Device, sensor system and method for determining a radar target list

Abstract

The present invention relates to a device for determining a radar target list, the device having: an input interface for receiving pre-processed sensor data of a radar sensor, the sensor data having information about the power detected in a high point in the distance dimension and / or the speed dimension and the power detected in a predefined neighborhood of the high point in the distance dimension and / or the speed dimension; an analysis unit for determining adjacent high points having overlapping neighborhoods based on the pre-processed sensor data; an adaptation unit for adapting the neighborhoods of the adjacent high points; and an evaluation unit for determining a radar target list having information about targets in the field of view of the radar sensor based on the high points and their neighborhoods. The present invention also relates to a method for determining a radar target list and a sensor system for detecting targets in the surroundings of a vehicle.

Description

Field of the Invention

[0001] The present invention relates to a device for determining a radar target list. The present invention also relates to a sensor system for detecting targets in the surroundings of a vehicle, and to a method for determining a radar target list. Background Art

[0002] Today's vehicles (cars, transport vehicles, trucks, motorcycles, etc.) have multiple sensors that provide information to the driver and semi-automatically or fully automatically control various functions of the vehicle. The surroundings of the vehicle and other traffic participants are recorded by means of sensors. Based on the recorded data, a model of the vehicle surroundings can be generated and a response can be made to changes in this vehicle surroundings.

[0003] Here, an important sensor principle is radar technology. Radar sensors currently commonly used in the vehicle field operate as multi-pulse radar sensors (also known as chirp sequence radar sensors), where multiple frequency-modulated pulses are emitted at short intervals. Radar sensors typically include multiple transmitting elements and receiving elements (antenna arrays), and these transmitting elements and receiving elements form virtual channels (Rx / Tx antenna pairs) of the radar sensor. In each receiving channel, downmixing to baseband, filtering, and subsequent digitization of the obtained baseband signal are performed. By preprocessing the baseband signal of each receiving channel, targets (i.e., objects in the field of view of the radar sensor) can be detected and located.

[0004] Here, the preprocessing can in particular include: performing a Fourier transform or optimal filtering (so-called fast-time processing) on the baseband data of individual radar pulses, performing a further Fourier transform on multiple pulses (so-called slow-time processing), beamforming, signal power detection (e.g., based on the constant-force-alarm-rate method), and determining high points. Then, based on these preprocessed sensor data, a radar target list (also known as a point cloud) can be generated. This radar target list includes the following target parameters: distance, radial velocity or Doppler velocity, and (if available) azimuth and elevation angle, and this radar target list forms the basis for recognizing the surroundings.

[0005] There are various methods for determining high-resolution angles based on the phase differences of signals received on the elements of an antenna array. Examples are MUSIC, Root-MUSIC or ESPRIT. The maximum likelihood method is usually complex for calculations in the local domain, yet efficient calculations can be performed in the frequency domain. In this regard, the use of algorithms for high-resolution angle estimation in a 77 GHz radar sensor is described in "High-Resolution Angle Estimation for an Automotive FMCW Radar Sensor" by Schoor et al. A flexible method for performing computationally efficient high-resolution frequency estimation, which is based on decoupled frequency estimation in the Fourier domain, is disclosed in "Advances in Automotive Radar: A framework on computationally efficient high-resolution frequency estimation" by Engels et al. Here, preprocessing can be applied to the range dimension, the velocity dimension or the angle dimension.

[0006] Such methods typically require multiple snapshots of the target to determine the signal covariance matrix. However, this is usually not reliably achievable due to the sometimes rapid relative motion between the sensor and the target, as it is not possible to associate individual targets over multiple measurement periods during signal processing and the angles measured over multiple periods can also vary greatly. Another solution involves observing cells (neighborhoods or peak regions) adjacent to high points in the range-velocity domain. It is also possible to observe the vicinity of the target in the range-velocity-angle domain to estimate high-resolution frequencies.

[0007] The disadvantage of additionally observing the neighborhood when evaluating individual high points is that targets adjacent in the range-velocity domain usually influence each other. This can lead to inaccuracies or false detections. The solution using dual-target modeling results in a significantly increased computational complexity and can also cause target duplication. Summary of the Invention

[0008] Starting from this, the object of the present invention is to provide an efficiently computable and reliable method for generating a radar target list. It should enable a high angular resolution to be achieved in the angle dimension. Duplication of targets or other false detections and the resulting inaccuracies should be avoided.

