[0017]
number
[0018] Then, the deviation can be calculated according to the following formula 1.
[0019]
number
[0020] Let,be the measurement vector
[0021]
number
[0022] and the calculated scalar product
[0023]
number
[0024] The control vector multiplied by
[0025]
number
[0026] It can be calculated according to Equation 2 as the difference between
[0027]
number
[0028] Scalar product
[0029]
number
[0030] is the pair of azimuth and elevation angles
[0031]
number
[0032] For a combination of angles consisting of azimuth and elevation angles,
[0033]
number
[0034] Hermitian conjugate control vector of
[0035]
number
[0036] and the measurement vector
[0037]
number
[0038] Therefore, it may be calculated according to formula 1*. Then, the deviation
[0039]
number
[0040] Let,be the measurement vector
[0041]
number
[0042] and the calculated scalar product
[0043]
number
[0044] The control vector multiplied by
[0045]
number
[0046] It can be calculated according to Equation 2 as the difference between
[0047]
number
[0048] The calculated deviations are then statistically evaluated for all selected targets, whereby the calculated deviations are averaged or the median of the calculated deviations is calculated. During averaging, the calculated deviations may additionally be weighted by the respective signal-to-noise ratios of the relevant targets. Alternatively, a histogram may be created for the statistical evaluation.
[0049] Finally, the antenna diagram or the radar measurement value is corrected with the statistically evaluated deviation. Here, either the pre-calculated or measured antenna diagram is corrected or the future radar measurement value is directly corrected. In radar sensors that can detect a large number of targets simultaneously, a further processing step may be provided in which a correction of the current radar measurement value is performed in the same cycle in which the statistically evaluated deviation is calculated. This is particularly useful when a coating (ice, snow, slush, etc.) is detected on the radar sensor, but the radar measurement should be performed in the best possible way.
[0050] Radar sensor misalignment means that most objects are not in the calibration plane (e.g. at 0° elevation in the sensor coordinates). This leads to angular errors that cannot be corrected even with perfectly specified misalignment. Angle errors due to antenna diagram distortions are not directly corrected, since it is not possible to distinguish between physical misalignments and antenna diagram distortions. In the described radar sensor calibration, amplitude and / or phase deviations relative to the antenna diagram and thus correlation losses are corrected. This improves the start-up of the multi-target angle estimation algorithm and the detection of distortive blind spots. It also improves object formation that is hindered by correlation losses.
[0051] Only the azimuth angle or, if necessary, the elevation angle can be considered during the correction. Alternatively, the azimuth and elevation angle pair can be considered during the correction. In this case, a two-dimensional correction (2D correction map) that depends on the azimuth and elevation angles is realized.
[0052] The corrections can be applied to all angles, correcting for amplitude and / or phase deviations for the entire range of angles measured by the radar sensor. Such a global correction can be applied to both the antenna diagram and the radar measurement, as described above. With a global correction, the radar measurement of the target is already sufficient to calculate the deviations.
[0053] Alternatively, an angle-dependent correction may be provided, in which the amplitude and / or phase deviations are corrected for a given angle range. This allows different angle ranges with different deviations to be corrected separately. Such a correction may be applied only to the antenna diagram. In angle-dependent correction, radar measurements are made for multiple targets. In particular, one or more targets are measured in each angle range.
[0054] In order to avoid erroneous corrections in the event of a distortive agnosia, deviations can be recorded and statistically evaluated over a significantly longer period than is intended for the detection of a distortive agnosia.
[0055] The calibration may be performed separately for different temperature ranges. As described above, the deviation is calculated separately for each temperature range and the correction is performed. This allows also temperature effects that cause rapid changes to be corrected. The temperature is preferably determined by means of a temperature sensor, which is typically already provided in the radar sensor.
[0056] Exemplary embodiments of the invention are illustrated in the drawings and are explained in more detail in the following description. [Brief description of the drawings]
[0057] [Figure 1] 1 shows a flow chart of a first embodiment of the method according to the invention. [Diagram 2] 4 shows a flow chart of a second embodiment of the method according to the invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0058] 1 and 2 respectively show a flow chart of an embodiment of a method according to the invention for calibrating a radar sensor. In both cases, an antenna diagram of the radar sensor is stored before the radar sensor is put into use. 1 The antenna diagram is stored at multiple angles.
