WO2025229016A1

WO2025229016A1 - Method and radar system for radar-based localizing of an object by a radar system, method and electronic vehicle guidance system for guiding a vehicle at least partially automatically, data processing system, as well as computer program product

Info

Publication number: WO2025229016A1
Application number: PCT/EP2025/061755
Authority: WO
Inventors: Sandeep Kumar; Niko Moritz Scholz; Maximilian Poepperl; Artem LUKIN; Alan Jenkins
Original assignee: Valeo Schalter und Sensoren GmbH
Current assignee: Valeo Schalter und Sensoren GmbH
Priority date: 2024-04-30
Filing date: 2025-04-29
Publication date: 2025-11-06
Anticipated expiration: 2026-10-30
Also published as: DE102024112118A1

Abstract

The invention relates to a method for radar-based localizing of an object by a radar system (2), a method for guiding a vehicle (1) at least partially automatically, a data processing system (3), a radar system for localizing an object, an electronic vehicle guidance system for guiding a vehicle at least partially automatically, as well as a computer program product. The method for localizing comprises the following steps: - receiving first radar data (7) of a first antenna array of a first radar sensor (4) of the radar system (2); - receiving second radar data (8) of a second antenna array of a second radar sensor (5) of the radar system (2) that is different from the first antenna array; and - determining an azimuthal angle and/or an elevation angle and/or an angle of arrival for the object depending on input data (6), which comprises the first radar data (7) and the second radar data (8), by a trained learning algorithm (9).

Description

Method and radar system for radar-based localizing of an object by a radar system, method and electronic vehicle guidance system for guiding a vehicle at least partially automatically, data processing system, as well as computer program product

Aspects of the invention relate to a method for radar-based localizing of an object by a radar system, a method for guiding a vehicle at least partially automatically, a data processing system, a radar system for localizing an object, an electronical vehicle guidance system for guiding a vehicle at least partially automatically, as well as a computer program product.

Radar sensor systems are employed in many different applications for vehicles, in particular also for functions at raised speeds. Typical functions comprise adaptive cruise control, ACC. cross traffic alert, CTA, or dead angle assistants.

The requirements on the performance of environmental sensor systems, such as radar sensor systems, continually increase. In order to be able to meet the high requirements in particular for applications requiring long ranges, such as for instance applications at high vehicle speeds, it is in particular desirable to achieve a high angle resolution in order to be able to distinguish objects localized comparatively close to one another, for example on adjacent lanes, from each other. Further, a high accuracy in localizing objects, in particular with regard to their angle position, is desirable.

From the US 2022/0283265 A1 a radar system is known which includes a plurality of substantially identical transceiver sets, each of which form substantially identical, overlapping virtual antenna arrays.

Fuchs, Jonas & Gardill, Markus & Lubke, Maximilian & Dubey, Anand & Lurz, Fabian. (2022). A Machine Learning Perspective on Automotive Radar Direction of Arrival Estimation. IEEE Access. PP. 1 -1. 10.1109/ACCESS.2022.3141587 provides an overview of the current progress and work in the field of the estimation of the arrival direction based on Deep Learning in connection with the motor vehicle radar.

It is an objective of the invention to increase an angle resolution and/or an angle accuracy of the localizing during radar-based localizing of an object in an environment of a vehicle. This objective is solved by the subject matter of the independent claim. Advantageous further developments and preferred embodiments are subject matter of the dependent claims.

The invention is based on the idea to use data of at least two radar sensors for determining an angle of arrival by a machine learning algorithm as input data.

According to an aspect of the invention a method for radar-based localizing of an object, in particular in an environment of a vehicle, is indicated. The vehicle is in particular a vehicle, for example a car, a passenger car, a truck, or another utility vehicle. The method comprises the following steps:

- In particular, receiving first radar data of a first antenna array of a first radar sensor of the radar system. In particular, the first radar data is provided to an evaluation unit of the radar system, in particular received by the evaluation unit. The first radar data may in particular also be referred to as first measuring data.

- In particular, receiving second radar data of a second antenna array of a second radar sensor of the radar system that is different from the first antenna array. In particular, the second radar data is provided to the evaluation unit, in particular received by the evaluation unit. The second radar data may in particular also be referred to as second measuring data.

