Method for operating a driver assistance function

Abstract

Method for operating a driver assistance function of a vehicle (1) comprising at least one sensor (10) for detecting the vehicle environment and an evaluation device (21) for recognizing objects (8) from signals detected by the at least one sensor (10), characterized in that the method comprises the following steps: - Detection of the vehicle environment (11) using at least one sensor (10), - initial detection and initial classification of at least one specific object relevant for activating the driver assistance function by means of the evaluation device at a distance from the vehicle that is greater than a specified first threshold (12), - second detection and second classification of the at least one specific object by means of the evaluation device at a distance from the vehicle that is less than a specified second threshold (13), - if the first classification of the object differs from the second classification (14), classify the object as a falsely identified object (15), - Sending a signal with a number of data points about the object to a database (16, 35), and - if a driver assistance function was activated as a result of the initial classification and if the object was classified as a falsely identified object, adjust the driver assistance function (18).

Description

[0001] The present invention relates to a method for operating a driver assistance function of a vehicle and an assistance device for a vehicle. The invention further relates to a vehicle, for example a motor vehicle, a computer-implemented method, a computer program product, a computer-readable data carrier, and a data carrier signal.

[0002] Driver assistance functions are frequently used in vehicles, activating or deactivating specific functions depending on certain characteristics of the detected vehicle environment. This often requires object detection, which can be unreliable, especially when objects are detected at a great distance from the vehicle. Furthermore, the detection of the vehicle's surroundings can be dependent on the time of day, the season, and the weather. Therefore, it is essential to continuously validate object detection and further develop it to improve its reliability.

[0003] Document US 7 957 562 B2 describes the identification and classification of objects in the immediate vicinity of a vehicle using a camera. This involves adapting a depth map of the road surface.

[0004] Document US 10 748 032 B1 describes a method for improving the distance estimation of an object using a vehicle camera, whereby pitch calibration is performed and information from vehicle-to-vehicle (V2V) communication is used.

[0005] Document WO 2017 / 189 361 A1 describes a system and a procedure for detecting incorrect settings and other errors of a vehicle sensor. This uses V2V communication.

[0006] Document DE 10 2019 208 735 A1 describes a method for operating a driver assistance system in which sensor data from the vehicle's environment are successively recorded and evaluated, and control data for the semi-automated control of the vehicle are generated on the basis of the analyzed sensor data using a neural network.

[0007] Documents DE 10 2013 206 707 A1 and DE 10 2019 002 487 A1 describe environmental sensing systems of a vehicle.

[0008] Against the background described above, the object of the present invention is to provide an advantageous method for operating a driver assistance function of a vehicle. Further objects are to provide an advantageous driver assistance device, a vehicle, a computer-implemented method, a computer program product, a computer-readable data carrier, and a data carrier signal.

[0009] The aforementioned problems are solved by a method for operating a driver assistance function according to claim 1, a driver assistance device according to claim 8, a vehicle according to claim 9, a computer-implemented method according to claim 10, a computer program product according to claim 11, a computer-readable data carrier according to claim 12, and a data carrier signal according to claim 13. The dependent claims contain further advantageous embodiments of the invention.

[0010] The inventive method for operating a driver assistance function of a vehicle, for example a motor vehicle, relates to a vehicle comprising at least one sensor for detecting the vehicle's surroundings, in particular for detecting features of the vehicle's surroundings, and an evaluation unit for recognizing objects from signals detected by the at least one sensor. The method comprises the following steps: The vehicle's surroundings are detected by means of the at least one sensor. At least one specific object relevant for activating the driver assistance function is detected and classified by means of the evaluation unit at a distance from the vehicle that is greater than a defined first threshold value, i.e., above a defined minimum distance. Thus, an initial detection and initial classification of at least one object from a large distance takes place.In the next step, the evaluation unit performs a second detection and classification of at least one specific object at a distance from the vehicle that is less than a defined second threshold value; in other words, at a distance less than a defined maximum distance. Thus, the detection and classification of at least one object is repeated after the vehicle has approached it.

[0011] In a further step, if the first classification of the object differs from the second classification, the object is classified as a misidentified object. A signal containing data about the object is then transmitted to a database. This data can be stored locally in a database, for example, within the vehicle. Alternatively, or additionally, the data can be advantageously sent to a remote location, such as a cloud. If a driver assistance function was activated as a result of the first classification, the function is adjusted, for example, deactivated if the object was classified as a misidentified object.