[0009] To achieve this object, in a first aspect, the invention relates to a device for determining a radar target list, the device having:

[0010] an input interface for receiving preprocessed sensor data of a radar sensor, the sensor data having information about the power detected in a high point in the distance dimension and / or the speed dimension and about the power detected in a predefined neighborhood of the high point in the distance dimension and / or the speed dimension;

[0011] an analysis unit for determining mutually adjacent high points having overlapping neighborhoods based on the preprocessed sensor data;

[0012] an adaptation unit, the adaptation unit being used to adapt neighborhoods of the high points adjacent to each other; and

[0013] An evaluation unit is used to determine a radar target list with information about targets in a field of view of the radar sensor based on the high point and its neighborhood.

[0014] In another aspect, the invention relates to a sensor system for detecting objects in the surroundings of a vehicle, the sensor system having:

[0015] A radar sensor for generating and preprocessing sensor data, wherein the radar sensor is preferably designed as a frequency-modulated multi-pulse radar sensor; and a device as described above.

[0016] Other aspects of the present invention relate to: a method formed corresponding to the aforementioned device; and a computer program product having a program code, which, when implemented on a computer, is used to perform the steps of the method; and a storage medium, on which a computer program is stored, which, when implemented on a computer, causes the method described herein to be implemented.

[0017] The invention also provides preferred embodiments. It goes without saying that the features mentioned in the embodiments and the features still to be described below can be used not only in the respectively given combination, but also in other combinations or alone without departing from the scope of the invention. In particular, the sensor system, the method and the computer program product are implemented in a manner corresponding to the embodiments described herein for the device.

[0018] According to the present invention, pre - processed sensor data of a radar sensor is received. Here, for a receiving channel, power detected in the distance dimension and / or speed dimension can be received. It is also possible to receive only the power detected at the high points and in the surrounding environment (neighborhood) of the high points. First, it is analyzed whether it is possible to determine that two high points have overlapping neighborhoods based on the pre - processed sensor data. For this purpose, in particular, the positions of the high points and the size of the pre - defined neighborhood can be analyzed. According to the present invention, the neighborhoods of adjacent high points with overlapping neighborhoods are adapted. A radar target list is determined based on the high points and their adapted neighborhoods.

[0019] Therefore, compared with the methods to date, according to the present invention, an additional step of adapting the neighborhoods of adjacent high points is provided before establishing the target list. For each high point, the same neighborhood is not observed to determine whether the high point is a single target or multiple targets. Instead, it is first checked whether the neighborhood of the high point may contain information that cannot be traced back to this adjacent high point due to the influence of the adjacent high point. In order not to consider such inapplicable information, the neighborhood is adapted. By adapting the neighborhoods of adjacent high points, mutual influences that may worsen the result are avoided. Reliable identification of targets and high detection accuracy are achieved. The determined radar target list can have higher precision (e.g., in the angular dimension) through the previously implemented adaptation of the neighborhood.

[0020] Here, in a preferred design, the adaptation unit is designed to shorten the neighborhoods of the adjacent high points in the direction towards the respective other high point. The shortening is preferably limited by a pre - defined minimum length. Here, shortening should be understood as ignoring multiple parts of the neighborhood. In further processing, the part of the neighborhood of the high point that faces the respective other high point is not involved. Thus, the power in this part of the neighborhood is ignored when determining whether it is a single target or multiple targets. The information in this part of the neighborhood has a high error tendency. Therefore, by shortening the neighborhood, errors that may occur due to the influence of this part of the neighborhood during target identification are avoided. The reliability when determining the radar target list is improved.

[0021] Here, in a preferred design, the adaptation unit is designed to adapt the neighborhoods of the adjacent high points based on the power detected in the respective other high point and / or its neighborhood. Preferably, higher power detected in the respective other high point causes greater shortening of the neighborhood. The power detected in the adjacent high point can be considered during adaptation. If the adjacent high point is a very prominent high point, stronger adaptation can be performed, for example, through greater (wider) shortening. Thereby, the reliability during target identification is further improved. A particularly prominent high point may cause a greater influence and thus greater errors.