- In particular, determining an azimuthal angle and/or an elevation angle and/or an angle of arrival for the object depending on input data, which comprises the first radar data and the second radar data, by a trained learning algorithm.

For example for localizing the object an azimuthal angle of the object and/or an elevation angle of the object and/or an angle of arrival for the object is determined depending on the input data.

The method may for example be configured to be computer-implemented, that is purely computer-implemented. Unless stated otherwise, all steps of the computer-implemented method may be performed by a data processing system, which comprises at least one data processing device, in particular of a data processing system of the vehicle. In particular, the data processing device may also be referred to as evaluation unit. In particular, the at least one data processing device is configured or adapted for executing the steps of the computer-implemented method. For this purpose, the at least one data processing device may for example store a computer program, which comprises instructions, which, when executed by the at least one data processing device, cause the at least one data processing device to perform the computer-implemented method. The terms “data processing system” and “at least one data processing device” may be used interchangeably.

All data processing devices of the at least one data processing device may be part of the vehicle. It is, however, also possible that all data processing devices of the at least one data processing device are part of an external computing system outside the vehicle, for example a backend server or a cloud computer system. It is also possible that the at least one data processing device comprises both at least one vehicle data processing device of the vehicle as well as at least one external data processing device of the external computing system. The at least one vehicle data processing device may for example be comprised by one or several control units, ECUs, and/or by one or several zone control units, ZCUs, and/or one or several domain control units, DCUs, of the vehicle and/or by the radar system. In particular the evaluation unit of the radar system is part of the data processing system, for example one of the at least one data processing device.

In case that the at least one data processing device comprises two or more data processing devices, certain steps, which are performed by the at least one data processing device, may for example be understood such that different data processing devices perform various steps or different parts of a step. In particular, it is not required that each data processing device performs the steps completely. In other words, the performance of the steps may be distributed over the two or more data processing devices.

From each embodiment of the computer-implemented method a corresponding embodiment of a method for localizing an object, which is not purely computer- implemented, derives by including corresponding steps for generating the first radar data, for example by the radar system, in particular by the first radar sensor, and/or the second radar data, for example by the radar system, in particular by the second radar sensor.

For example the first antenna array comprises several first antenna elements and the second antenna array comprises several second antenna elements. In particular, the first antenna elements and the second antenna elements are spaced apart from each other, in particular they do not overlap. For example, a distance between the first antenna array and the second antenna array amounts to 15 centimeters and 2 meters. The antenna elements may in particular also be referred to as elements. In particular, the first radar sensor and the second radar sensor are arranged spaced apart from each other and are not overlapping. If applicable, the fields of vision of the first and the second radar sensor may overlap. Preferably, the radar sensors are multiple- input-multiple-output radar sensors, MIMO radar sensors. In particular, the radar sensors are monostatic MIMO radar sensors, in particular no bistatic MIMO radar sensors. For example, a behavior of the radar sensors may be described by virtual antenna arrays. For example, the first radar sensor comprises n transmitters and m receivers. For example, the first virtual antenna array comprises n x m virtual antenna elements. In particular, this also applies in analogy to the second virtual antenna array. A virtual antenna element in particular derives from a combination of a transmitter and a receiver. In particular the radar data are receiving data of the, in particular virtual, antenna elements. Preferably, also the virtual first antenna elements and the virtual second antenna elements, in particular their virtual positions relative to one another, are spaced apart from each other and in particular do not overlap. Virtual positions of the virtual antenna elements may for example be calculated by a convolution operation between the discrete, in particular real, elements, of which for example the array consists. A property of this convolution is that a single virtual array may be calculated from many possible, in particular real, antenna arrays that are not degenerate. A property of this convolution is that a single virtual array may be calculated from many possible, non-degenerate real antenna arrays.

A virtual element is in particular a receiving channel that is a virtual receiving element. A physical receiving element may for example receive data from several transmitter elements. These are for example treated as if they were separate virtual elements.

The first radar sensor and the second radar sensor each are in particular primary radar systems, for example frequency-modulated continuous wave radars, also referred to as FMCW radars. The first radar sensor and the second radar sensor may for example be FMCW-MIMO radars (MIMO: multiple input multiple output), in particular 4D-FMCW- MIMO radars.