[0012] The method according to the invention has the advantage of being independent of direct user input. This means that the user, for example the driver, is not prompted to manually enter information. Nevertheless, data is collected, analyzed, and the necessary information or environmental characteristics are stored in a database. If a database accessible to a large number of vehicle users is used, for example, a database stored in a cloud, the advantage arises that the database can be used by a large number of users, is simultaneously maintained by uploading data from a large number of vehicles, and is continuously improved in terms of its reliability. A corresponding cloud can therefore be designed to receive data from a vehicle fleet.It can be configured to allow only certain vehicles to upload data, but it can also generally allow every new vehicle to have a data connection, i.e., to both upload and download data.

[0013] By analyzing information stored in the cloud, a digital map, such as a roadbook, can be generated. The roadbook's content can be stored in the cloud and accessed via mobile devices. The roadbook is a digital map that includes only the types of information necessary for a specific application. For example, valid traffic signs, symbols, and / or irrelevant signs that could cause false positive detections—such as reflectors or reflective objects that could potentially cause misclassified light detections—can be stored in the roadbook to help prevent these false positives. The position and characteristics of these objects can also be stored within the roadbook.

[0014] The data is preferably validated before being stored in the roadbook. For example, a specific number of vehicles passing the relevant object may be required to trigger the recording of a corresponding feature. This number can be defined based on specific features and may vary for different features or objects. Once the roadbook is uploaded, the information can be made available via the cloud and downloaded by any vehicle. The information from the roadbook helps overcome limitations imposed by the sensors used. While a camera sensor should reliably provide the relevant information, it may turn out, especially at large distances from the object being detected, that the information provided by the sensor is inaccurate.For example, a reflector can reflect the light from the vehicle's own headlights, and the resulting signal, detected by the camera, might be misinterpreted as another vehicle or the lights of another road user. Once the vehicle has approached the object, the sensor, such as the camera, is able to identify and correctly classify it. However, driver assistance functions may have already been activated as a result of the initial detection. By using information from the roadbook, a detection algorithm can compare the current sensor data with information from the roadbook at specific positions in front of the vehicle. This enables correct classification even at greater distances and reliably corrects false positive detections.

[0015] In an advantageous embodiment of the inventive method, the vehicle's surroundings are detected stepwise or continuously using at least one sensor. Preferably, the at least one sensor is designed to detect the vehicle's surroundings in the direction of travel optically, for example by means of a camera, and / or acoustically.

[0016] The object being recognized can be, for example, a traffic sign and / or a marking, such as a lane marking, and / or a road user, such as a motor vehicle, moped or motorcycle, a cyclist, or a pedestrian. The evaluation device used for object recognition, in particular the software used in this context, is preferably designed for such object recognition.

[0017] The evaluation of the captured signals, and in particular object detection and classification, can be performed inside or outside the vehicle, manually and / or automatically. Conventional detection algorithms and / or machine learning algorithms, especially those using artificial neural networks (deep learning), can be employed.

[0018] Data about the object can include its position and / or at least one characteristic of the object, which can be sent to a database. This can be achieved, for example, by sending or transmitting at least one photograph, a video, or a video sequence to the database. Furthermore, information about the object can be transmitted or sent indicating whether it is static or moving.

[0019] In another variant, after the initial classification, data on the detected object can be compared with data from a database, for example, the aforementioned database. If the detected object corresponds to an object classified as incorrectly detected in the database, the driver assistance function can be adjusted, for example, deactivated. In this case, the adjustment, particularly the deactivation, can be performed before the object is detected and classified a second time. This approach has the advantage that an existing database, maintained or further developed using the inventive method, is efficiently used to detect and eliminate false positive detections.

[0020] Furthermore, the signals sent to the database by a plurality, preferably a large number, of users can be evaluated regarding their reliability in classifying certain objects as incorrectly identified. A corresponding confidence interval can be continuously adjusted for this purpose. Individual data records can be released to users, for example, when predefined thresholds are exceeded, such as the number of uploaded data records or entries relating to a specific object.