[0022] In a preferred design, the adaptation unit is designed to adapt the neighborhoods of adjacent high points based on the distance between adjacent high points. Preferably, a larger distance causes a greater shortening of the neighborhood. High points with very small distances between each other result in strong mutual influences and may cause errors when determining the radar target list. Therefore, adapting the neighborhoods based on the distance can improve the reliability when determining the radar target list. Errors and false associations are avoided.

[0023] In a preferred design, the adaptation unit is designed to expand the neighborhoods of adjacent high points in a direction away from the respective other high point. Preferably, the size of the neighborhood remains constant. In other words, adapting the neighborhood can include shifting away from the respective other high point. This shifting allows the size of the neighborhood to remain constant in order to enable efficient further calculations. Computability can be improved.

[0024] In a preferred design, the adaptation unit is designed to determine a weight coefficient for the neighborhoods of adjacent high points. The evaluation unit is designed to determine the radar target list based on the determined weight coefficient. The adaptation can also include determining the weight coefficient. In particular, those parts of the neighborhood of a high point facing the adjacent high point can be incorporated with only a small weight during further processing in order to compensate for possible errors in this range. Therefore, the individual cells are considered with a small weight. A higher reliability is achieved when determining whether a high point is one target or multiple targets.

[0025] In a preferred design, the input interface is designed to receive preprocessed sensor data with a list of high points. The list of high points includes, for each high point, the high point position, the high point power, and a plurality of ambient powers in a predefined neighborhood of the high point. By only receiving the list of high points, the amount of information to be transmitted is minimized. Efficient computability is achieved.

[0026] In a preferred design, the preprocessed sensor data includes the power detected in the distance dimension and the speed dimension. It is particularly advantageous that a two-dimensional evaluation or determination of the neighborhood can be achieved in the distance dimension and the speed dimension. Then, for example, the neighborhood can include a side length of 3, 5, 7, or 9 cells in the speed dimension direction and in the distance dimension direction, where the high point corresponds to the middle cell accordingly. By reliably identifying the target based on the list of high points, an improved resolution is achieved when determining the angle subsequently.

[0027] In a preferred design, the preprocessed sensor data additionally includes the power detected in the azimuth dimension and / or the elevation dimension. The method according to the invention can also be extended to the azimuth dimension and / or the elevation dimension (angle dimension). Thereby, a further improved and reliable detection of radar targets within the visual range of the sensor is achieved. The angular resolution can be increased.

[0028] In a preferred design, the evaluation unit is designed to determine the radar target list based on a comparison of the high points and their neighborhoods with a predefined model. In particular, for each high point, a comparison with the model can be made to determine whether the high point is one target or multiple targets. By using the comparison method, efficient computability and reliable target detection are achieved.

[0029] In a preferred design, the evaluation unit is designed to determine the radar target list based on the MUSIC method, the Root-MUSIC method, the ESPRIT method, or the maximum likelihood method. These methods achieve an accurate determination of the targets.

[0030] The radar sensor emits radar signals and receives the reflections of the radar signals on objects (also referred to as targets) within the field of view of the radar sensor. The object can be, for example, another vehicle, but can also be another traffic participant (pedestrian, cyclist, etc.) or a stationary object (tree, house, traffic sign, etc.). The field of view represents the area within which objects can be recorded. The radar sensor can include multiple individual sensors, which, for example, enable a 360° omnidirectional view and can thus present a comprehensive image of the vehicle's surroundings. The preprocessed sensor data represents the scanned and preprocessed baseband signal. The preprocessing steps can particularly include optimal filtering or Fourier transform of the scanned values for each pulse and Fourier transform over multiple pulses. The preprocessed sensor data particularly can include a two-dimensional index of the power of each cell in the direction of the distance dimension and the speed dimension. In this regard, the sensor data can exist in the form of a matrix and provide an image of the surroundings. It is also possible to receive only the high points and their neighborhoods in the form of a list of high points. Thereby, the amount of data to be transmitted can be minimized. The predefined neighborhood should particularly be understood as an index of the number of cells next to or around the cell corresponding to the high point. For example, the predefined neighborhood in the distance dimension and the speed dimension can include two cells on each side of the high point. An overlapping neighborhood is a neighborhood that partially includes the same cells. A shortened neighborhood should particularly be understood as deleting a part of the neighborhood or weighting a part of the neighborhood with a zero weight. Description of the Drawings

[0031] Next, the present invention will be described and explained in detail with reference to several selected embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0032] Figure 1 A schematic diagram of a sensor system according to the present invention in a vehicle is shown;