Localizing the object comprises determining the angle of arrival, AoA, also referred to as direction of arrival, of the object. The angle of arrival may, in particular, be taken as combination of the elevation angle and the azimuthal angle. Consequently, in such embodiments the azimuthal angle and the elevation angle of the object with regard to the vehicle or with regard to the radar sensors may be determined with increased accuracy or resolution, respectively. Radar data may, in particular also in the case of different kinds of radar sensors, be of different kind or be differently generated. The radar data may for example be raw radar data or correspond to the temporal course of a respective receiver signal from one or several receiving antennae of the radar sensor or to certain key figures of such signals, such as pulse amplitudes or pulse widths or pulse transit times, or digital data scanned by an analog-to-digital converter, ADC converter. The radar data may for example also be further preprocessed data, wherein the preprocessing may comprise for example a Fourier transformation, in particular of the digital data sampled by the ADC converter.

For example in the case of FMCW radars the radar data may for example correspond to for example to the Fourier-transformed receiving signals sampled by the ADC converter. In this case at times it is also referred to radar frequency data.

The trained learning algorithm may, in particular, also be referred to as trained machine learning model, MLM. In particular, it is a deep learning algorithm.

A trained MLM may be taken as algorithm, in particular computer-implemented algorithm, which can imitate concrete functions or in a wider sense functions rendered possible by the capacity of the human mind. A trained MLM may also for example be referred to as “trained function”.

When training an MLM generally parameters of the MLM are adapted or updated. The training may in general be effected in supervised, partially supervised, or unsupervised manner. The training may also comprise reinforcement learning or representation learning and/or further known training methods. In particular, the parameters of the MLM may iteratively be adapted over several training steps. In particular, for training a predefined loss function may be minimized. For adapting the parameters, in case the MLM is an artificial neural network, ANN, a backpropagation algorithm may be employed.

An MLM may in particular comprise an ANN, a support-vector machine, a k-means cluster algorithm, a decision tree, and so on. In particular an ANN may be or comprise a deep neural network and/or a convolutional neural network, CNN, in particular a deep CNN, and/or a recurrent neural network, RNN, in particular a recurrent CNN, and/or a transformer network and/or a generative adversarial network, GAN. The MLM may for example be an ANN or be based thereon. The first radar data and the second radar data are in particular input data, which are supplied to the MLM, in order to emulate the third radar data. The learning algorithm is also applied to the input data. The input data in particular depends on the first radar data and the second radar data or comprises these or consists of these.

In particular, the learning algorithm is applied to the input data. If applicable, the learning algorithm outputs the angle of arrival and/or the azimuthal angle and/or the elevation angle directly. Alternatively, the angle is calculated depending on an output of the learning algorithm.

By localizing the object by the trained machine learning algorithm with at least two radar data, in particular the first radar data and the second radar data, the angle resolution and/or accuracy of the determined angle is increased.

In an embodiment the first radar data is generated depending on a part of a first signal sent by the first radar sensor that is reflected from the object. Alternatively or additionally, the second radar data is generated depending on a part of a second signal sent by the second radar sensor that is reflected from the object.

For example a part of the first signal which is reflected from the object is received by the first radar sensor. For example the first radar sensor generates the first radar data depending on the received part of the first signal. Prior to receiving the second radar data, if applicable, a second signal is sent by the second radar sensor and a part of the second signal which is reflected from the object is received by the second radar sensor. For example the second radar sensor generates the second radar data depending on the received part of the second signal. In particular the generated first and second radar data are provided to the data processing system, for example the evaluation unit. If applicable, this embodiment may be a measuring method.

In an embodiment the first radar signal and the second radar signal are radar signals that are emitted incoherently or in a master-slave configuration. In particular, the radar signals are emitted according to the configuration.

For example in an incoherent configuration each element of the radar sensor has a signal shape generator of its own with an individual signal shape as a consequence. This individual signal shape is for example the basis for an assignment of reflected echo signals to their source, that is an element.

In the case of a master-slave configuration for example an, in particular single, element of the antenna array sends whilst the other elements receive. If applicable, in a next cycle another element of the antenna array sends.

In an embodiment the first radar data correspond to data of a virtual first antenna array. The second radar data correspond to data of a virtual second antenna array.

