DE102022204776B3 - Method for locating a vehicle within a SAR image - Google Patents

Abstract

The invention relates to a method for locating a vehicle (1) within an SAR image of a radar system of the vehicle (1), in which the radar system emits radar signals which are reflected by objects in the area surrounding the vehicle (1), the radar system using Radar detections create a SAR signature (6) for the detected objects and based on this a SAR image of the area surrounding the vehicle (1) is generated, the radar system records individual detections (7) and the individual detections are compared with the SAR image in order to Locate the vehicle (1) in the SAR image.

Description

The present invention relates to a method for locating a vehicle within an SAR image of a radar system of the vehicle, a radar system for a vehicle for generating a SAR image, a vehicle having a corresponding radar system and use of the radar system according to the invention.

Technological background

Radar techniques are known from the prior art, by means of which high-resolution radar information, also referred to below as radar image, can be obtained. Such high-resolution radar images are particularly relevant for autonomous driving of vehicles. Synthetic Aperture Radar (SAR) in particular enables high spatial resolution. The calculation of high-resolution radar information, especially in the case of SAR radar systems, requires a great deal of computing effort because, in addition to the highly precise knowledge of the trajectory of the vehicle on which the radar system is located, each grid area or pixel of the radar image is calculated taking into account several ramp-like or pulse-like radar signals must become. The system requirements for a radar system with high spatial resolution and real-time or quasi-real-time requirements are therefore very high, particularly with regard to the size of the system memory and the processor capacity.

State of the art

For highly automated driving functions such as the Highway Pilot or fully automated parking without driver intervention, redundant sensor systems are required that allow highly accurate and robust localization. The localization can be realized, for example, on the basis of visual landmarks (as described, for example, in S. Luthardt, V. Willert and J. Adamy, "LLama-SLAM: Learning High-Quality Visual Landmarks for Long-Term Mapping and Localization", 21st International Conference on Intelligent Transportation Systems (ITSC), Maui, Hawaii, USA, 2018, p Visibility such as rain, fog or snowfall may no longer reliably detect relevant landmarks. Such localization systems based on visual landmarks can therefore fail more easily in critical scenarios or deliver inaccurate results, for example if the weather is bad and the relevant landmarks are not in sight are.

In contrast, so-called radar-based SLAM (Simultaneous Localization and Mapping) methods have an advantage. Here, however, the achievable angular resolution and the associated accuracy of the radar map is limited by the limited number of antennas and aperture size in the radar sensor. High-resolution radar sensors are also associated with increased costs and manufacturing effort. For example, in Jürgens, N. Koch and M. -M. Meinecke, "Radar-based Automotive Localization using Landmarks in a Multimodal Sensor Graph-based Approach," 21st International Radar Symposium (IRS), 2020, pp. 373-378, doi: 10.23919/IRS48640.2020.9253921 described a system which landmarks on Detected based on radar data. The DE 10 2020 205 187 A1 describes a radar device with at least one radar sensor and a computing device, which is designed to evaluate the radar measurement data of the radar sensor using SAR methods to generate a radar image, the radar image is later evaluated using a SLAM method to generate a map of the surroundings and the locate at least one radar sensor on the map of the area. Absolute localization can be achieved with the radar device by comparing the generated map of the surroundings or a map section with a stored known map of the surroundings, a comparison of a few landmarks taking place.

Furthermore, previous systems usually work with accumulated radar point clouds. A special approach is also the generation of SAR images using a Synthetic Aperture Radar (SAR), as described for example in H. Iqbal, A. Loeffler, M. Mejdoub, F. Gruson, "Realistic SAR Implementation for Automotive Applications", 17th European Radar Conference, Utrecht. The use of SAR promises a significantly higher resolution of the generated map with little hardware effort. It can z. B. already installed side radars can be used, with theoretically only one radar with only one antenna is sufficient. For example, also in DE 10 2015 201 828 A1 and DE 10 2017 123 902 A1 Methods for radar systems are described in which images are generated on the basis of SAR. Further describes DE 10 2018 124 215 A1 a method in which location information is assigned to the received information received from the respective radar sensors, which information forms a location reference for the received information.

The DE 10 2017 221 691 A1 describes a method for self-localization of a vehicle in its environment, with the environment being detected by means of radar sensors while the vehicle is driving and a map based on reflex points is created, whereby non-stationary reflex points are sorted out and only stationary reflex points are used. The reflex points are then compared with those on the map to determine the vehicle's position on the map.

Out of JP 5 047 290 B2 a method is known in which a radar sensor with a synthetic aperture generates a SAR image when the vehicle moves e.g. B. moved along a parking lot, and by processing the width and length of the parking space is determined.