[0021] The assistance device according to the invention is designed for a vehicle which comprises at least one sensor, for example a front camera, for detecting the vehicle's surroundings, in particular for detecting features of the vehicle's surroundings, and an evaluation unit for recognizing objects from the signals detected by the at least one sensor. The assistance device is designed to carry out a previously described method according to the invention. The vehicle according to the invention comprises a previously described assistance device according to the invention and / or is designed to carry out a previously described method according to the invention. The vehicle can be a motor vehicle, for example a passenger car, a truck, a moped, a motorcycle, a bus or a minibus, a rail vehicle, or a ship.The assistance device and the vehicle according to the invention have the features and advantages already mentioned in connection with the method according to the invention.

[0022] The computer-implemented method according to the invention comprises instructions that, when the program is executed by a computer, cause it to execute a method according to the invention as described above. The computer program product according to the invention comprises instructions that, when the program is executed by a computer, cause it to execute a method according to the invention as described above. A computer program product according to the invention, as described above, is stored on the computer-readable data carrier. The data carrier signal according to the invention transmits the computer program product according to the invention. The computer-implemented method, the computer program product, the computer-readable data carrier, and the data carrier signal have the features and advantages already mentioned in connection with the method according to the invention.In particular, they make it possible to utilize the inventive method described above as part of a retrofit for existing vehicles.

[0023] The present invention offers the following long-term advantages: The acquired measured values ​​and / or data can be analyzed directly in the respective vehicle, and potential sources of false positive detections can be uploaded to a cloud. For example, a reflector at a great distance may be detected or recognized as a taillight. As the vehicle approaches the reflector, the sensor system and evaluation unit can recognize that it is a reflector and not a taillight and can upload the position and characteristics of this potential source of false positive detection to the cloud, thus ensuring that other vehicles avoid a similar false positive detection from a great distance.

[0024] In currently used systems, the evaluation and analysis of the vehicle's surroundings is performed entirely autonomously. This results in multiple vehicles making similar false positive detections. The present invention allows unnecessary actions by driver assistance functions due to false detections to be avoided or at least reduced. This improves driving comfort and increases user safety. For example, a driver assistance function might automatically dim the high beams following a false positive detection of a feature in the vehicle's surroundings as an oncoming vehicle. The present invention prevents such unnecessary dimming.

[0025] The data collected from the vehicle's environment, such as the detected signals, can be analyzed offline within the vehicle and processed for later use by an advanced CPU. Alternatively or additionally, automated analysis and evaluation can be performed using regularly updated software and / or various analytical approaches. Manual analysis is also possible.

[0026] Furthermore, the collected data can be analyzed based on data from various vehicles, taking into account different driving habits of different users, different times of day, seasons, weather conditions, etc., preferably in a suitable cloud environment. This improves the quality and reliability of the data made available to other users via such a database. With a large number of users uploading data to such a database, individual false detections or misinterpretations become negligible. Overall, this approach provides access to highly accurate data, which can be used as reference data in various scenarios, both online and offline, during a specific journey, reliably reducing false positives.

[0027] Uploading or downloading data can be done via mobile devices or mobile network services, such as those available in modern vehicles. Download options include: online data transfer in areas with good network coverage, and regular synchronization, for example, once a day, week, or month, within a specific radius of the vehicle. The download process can be tailored to the individual user's habits, particularly the user profile of the specific vehicle. For example, if the vehicle is frequently used in a particular area, downloads can be restricted to that area and occur more frequently than in other areas.Furthermore, the download can be performed in stages; for example, if the distance from a starting point exceeds a defined radius, such as 80 km, the data will be downloaded for a further defined radius, such as 100 km.

[0028] Data can be uploaded in the following ways: Online data transfer can be carried out in regions with good network coverage; data packages of a predefined size can be uploaded as soon as they are complete; the data for a complete route can be uploaded after the route has been completed.

[0029] Ideally, data from an entire vehicle fleet is provided and analyzed, with the decision of whether to include specific data in the roadbook depending on predefined quality criteria. Given the high volume of data involved, occasional upload or data processing difficulties are not a significant factor.

[0030] The invention is explained in more detail below with reference to exemplary embodiments and the accompanying figures. Although the invention is illustrated and described in detail by the preferred embodiments, the invention is not limited by the disclosed examples and other variations can be derived from them by a person skilled in the art without departing from the scope of protection of the invention.