A virtual antenna array is in particular a fictitious unit. The elements of a real array that are present in real behave in the way of a larger virtual array with more elements than the real one. An increased directivity and an improved angle resolution may be achieved that would be the case without the virtual elements.

In an embodiment the first radar data and the second radar data, in particular the respective radar frequency data, correspond to complex numbers.

In an embodiment the input data are provided as at least one vector. Each vector element of the at least one vector is a real part or an imaginary part or an absolute value or a phase of one of the complex numbers.

For example the first radar data and the second radar data are all integrated into a single vector so that each complex number of the first radar data and the second radar data provides at least one vector element, for instance the real part, the imaginary part, the absolute value, or the phase. For example the first radar data comprises two complex numbers Re11 + i*lm11 , Re12+i* Im12 and the second radar data comprises for example two second complex numbers Re21+i* Im21 , Re22+i* Im22. If applicable, the vector comprises the vector elements (Re 11 , Re12, Re21 , Re22, Im11 , Im12, Im21 , Im22). This may be extended in analogy to more than two complex numbers.

Alternatively, the input data are provided as n vectors, wherein n corresponds to a number of radar sensors. For example the input data in the case of two radar sensors are provided as two vectors. In this connection the first vector for example comprises the first radar data and the second vector comprises the second radar data. For the named example then two vectors are provided, wherein a first vector comprises the elements (Re11 , Re12, Im11 , Im12) and the second vector comprises the elements (Re21 , Re22, Im21 , Im22). This may be extended in analogy to more than two complex numbers.

It is also possible that the input data is provided as several vectors, wherein an allocation of the radar data into the vectors is not according to the radar sensors, but rather for example according to the kind of a component of the complex numbers. For example, the at least one real part vector comprises all real parts of the first and the second radar data. For example at least one imaginary part vector comprises all imaginary parts of the first and the second radar data. For example at least one absolute value vector comprises all absolute values of the first and the second radar data. For example at least one phase vector comprises all phases of the first and the second radar data. In this connection it is also possible that the real parts of the first radar data are provided in a first real part vector and the real parts of the second radar data are provided in a second real part vector. The same applies in analogy to the imaginary part vector, the phase vector, and the absolute value vector. For the named example, for example a real part vector with the vector elements (Re11 , Re12, Re21 , Re22) and an imaginary vector with the vector elements (Im11 , Im12, Im21 , Im22) is provided. Alternatively, in the example two real part vectors are provided, wherein a first real part vector comprises the elements (Re11 , Re12) and a second real part vector comprises the vector elements (Re21 , Re22).

In preferred embodiments the real part vector and the imaginary part vector or alternatively the phase vector and the absolute value vector are provided to the machine learning algorithm. In these two variants the two vectors contain all information of the complex numbers.

In an embodiment the machine learning algorithm is designed as regression algorithm. Depending on the input data the machine learning algorithm outputs the angle of arrival for the object. For example the learning algorithm is designed as regressing artificial neural network, in short regression network. Therein the input data, in particular the at least one input vector, is transferred into a regression problem by a covariance matrix.

In an embodiment the machine learning algorithm depending on the input data generates an angle spectrum as output. The angle spectrum is for example a two-dimensional matrix. A first dimension of the matrix corresponds to the azimuthal angle. A second dimension of the matrix corresponds to the elevation angle. Elements of the matrix indicate probability values. Depending on the generated angle spectrum the azimuthal angle and/or the elevation angle and/or the angle of arrival of the object is determined. This embodiment may, in particular, be referred to as spectrum-based approach. For example the angle spectrum is determined as output of the MLM and the corresponding angle with the highest probability value depending on the output, in particular as post processing step. Alternatively, determining the corresponding angle is effected by the MLM, which means that the output of the MLM comprises the corresponding angle.

A further aspect relates to a method for guiding a vehicle at least partially automatically. In particular a method according to one of the above-named aspects or an embodiment thereof and in particular depending on the localizing of the object at least one control signal for guiding the vehicle at least partially automatically is generated. In particular, depending on the localizing of the object means that depending on a result of the localizing, that is for example depending on distance and/or azimuthal angle and/or elevation angle, the assistance information and/or the control signals are generated. Since by the method according to the invention and their embodiments the localizing is improved, the vehicle may also be guided in an improved manner.