Out of DE 10 2020 207 879 A1 method for operating a radar sensor in a motor vehicle is known, in which, in a SAR measurement mode, objects, including stationary objects, are located with high angular resolution in a SAR measurement mode based on the principle of the synthetic aperture, with the same radar sensor being operated with a time delay or at the same time in the SAR measurement mode and in a Doppler measurement mode is operated, the relative speeds of objects, including moving objects, being measured in the Doppler measurement mode with a time resolution that is greater than the time resolution in the SAR measurement mode. It is possible to switch between SAR measurement mode and Doppler measurement mode depending on the traffic situation.

Also describes the DE 10 2014 217 847 A1 a driver assistance system for updating a digital map, which determines a deviation between a relative position of a landmark detected by the vehicle-internal environment sensor device and the relative position obtainable from the digital map. This deviation can then be verified through communication between neighboring vehicles and then used to update the digital map.

Object of the present invention

Proceeding from this, it is the object of the invention to specify a method by means of which reliable and highly precise localization in automotive SAR images is made possible.

Solution according to the invention

The above object is achieved by the combination of features of claim 1 and the independent claims. Further developments of the invention are claimed in the dependent claims.

In the method according to the invention for locating a vehicle within an SAR image of a radar system of the vehicle, the radar system emits radar signals which are reflected on objects in the vicinity of the vehicle and are received again by the radar system. The radar system can then use the radar detections to create an SAR signature for the detected objects and, based on this, generate an SAR image of the vehicle's surroundings. Furthermore, the radar system also records individual detections or individual reflections, which are compared with the SAR image, with the individual detections being incorporated into the SAR image or the SAR signatures recorded in the SAR image being superimposed or compared with the individual detections then locate the vehicle in the SAR image. The method according to the invention provides additional functionality for more precise localization (autonomous driving level 3/4). This results in the advantage that a significantly higher lateral resolution is achieved than in the conventional radar cycle. In addition, the method provides a cost-effective implementation of high-precision localization tasks without additional sensors, which enables reliable and high-precision localization in automotive SAR images.

A first mode is preferably provided in which the radar system records radar detections for an SAR signature and a second mode is provided in which the radar system records individual detections.

The first mode and the second mode can expediently take place alternately one after the other or in parallel next to one another.

According to a further embodiment of the invention, other objects can also be localized in the SAR image using the comparison of the individual detections with the SAR image.

Furthermore, navigation data can also be recorded and/or received (e.g. by the navigation system), which are stored in the SAR image, in particular pixel by pixel.

The navigation data can expediently be provided for locating the vehicle or objects in the vicinity of the vehicle.

Furthermore, the present invention claims a radar system for a vehicle, comprising at least one radar sensor, which generates a SAR image, with the vehicle being localized using a method according to the invention.

In addition, the present invention claims a vehicle comprising a radar system according to the invention.

Furthermore, the present invention also includes the use of a radar system according to the invention for checking the functionality of the radar system. In this case, radar detections are recorded when the vehicle is stopped (e.g. when the vehicle is switched off/parked), a SAR image being created and the vehicle and/or objects in the SAR image being localized using the method according to the invention. The first SAR image can then be stored in a memory of the radar system, for example. Radar detections can then be recorded when the vehicle is started, with the vehicle and/or objects in the SAR image being localized using the radar detections in the last SAR image. Localization is important here to determine that the vehicle has not moved between stop and start. The radar detections can then be compared with the radar detections in the SAR image, so that deviations between them are recorded (e.g. deviations in the image or the localization), whereby conclusions can be drawn about the functionality of the radar system based on the deviations. For example, a misalignment or a sensor misalignment can be assumed if the environment in the SAR image changes (eg with regard to stationary objects such as crash barriers, roadside buildings or the like) but the vehicle 1 has the same localization. In a practical way, the radar system can then inform the driver of the vehicle of such a misalignment.

Description of the invention based on exemplary embodiments

• 1 eine vereinfachte Darstellung eines Verkehrsszenarios, bei dem ein Fahrzeug mittels Radarsensorik ein SAR-Bild erstellt gemäß einer Ausgestaltung des erfindungsgemäßen Verfahrens;
• 2 eine vereinfachte Darstellung des Verkehrsszenarios aus 1 mit der eingetragenen SAR-Signatur des erstellten SAR-Bildes;
• 3 eine vereinfachte Darstellung des Verkehrsszenarios aus 1, bei dem das Fahrzeug mittels Radarsensorik Einzeldetektionen erstellt, sowie
• 4 eine vereinfachte Darstellung des SAR-Bildes aus 2 mit den eingetragenen Einzeldetektionen aus 3.

The invention is explained in more detail below using expedient exemplary embodiments. Show it:

• 1 a simplified representation of a traffic scenario in which a vehicle uses radar sensors to create a SAR image according to an embodiment of the method according to the invention;
• 2 a simplified representation of the traffic scenario 1 with the entered SAR signature of the created SAR image;
• 3 a simplified representation of the traffic scenario 1 , in which the vehicle creates individual detections using radar sensors, as well as
• 4 a simplified representation of the SAR image 2 with the entered individual detections 3 .