[0033] Figure 2 An illustration of a device according to the present invention is shown;

[0034] Figure 3 A schematic diagram of adapting to the neighborhood of high points adjacent to each other according to the present invention is shown; and

[0035] Figure 4 A schematic diagram of a method for determining a radar target list according to the present invention is shown. Detailed Description of the Invention

[0036] In Figure 1 a sensor system 10 for detecting targets in the surroundings 12 of a vehicle 14 according to the present invention is schematically shown. In the example shown, the sensor system 10 is an automotive radar system integrated into the vehicle 14. Objects 16 in the surroundings 12 of the vehicle can be detected as individual targets 17. It is also possible to implement such that individual objects 16 form multiple targets 17 in the radar target list.

[0037] The sensor system 10 includes a radar sensor 18 and a device 20. The radar sensor 18 is preferably designed as a frequency-modulated multi-pulse radar sensor (FMCW radar sensor) and generates sensor data reflecting the surroundings 12 of the vehicle 14. A radar target list is determined in the device 20, and this radar target list can be used to generate an image of the surroundings in order to, for example, enable autonomous operation of the vehicle 14 or support the driver.

[0038] In Figure 2 a device 20 according to the present invention is schematically shown. The device 20 can in particular be integrated into the radar sensor. It is also possible to implement the device as an additional module for the radar sensor or the vehicle controller. In addition, the device 20 can be implemented as software executed by a processor of the radar sensor or by a processor of the vehicle controller.

[0039] The device 20 includes an input interface 22, an analysis unit 24, an adaptation unit 26, and an evaluation unit 28. Here, the individual units and interfaces can in particular be implemented individually or in combination or completely or partially in software and / or hardware.

[0040] The input interface 22 is connected to the radar sensor and receives the data of the radar sensor after preprocessing the raw data in the multi-pulse radar sensor. In particular, the received preprocessed sensor data includes information about the power detected in the peaks in the range dimension and / or velocity dimension and the power detected in the predefined neighborhood of the peaks in the range dimension and / or velocity dimension. On the one hand, a list of peaks can be received directly. On the other hand, a complete range and velocity matrix can also be received, which not only reflects the peaks and their neighborhoods, but also all other cells in the range dimension and velocity dimension. In particular, the sensor data here includes power information for each discrete cell in the range dimension (e.g., a cell size of 1 meter) and in the velocity dimension (which can also be called the Doppler dimension, e.g., a cell size of 0.5 meters per second).

[0041] In the analysis unit 24, it is determined that adjacent peaks have overlapping neighborhoods. For this purpose, the positions of the peaks in the range dimension and velocity dimension are observed, and it is determined whether the neighborhoods of two peaks partially include the same cells. Here, the analysis unit can in particular be designed for geometric evaluation.

[0042] If it has been previously determined that adjacent peaks have overlapping neighborhoods, then in the adaptation unit 26, the neighborhoods of these adjacent peaks are adapted. In particular, at least one of the neighborhoods can be shortened. Here, shortening should be understood as not observing or ignoring the power in at least one cell of the neighborhood cells. Additionally or alternatively, the weight of the cells in the overlapping range can also be reduced.

[0043] Optionally, the adaptation unit 26 can here be designed to perform the adaptation based on the power detected in the respective other peak. Here, a higher power detected in the respective other peak can cause a greater shortening. The distance between the two adjacent peaks can also be considered when adapting the neighborhood. In particular, a smaller distance can cause a greater shortening of the neighborhood.

[0044] Subsequently, in the evaluation unit 28, a radar target list with information about the targets in the field of view of the radar sensor is determined based on the peaks and the adapted neighborhoods. For this purpose, the evaluation unit 28 can for example be designed to use the MUSIC method, the Root-MUSIC method, the ESPRIT method or the maximum likelihood method.

[0045] In Figure 3The distance and speed ranges are schematically shown herein. Here, the distance r between the sensor and the target is plotted on the x-axis, where one cell can correspond to, for example, one meter. The relative angular velocity v (Doppler) of the target is plotted on the y-axis, where one cell can correspond to a speed of, for example, 0.5 m / s. In the example shown, the distance and speed ranges have two high points H1, H2, which correspond to cells with a detected (compared to their surroundings) relatively high power. The neighborhood accordingly includes a length of five cells in both the distance dimension and the speed dimension. Here, the neighborhood of the high point H1 is represented by a vertical line. The neighborhood of the high point H2 is represented by a horizontal line. There are cells that belong to both neighborhoods and are represented by the vertical and horizontal lines between the high points H1 and H2.