A further aspect of the invention relates to a data processing system, which is adapted for performing a method according to the above-referenced aspects or an embodiment thereof. In particular, the data processing system performs the corresponding method. The data processing system comprises at least one data processing device. The at least one data processing device is in particular adapted for performing a method according to the invention.

The terms “data processing system” and “at least one data processing device” may be interchangeably used within the scope of the present disclosure. In the present disclosure a data processing device may for example be taken as a device with processing circuitry for processing data. This means that a data processing device may perform computing operations in order to process data. An indicated access to a data structure, for example a look-up table, LUT, or a database, may equally be regarded as computing operation.

A data processing device may in particular comprise one or several computers, one or several microcontrollers and/or one or several integrated circuits, for example one or several application-specific integrated circuits, ASIC, one or several field-programmable gate arrays, FPGA, and/or one or several systems-on-a-chip, SoC. A data processing device may also contain one or several processors, for example one or several microprocessors, one or several central processing units, CPU, one or several graphic processor units, GPU, and/or one or several signal processors, in particular one or several digital signal processors, DSP. The data processing device may also comprise a physical or virtual cluster of computers or other ones of the named devices.

A data processing device may also comprise one or several hardware and/or software interfaces, for example for the reception and/or the provision of data.

A data processing device may also comprise one or more one or more memory units. Therein, a memory unit may be implemented as a volatile data memory, for example as a dynamic random access memory, DRAM, or a static random access memory, SRAM, or as a non-volatile data memory, for example a read-only memory, ROM, a programmable read-only memory, PROM, an erasable programmable read-only memory, EPROM, an electrically erasable programmable read-only memory, EEPROM, a flash memory or flash EEPROM, a ferroelectric random access memory, FRAM, as a magnetoresistive random access memory, MRAM, or as a phase-change random access memory, PCRAM.

A further aspect of the invention relates to a radar system for localizing an object. The radar system comprises a first radar sensor and a second radar sensor and a data processing system according to the invention. The first radar sensor is configured to generate first radar data of a first antenna array of the first radar sensor and to provide it to the data processing system, for example the evaluation unit. The second radar sensor is configured to generate the second radar data of a second antenna array of the second radar sensor that is different from the first antenna array and to the data processing system, for example the evaluation unit.

It is possible that the radar system comprises one or several pairs of further first and further second radar sensors, the antenna arrays or which preferably each are spaced apart from one another. If applicable, what has been set out as to the first and second radar sensor applies in analogy to the further first and second radar sensors. In particular, the radar system is configured for performing a method according to one of the above- named aspects or an embodiment.

In an embodiment the first radar sensor and the second radar sensor are configured as monostatic MIMO radar systems and/or frequency-modulated continuous wave radars (FMCW). These facilitate an improved localizing of the object. A further aspect of the invention relates to an electronic vehicle guidance system for guiding a vehicle at least partially automatically. The electronic vehicle guidance system comprises a radar system according to the above-named aspect or an embodiment thereof. The electronic vehicle guidance system is for example configured to guide the vehicle at least partially autonomously or at least partially automatically depending on the localizing of the object. The at least one control signal may for example be provided to one or several actuators of the vehicle, including for example one or several braking actuators and/or one or several steering actuators and/or one or several propulsion motors of the vehicle. The one or several actuators may influence a longitudinal and/or transverse control of the vehicle in order to guide the vehicle at least partially automatically.

An electronic vehicle guidance system may also be taken as an advanced driver assistance system, ADAS, which assists the driver in partially automated or partially autonomous driving. In particular, the electronic vehicle guidance system may implement a partly automatic or partly autonomous driving mode according to levels 1 to 4 of the SAE J3016 classification. Here and in the following, SAE J3016 refers to the respective standard dated April 2021 . The assistance information may be output via an output device of the vehicle, for example a display and/or an audio output system and/or a haptic output system.

In an embodiment the data processing system is configured to generate at least one control signal for guiding the vehicle at least partially automatically depending on the localizing of the object. Alternatively or additionally, the data processing system is configured to generate assistance information for assisting a person guiding the vehicle in guiding the vehicle.