[0059]
number
[0060] Each of the control vectors
[0061]
number
[0062] In this embodiment, the angle
[0063]
number
[0064] is the azimuth angle. In a further embodiment, not shown, the angle may be an elevation angle. Also, in another embodiment, also not shown, multiple angles
[0065]
number
[0066] Instead, multiple combinations of azimuth and elevation angles are used. In the first example of Fig. 1, the radar sensor performs radar measurements 10 on one or more targets, in this example after it has been put into service. Suitable targets are selected for calibration, for example targets whose signal-to-noise ratio exceeds a threshold. The received signals acquired in the radar measurements 10 are stored in a measurement vector for each target,
[0067]
number
[0068] 11. The scalar product is
[0069]
number
[0070] is the Hermitian conjugate control vector of the antenna diagram
[0071]
number
[0072] and the measurement vector
[0073]
number
[0074] Then, the measurement vector
[0075]
number
[0076] and the scalar product
[0077]
number
[0078] The control vector of the antenna diagram multiplied by
[0079]
number
[0080] From this, the difference is formed according to Equation 2,13 and thus the deviation
[0081]
number
[0082] is calculated.
[0083]
number
[0084] Then, the calculated deviation
[0085]
number
[0086] is averaged over all targets,14 and the mean deviation
[0087]
number
[0088] In the averaging, the calculated deviations may be weighted by the respective signal-to-noise ratios of the relevant targets. Alternatively, other types of statistical evaluations may be used, such as, for example, median calculations or histograms.
[0089] In this first embodiment, a global correction is performed in which the amplitude and / or phase deviations are corrected for the entire angular range 15. The antenna diagram is then averaged over the deviations
[0090]
number
[0091] 16 or future radar measurements will be averaged with the deviation
[0092]
number
[0093] 17. Furthermore, if the radar sensor detects multiple targets simultaneously, the current radar measurement is corrected with the same period 18. The second embodiment of Fig. 2 differs from the first embodiment in that different angular ranges are examined separately. After putting the radar sensor into use, in this embodiment radar measurements 20 are performed on a number of targets in different angular ranges. The received signals are then converted into measurement vectors 20 for each target.
[0094]
number
[0095] 21, deviation is saved
[0096]
number
[0097] is calculated for each target according to Equation 1 and Equation 2 above,22,23 and the deviation
[0098]
number
[0099] is the averaged deviation
[0100]
number
[0101] , or is otherwise statistically evaluated as described above. See the description of the first embodiment. In this second embodiment, an angle-dependent correction is performed in which the amplitude and / or phase deviations are corrected only for a given angular range 25. The antenna diagram is then plotted against the averaged deviations.
[0102]
number
[0103] This is corrected to 26. In both embodiments, steps 10 through 18 or steps 20 through 26 can be repeated for different temperature ranges.
Claims
1. 1. A method for calibrating a radar sensor, comprising: before starting to use said radar sensor, a control vector [0010] (1) storing an antenna diagram enabling the assignment of performing (10; 20) radar measurements on one or more targets; The received signal for each target is converted into a measurement vector for said target. [0025] and storing the data (11;21) in the above steps, - the control vector for each target [0030] The measurement vector from [0045] Deviation of [0050] (12, 13; 22, 23) a step of calculating - the calculated deviation for all targets [006] Statistically evaluating (14;24) - said statistically evaluated deviations [0070] and correcting (15, 16, 17, 18; 25, 26) said antenna diagram or said radar measurements with A method comprising:
2. The measurement vector [0080] The control vector [0090] When calculating the deviation from (12, 13; 22, 23), the Hermitian conjugate control vector [0010] and the measurement vector ##EQU00011## The scalar product with ##EQU00012## is calculated (12; 22), and the deviation ##EQU00013## is the measurement vector ##EQU14## and the calculated scalar product ##EQU00015## 2. The method according to claim 1, characterized in that the control vector is calculated as the difference between the control vector multiplied by (13; 23).
3. A correction (25) according to the angle of the antenna diagram is performed based on the statistically evaluated deviation. ##EQU00016## 3. The method according to claim 1 or 2, characterized in that it is carried out at
4. 4. A method according to claim 1, characterized in that the calibration is performed for a combination of angles consisting of azimuth and elevation angles.
Citation Information
Patent Citations
Method and device for determining the installation position of a vehicle sensor
DE102019127576A1
How to calibrate mimo radar sensors for automobiles
JP2017513027A
Calibration device of laser device, calibration system, calibration method, and lader device
JP2021139852A
A method for calibrating MIMO radar sensors for automobiles
JP2021529309A
Device and method for calibrating a multiple-input multiple output radar sensor
WO2020148179A1