[0046] In the methods heretofore used to improve the angular resolution in automotive radar applications, the neighborhoods of the high points in the distance and / or speed ranges are observed and evaluated. In particular, modeling is used to determine whether a high point is a single target or multiple targets are presented as a single high point.

[0047] Here, in particular in the case of partial overlap of the neighborhoods, errors or inaccuracies may occur. If, for example, there are two targets with a distance of three cells in the speed dimension in the case of the same speed and similar radar cross-sections (with azimuth angles from +5° and -5°), inaccuracies may occur when determining the radar target list. First, the high points in the distance and speed ranges should be correctly detected. However, when determining the high-resolution angle based on the covariance matrix based on the vicinity of a length of five cells, two targets may be detected respectively because the neighborhoods overlap by two cells each. That is, four targets may be generated from the two actually existing targets in various combinations of distance and azimuth angle. If additionally it is assumed that the azimuth angles are also the same, then two high points may further be detected in the distance, speed, and azimuth angle ranges. Then it may be necessary to perform a relatively complex test for the double-target hypothesis because there is also another target in the spectrum for each of the two high points.

[0048] According to the present invention, it is proposed that when detecting a high point, a vicinity observation is performed, and an adaptive high-point neighborhood is determined based thereon. For this purpose, after detecting a high point, the overlap of the neighborhood of each high point with the neighborhoods of other high points is checked. If there is an overlap with other neighborhoods, the corresponding dimension is adaptively adjusted. Here, the check is performed in terms of its dimension. Here, the adjustment may particularly include shortening the neighborhood in the direction towards the adjacent target. Here, shortening should be understood as truncating the neighborhood so that the neighborhoods of adjacent high points are no longer shortened or are only shortened within one's own neighborhood. Thus, when further evaluating the currently observed high point, the information of the adjacent target that has been resolved when detecting the high point is no longer considered or is only considered less.

[0049] InFigure 3 Corresponding examples are shown. If it is determined that the neighborhoods of adjacent high points overlap, then the neighborhoods are adapted. In the example shown, when further processing and determining the neighborhood, for example, for the high point H1, ten overlapping cells may be shortened. That is, when further processing, these ten cells represented by horizontal and vertical lines may be ignored or considered with reduced weights. Similarly, only five cells facing another high point H2 may be ignored.

[0050] In Figure 4 A method for determining a radar target list according to the present invention is schematically shown. The method includes the following steps: receiving S10 preprocessed sensor data, determining S12 adjacent high points, adapting S14 neighborhoods, and determining S16 a radar target list. The method can be implemented, for example, as software implemented on a vehicle controller. Preferably, the method is applied based on its current output during the normal operation of the radar sensor.

[0051] The present invention is generally described and illustrated with reference to the accompanying drawings and the description. The description and illustration should be understood as examples and not restrictive. The present invention is not limited to the disclosed embodiments. Those skilled in the art will arrive at other embodiments or variations when using the present invention and when accurately analyzing the scope of the present invention disclosure.

[0052] In the present invention, the words "comprising" and "having" do not exclude the existence of other elements or steps. The indefinite article "a" or "an" does not exclude the existence of a plurality. A single element or a single unit can perform the functions of multiple units mentioned in the present invention. Elements, units, interfaces, devices, and systems can be implemented partially or completely in hardware and / or software. Several measures separately mentioned in multiple different scenarios should not be understood as not being able to use the combination of these measures in an equally advantageous manner. A computer program can be stored / run on a non-volatile data carrier, for example, stored / run on an optical memory or a solid-state drive (SSD). The computer program can run together with hardware and / or as part of software, for example, run via the Internet or via a wired or wireless communication system. The reference signs in the present invention should not be understood as limiting.