A further aspect of the invention relates to a vehicle, which comprises the vehicle guiding system according to the invention or its embodiments.

According to the further aspect of the invention a computer program comprising instructions is indicated. If the instructions are executed by a data processing system, the instructions cause the data processing system to perform a method according to the invention. The instructions may be provided as program code, for example. The program code may for example be provided as binary code or assembler and/or as source code of a programming language, for example C, and/or as program script, for example Python.

According to a further aspect of the invention a computer-readable storage medium which stores a computer program according to the invention is indicated.

The computer program and the computer-readable storage medium are each computer program products comprising the instructions.

Further features of the invention derive from the claims, the figures, and the description of the figures. The features and feature combinations previously mentioned in the description as well as the features and feature combinations shown in the following in the figures alone may be used not only in the respectively indicated combination but also in other combinations without leaving the scope of the invention. Thus, also implementations of the invention are to be regarded as comprised and disclosed that are not explicitly shown and explained in the figures, however, derive by separated feature combinations from the explained implementations and can be generated therefrom. Also implementations and feature combinations are to be regarded as disclosed that thus do not comprise all features of an originally formulated independent claim. Moreover, embodiments and feature combinations are to be regarded as disclosed, in particular by the above set out implementations that go beyond the feature combinations set out in the back-references of the claims or deviate therefrom.

In the following embodiments of the invention are described. These show in:

Fig. 1 a vehicle with an embodiment of a radar system according to the invention;

Fig. 2 a flow diagram of an embodiment of a method for radar-based localizing of an object according to the invention;

Fig. 3 a flow diagram of a further embodiment of a method for radar-based localizing of an object according to the invention; and

Fig. 4 a flow diagram of a further embodiment of a method for radar-based localizing of an object according to the invention. In the figures same reference signs each designate elements having the same function.

Fig. 1 shows a vehicle 1 with an embodiment of a radar system 2 according to the invention. The radar system 2 comprises for example an embodiment of a data processing system 3 according to the invention. If applicable, the radar system 2 comprises a first radar sensor 4 and a second radar sensor 5, which are attached in different places, however, in particular with overlapping fields of view, to the vehicle 1 . For example the first radar sensor 4 and the second radar sensor 5 both are front radars, or both are rear radars, or both are radars of the vehicle 1 attached to a left or right side of the vehicle 1 . For example the first radar sensor 4 and the second radar sensor 5 are configured as MIMO radars and/or as FMCW radars. For example the first radar sensor 4 is configured to generate first radar data 7 and the second radar sensor 5 is configured to generate second radar data 8.

Fig. 2 shows a schematic representation of an embodiment of a flow diagram of a method for radar-based localizing of an object by a radar system 2 according to the invention. For example the data processing system 3 receives first radar data 7 of a first antenna array of the first radar sensor 4. For example the data processing system 3 receives second radar data 8 of a second antenna array of the second radar sensor 5 that is different from the first antenna array. In particular, the data processing system 3 determines an azimuthal angle and/or an elevation angle and/or an angle of arrival for the object depending on the input data 6, which comprises the first radar data 7 and the second radar data 8, by a trained learning algorithm 9. The azimuthal angle and/or the elevation angle and/or the angle of arrival may be referred to as output data 10.

In an embodiment the first radar data 7 and the second radar data 8 corresponds to complex numbers. For example the first radar data 7 and the second radar data 8 each are integrated into a vector and, as exemplarily shown in Fig. 2, provided as two vectors to the machine learning algorithm 9, which may also be referred to as neural network.

In the embodiment of the flow diagram of the method for localizing an object shown in Fig. 3, the first radar data 7 and the second radar data 8 is for example integrated into a single vector. Each element of the single vector is for example a real part 11 or an imaginary part 12 or an absolute value or a phase of the complex numbers. Alternatively, all real parts 11 of the input data 6 are integrated into a vector and all imaginary parts 12 of the input data 6 are integrated into a further vector, as it is exemplarily shown in Fig. 4. In an alternative embodiment all absolute values of the input data 6 are integrated into the vector and all phases of the input data 6 into the further vector.

The output data 10 may for example be output as list of azimuthal angles and/or elevation angles and/or angles of arrival, as exemplarily shown in Fig. 3.