In 1 a traffic situation is shown in which a vehicle 1 is equipped with a radar system according to the invention. For this purpose, the radar system comprises a radar sensor 2 with a longer range (e.g. up to 300 m) and 4 short-range radars 3a-3d, which are arranged at the corners of the vehicle. However, it is also conceivable that the radar system has more or fewer radar sensors. The radar system used here can both create a SAR image and generate individual detections. This means that obstacles in the area around the vehicle can be detected, e.g. B. Roadside development, a parked vehicle 5, curbs and the like. For the solution according to the invention, localization is also made possible with the aid of individual radar detections in the previously created SAR image.

According to 2 a SAR signature 6 is first created and a SAR image is generated from it, which z. B. can be done using a conventional method for creating a (highly accurate) SAR image. For example, SAR signatures 6 can be parking spaces and the surroundings of a parking space, curbs, vegetation, other road users such as parked cars 5 and the like. The vehicle 1 thus creates a map or a SAR image based on SAR data or a SAR signature while driving. The SAR image can then be stored with navigation or GPS data from the navigation system, the on-board computer of the vehicle 1 or the like, or the GPS coordinates can be assigned to the pixels in the SAR image.

After that, individual detections 7 are recorded, as in 3 shown. This can e.g. This can be done, for example, by switching the radar measurement method from a first mode (SAR mode) to a second mode (single detection mode) or by detecting them side by side in parallel. The objects are then localized in the SAR image using individual radar detections, with the individual detections being incorporated into the SAR image or the SAR signatures recorded in the SAR image being superimposed or matched with the individual detections 7 in order to identify the objects locate as in 4 shown. The vehicle can then be localized in the SAR image or also objects in the SAR image by means of the radar detection. Furthermore, the navigation or GPS data can be used to globally localize the vehicle 1 or other objects in the vehicle environment.

The method can be used in a practical way in parking situations: when driving past a parking space, this can be precisely measured using SAR, with a map of the surroundings or a SAR image being generated. By driving past the parking space completely, it is then possible to measure it completely. In a surprising way, the following actual Parking bring benefits to use radar single detections for localization in the SAR image. For example, due to the way SAR works, the measurement is only complete after the vehicle has passed objects. It is particularly advantageous here to link the SAR and radar single detection mode for localization.

Furthermore, the present method can be used to detect a sensor misalignment in that, after the vehicle has been restarted, the radar detections can be compared or adjusted with the SAR image stored when the vehicle stopped. If deviations are then present, these can be used for automatic sensor alignment, in which the sensor adjusts itself automatically, or it can be used to indicate a misalignment to the driver so that he can visit a workshop.

Claims (9)

A method for locating a vehicle (1) within a SAR image of a radar system of the vehicle (1), in which the radar system emits radar signals which are reflected by objects in the vicinity of the vehicle (1), the radar system uses radar detections to create an SAR signature (6) for the detected objects and, based on this, generates an SAR image of the vehicle's surroundings (1), the radar system records individual detections (7), and the individual detections (7) are compared with the SAR image, the individual detections (7) being incorporated into the SAR image by the SAR signatures (6) recorded in the SAR image being overlaid with the individual detections (7) in order to locate the vehicle (1) in the SAR image. procedure after claim 1 , characterized in that a first mode is provided in which the radar system detects radar detections for a SAR signature and a second mode is provided in which the radar system detects individual detections. procedure after claim 2 , characterized in that the first mode and the second mode take place alternately one after the other or in parallel. Method according to one of the preceding claims, characterized in that other objects are also localized in the SAR image on the basis of the comparison of the individual detections with the SAR image. Method according to one of the preceding claims, characterized in that navigation data are recorded and/or received which are stored in the SAR image. procedure after claim 5 , characterized in that the navigation data are intended to locate the vehicle (1) or objects in the vicinity of the vehicle (1). Radar system for a vehicle (1), comprising at least one radar sensor (2, 3a-3d) which generates an SAR image and localizes the vehicle (1) using a method according to one of the preceding claims. Vehicle (1), comprising a radar system claim 7 . Using a radar system according to claim 7 for checking the functionality of the radar system, with radar detections being recorded when the vehicle (1) stops, with a SAR image being created and the vehicle (1) and/or objects in the SAR image using a method according to one of Claims 1 - 6 is/are localized, and radar detections are recorded when the vehicle (1) starts, the radar detections when the vehicle (1) starts are compared with the SAR image, so that deviations between the radar detections when the vehicle (1) starts and the Radar detections are recorded in the SAR image, and conclusions are drawn about the functionality of the radar system based on the deviations.

Images