[0031] The figures are not necessarily detailed or to scale and may be enlarged or reduced to provide a better overview. Therefore, the functional details disclosed here are not to be understood as limiting, but merely as an illustrative basis to guide those skilled in this field of technology in using the present invention in a variety of ways.

[0032] The expression "and / or" used here, when used in a series of two or more elements, means that each of the listed elements can be used alone, or any combination of two or more of the listed elements can be used. For example, when describing a composition containing the components A, B, and / or C, the composition can contain A alone; B alone; C alone; A and B in combination; A and C in combination; B and C in combination; or A, B, and C in combination. Fig. Figure 1 schematically illustrates the idea underlying the present invention in the form of a diagram. Fig. Figure 2 schematically shows a roadway as viewed through the windshield of a motor vehicle. Fig. Figure 3 schematically shows a method according to the invention in the form of a flowchart. Fig. Figure 4 schematically shows a vehicle according to the invention. Fig. Figure 5 schematically shows a flowchart of a further embodiment of the method according to the invention.

[0033] The Fig. Figure 1 schematically illustrates the underlying idea of ​​the present invention in the form of a diagram. Vehicles 1, for example, motor vehicles in a fleet, are each connected to a cloud 2 for data transmission 3. In other words, the individual vehicles 1 are designed to download data from and / or upload data to the cloud 2. The data transmission is indicated by arrows 3.

[0034] Within Cloud 2, uploaded data is processed and validated. This is indicated by an arrow (4). Preferably, a roadbook is created using the processed and analyzed data and the information stored in the cloud. The content of the roadbook is stored in Cloud 2 and can be accessed by individual vehicles (1) and / or mobile devices.

[0035] The roadbook is a digital map containing information required for specific applications, particularly driver assistance functions. This information can include, for example, valid traffic signs, symbols, displays, or irrelevant signs that could cause false positive detection of a traffic sign, reflectors that could trigger false positive light detections, and so on. The exact position and characteristics of these objects can be stored in the roadbook. Preferably, the data and information are validated within Cloud 2 before being stored in the roadbook. Suitable validation criteria can be defined as part of this validation process, such as a predefined minimum number of data points or detections of the same object uploaded to the cloud.

[0036] An example of a false positive detection is given below using the following: Fig. 2 explained. Fig. Figure 2 schematically shows a roadway 5, viewed through the windshield 9 of a motor vehicle 1. The motor vehicle 1 has a driver assistance function that automatically dims the headlights of the motor vehicle 1 when oncoming traffic is detected by means of a camera of the motor vehicle 1 (not shown).

[0037] In the Fig. 2. Using a front camera, two adjacent lights are detected within the circled image area 6 at a distance greater than a defined threshold or minimum distance and recognized as an object 8. These are classified by an evaluation unit as possible headlights of an oncoming vehicle. Triggered by this classification, the driver assistance function is activated and the headlights are automatically dimmed.

[0038] After traveling a further distance, which results in the vehicle 1 being at a distance from the detected object 8 that is less than a defined second threshold value, i.e., less than a maximum distance, the vehicle's surroundings are re-scanned by the camera, and in this context, the adjacent light points 8 are again detected and classified. This is in the Fig. 2 shown in section 7.

[0039] During the second detection and classification of the object, the classification differs from the first classification, resulting in the conclusion that the adjacent light points are not the headlights of an oncoming vehicle, but rather construction site lighting. The first and second classifications of the same object 8 thus diverge. In this case, the already activated driver assistance function is deactivated, and the headlights are switched back on. Furthermore, the detected object, in this case the adjacent light points, is classified as a "falsely detected" object, for example, incorrectly detected oncoming traffic. The position of the object, i.e., the construction site lighting, and its characteristics, such as two adjacent lamps that could lead to a potentially incorrect classification, can be uploaded to Cloud 2.Additionally or alternatively, the data and information can be stored in a database within the vehicle 1 and, if necessary, uploaded to Cloud 2 at a later time.