Alternatively, the output data 10 is output as two-dimensional matrix, as exemplarily shown in Fig. 4. For example a dimension of the matrix comprises the elevation angles and a second dimension of the matrix comprises the azimuthal angle. For example the neural network determines combinations of azimuthal angles and elevation angles and marks these for example with a binary 1 in the matrix and the inapplicable combinations with a binary 0.

The explained different data formats of the input data 6 and the output data 10 are to be taken as capable of being combined with one another and merely exemplary. For example input data 6, which are integrated into a single vector, as shown in Fig. 3, may be provided to the machine learning algorithm 9 and the machine learning algorithm for example outputs the two-dimensional matrix as output data 10, as shown in Fig. 4.

Claims

1 . Method for radar-based localizing of an object by a radar system (2), comprising the steps: receiving first radar data (7) of a first antenna array of a first radar sensor (4) of the radar system (2); receiving second radar data (8) of a second antenna array of a second radar sensor (5) of the radar system (2) that is different from the first antenna array; and determining an azimuthal angle and/or an elevation angle and/or an angle of arrival for the object depending on input data (6), which comprises the first radar data (7) and the second radar data (8), by a trained learning algorithm (9).

2. Method according to claim 1 , wherein the first radar data (7) is generated depending on a part of a first signal sent by the first radar sensor (4) that is reflected from the object and/or the second radar data (8) is generated depending on a part of a second signal sent by the second radar sensor (5) that is reflected from the object.

3. Method according to claim 2, wherein the first radar signal and the second radar signal are radar signals emitted incoherently or in a master-slave configuration.

4. Method according to any one of the preceding claims, wherein the first radar data (7) corresponds to data of a virtual first antenna array and the second radar data (8) corresponds to data of a virtual second antenna array.

5. Method according to any one of the preceding claims, wherein the first radar data (7) and the second radar data (8) correspond to complex numbers.

6. Method according to claim 5, wherein the input data (6) are provided as at least one vector, wherein each vector element of the at least one vector is a real part (11) or an imaginary part (12) or an absolute value or a phase of one of the complex numbers.

7. Method according to any one of the preceding claims, wherein the machine learning algorithm (9) is designed as regression algorithm and outputs the angle of arrival for the object depending on the input data (6).

8. Method according to any one of claims 1 to 6, wherein the machine learning algorithm (9) generates an angle spectrum depending on the input data (6), wherein the angle spectrum is a two-dimensional matrix, wherein a first dimension of the matrix corresponds to the azimuthal angle and a second dimension of the matrix corresponds to the elevation angle and elements of the matrix indicate probability values, wherein the azimuthal angle and/or the elevation angle of the object is determined depending on the generated angle spectrum.

9. Method for at least partially automatically guiding a vehicle (1 ), wherein a method according to any one of the preceding claims is performed and depending on the localizing of the object at least one control signal for guiding the vehicle (1 ) at least partially automatically is generated; and/or assistance information for assisting a person guiding the vehicle (1) in guiding the vehicle (1 ) is generated.

10. Data processing system (3), which is adapted for performing a method according to any one of the preceding claims.

11 . Radar system (2) for localizing an object, comprising a first radar sensor (4) and a second radar sensor (5) and a data processing system (3) according to claim 10, wherein the first radar sensor (4) is configured to generate first radar data (7); the second radar sensor (5) is configured to generate second radar data (8).

12. Radar system (2) according to claim 11 , wherein the first radar sensor (4) and the second radar sensor (5) are configured as monostatic MIMO radar systems (2) and/or frequency-modulated continuous wave radars (FMCW).

13. Electronic vehicle guidance system for guiding a vehicle (1 ) at least partially automatically, comprising a radar system (2) according to claim 11 , wherein the electronic vehicle guidance system is configured to guide the vehicle (1) depending on the localizing of the object.

14. Electronic vehicle guidance system according to claim 13, wherein the data processing system (3) is configured to generate depending on the localizing of the object at least one control signal for guiding the vehicle (1 ) at least partially automatically; and/or assistance information for assisting a person guiding the vehicle (1) in guiding the vehicle (1 ),

15. Computer program product comprising instructions, which, when they are executed by a data processing system (3) cause the data processing system (3) to perform a method according to any one of claims 1 to 9.