[0053] List of reference numerals

[0054] 10 Sensor system

[0055] 12 Surroundings

[0056] 14 Vehicle

[0057] 16 Object

[0058] 17 Target

[0059] 18 Radar sensor

[0060] 20 Device

[0061] 22 Input interface

[0062] 24 Analysis unit

[0063] 26 Adaptation unit

[0064] 28 Evaluation unit

Claims

1. An apparatus (20) for determining a radar target list, the apparatus having: an input interface (22) for receiving pre - processed sensor data from a radar sensor (18), the sensor data having information about the power detected at high points (H1, H2) in the distance dimension and / or the velocity dimension and the power detected in a pre - defined neighborhood of the high points in the distance dimension and / or the velocity dimension; an analysis unit (24) for determining adjacent high points with overlapping neighborhoods based on the pre - processed sensor data; an adaptation unit (26) for adapting the neighborhoods of the adjacent high points, wherein the adaptation unit (26) is designed to shorten the neighborhoods of the adjacent high points (H1, H2) in the direction towards the respective other high point; and an evaluation unit (28) for determining a radar target list having information about targets (17) in the field of view of the radar sensor based on the high points and their neighborhoods.

2. The apparatus (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the power detected in the respective other high point and / or its neighborhood.

3. The apparatus (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the distance between the adjacent high points.

4. The apparatus (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the distance between the adjacent high points.

5. The apparatus (20) according to claim 1, wherein the adaptation unit (26) is designed to expand the neighborhoods of the adjacent high points (H1, H2) in the direction away from the respective other high point.

6. The apparatus (20) according to claim 1, wherein the adaptation unit (26) is designed to determine a weight coefficient for the neighborhoods of the adjacent high points (H1, H2); and the evaluation unit (28) is designed to determine the radar target list based on the determined weight coefficient.

7. The apparatus (20) according to claim 1, wherein the input interface (22) is designed to receive pre - processed sensor data having a list of high points; and the list of high points includes, for each high point (H1, H2), the high point position, the high point power, and a plurality of ambient powers in the pre - defined neighborhood of the high point.

8. The apparatus (20) according to claim 1, wherein the pre - processed sensor data includes the power detected in the distance dimension and the velocity dimension.

9. The apparatus (20) according to claim 8, wherein the pre - processed sensor data additionally includes the power detected in the azimuth dimension and / or the elevation dimension.

10. The device (20) according to claim 1, wherein the evaluation unit (28) is designed to determine the radar target list based on a comparison of the high points (H1, H2) and their neighborhoods with a predefined model.

11. The device (20) according to claim 1, wherein the evaluation unit (28) is designed to determine the radar target list based on the MUSIC method, the Root-MUSIC method, the ESPRIT method, or the maximum likelihood method.

12. The device (20) according to claim 1, wherein the adaptation unit (26) is designed to shorten the neighborhoods of the adjacent high points (H1, H2) in the direction towards the respective other high point; the shortening is limited by a predefined minimum length.

13. The device (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the power detected in the respective other high point and / or its neighborhood; and higher power detected in the respective other high point causes a greater shortening of the neighborhood.

14. The device (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the distance between the adjacent high points; and a smaller distance causes a greater shortening of the neighborhood.

15. The device (20) according to claim 1, wherein the adaptation unit (26) is designed to adapt the neighborhoods of the adjacent high points (H1, H2) based on the distance between the adjacent high points; and a smaller distance causes a greater shortening of the neighborhood.

16. The device (20) according to claim 1, wherein the adaptation unit (26) is designed to expand the neighborhoods of the adjacent high points (H1, H2) in the direction away from the respective other high point; and the size of the neighborhood remains constant.

17. A sensor system (10) for detecting a target (17) in the surroundings (12) of a vehicle (14), the sensor system having: a radar sensor (18) for generating and preprocessing sensor data; and a device (20) according to one of the above claims.

18. The sensor system according to claim 17, wherein The radar sensor is designed as a frequency-modulated multi-pulse radar sensor.

19. A method for determining a radar target list, the method having the following steps: receiving (S10) preprocessed sensor data of a radar sensor (18), the sensor data having information about the power detected in high points (H1, H2) in the distance dimension and / or the speed dimension and the power detected in a predefined neighborhood of the high points in the distance dimension and / or the speed dimension; determining (S12) adjacent high points with overlapping neighborhoods based on the preprocessed sensor data; Adapt (S14) the neighborhoods of said high points adjacent to each other, wherein the neighborhoods of said adjacent high points (H1, H2) are shortened in the direction towards the respective other high point; And Determine (S16) a radar target list having information about a target (17) in the field of view of the radar sensor based on said high points and their neighborhoods.

20. A computer program product having program code which, when implemented on a computer, is for performing the steps of the method according to claim 19.

Images