[0040] Within Cloud 2, uploaded information and data can be verified, for example, through offline analysis or by evaluating multiple comparable data sets for the same object, uploaded by different vehicles or users. Following verification and / or validation, the relevant information can be made available for download to other users. Other vehicles 1 approaching, for example, construction site lighting, are thus informed via relevant data from Cloud 2 that the construction site lighting could potentially be detected and classified as a false positive for oncoming traffic. Data captured by a front camera or front sensor of another vehicle 1 can then be compared with the data from Cloud 2, and the classification of the object detected from a great distance can be adjusted immediately.

[0041] The Fig. Figure 3 schematically shows a method according to the invention in the form of a flowchart. In a first step 11, the vehicle environment of a vehicle, which comprises at least one sensor for detecting the vehicle environment and an evaluation device for recognizing objects from the measured values ​​acquired by the at least one sensor, is detected by means of the at least one sensor. The sensor can, for example, be a camera, preferably a front camera of the vehicle 1.

[0042] In a second step 12, the evaluation unit performs an initial detection and classification of at least one specific object relevant for activating the driver assistance function. The object 8 is detected at a distance from the vehicle 1 that is greater than a defined initial threshold. In the Fig. In the example shown in Figure 2, this step corresponds to recognizing the image section 6 and subsequently classifying the light pattern detected at a large distance as the headlights of an oncoming motor vehicle.

[0043] In the next step 13, a second detection of the already detected and classified object takes place at a distance from the vehicle that is less than a defined second threshold. In the Fig. In the example shown, this corresponds to capturing image section 7 using a camera. Furthermore, a second classification of the object, for example the detected light pattern, is carried out using the evaluation unit. In the example shown in the Fig. In the second example shown, the second classification would lead to the result that the detected light pattern is classified as construction site lighting.

[0044] In step 14, the object's first classification is compared with its second classification. If the first and second classifications do not differ, the process ends in step 19. If they do differ, the object is classified as incorrectly identified in step 15. Optionally, further characteristics and data of the object can be recorded and stored. For example, whether the object is static or moving can be recorded and stored.

[0045] In step 16, at least one signal containing data about the object is sent to a database, for example, a cloud 2 and / or a locally stored database. In the case of a locally stored database, this can be stored on a mobile device and / or in a storage device located in the vehicle. Data about the object 8 can include, for example, the object's precise position, one or more photos and / or video sequences of this object, preferably including the distance to the object from which the photo or video was taken.

[0046] Step 17 checks whether a driver assistance function was activated following the initial classification of the detected object. If not, the process ends or returns to the previous step. If a driver assistance function was activated, and the object was classified as incorrectly identified, step 18 adjusts the already activated function, for example, by deactivating it.

[0047] The Fig. Figure 4 schematically shows a vehicle, for example, a motor vehicle. The vehicle 1 comprises a driver assistance device 20 and a sensor 10 for detecting the vehicle's surroundings, for example, a camera, preferably a front camera. The driver assistance device comprises the at least one sensor 10 and an evaluation unit 21. The evaluation unit 21 is designed to recognize objects from signals detected by the at least one sensor 10. It can, for example, include corresponding object recognition software and / or software with machine learning algorithms, in particular using artificial neural networks (deep learning). The evaluation unit 21 is further designed to classify detected objects. The driver assistance device 20 is designed to perform a method according to the invention, for example, a method based on the Fig. 3. To carry out the procedure explained.

[0048] The Fig. Figure 5 schematically shows a flowchart of another embodiment of the method according to the invention. Block 31 describes the evaluation routine, for example, performed offline within a single motor vehicle. In step 32, a front camera of the vehicle detects an object at a great distance, in particular a distance above a defined minimum distance. The object is incorrectly classified during the first classification.

[0049] In step 33, the incorrect classification of the object activates a driver assistance function of at least one of the vehicle's driver assistance systems. Subsequently, in step 34, the same object is detected again using the vehicle's front camera, but from a distance that is less than a defined maximum distance. The object is classified a second time, and optionally, its position (whether moving or stationary) is determined.

[0050] If the second classification differs from the first, i.e., if it is a false positive detection, then, as indicated by an arrow, 35 data points are uploaded to Cloud 2. The uploaded information regarding the false positive detection preferably includes the exact position of the falsely positively classified object, a number of features, preferably detailed features, optionally an assessment of whether the object is stationary or moving, and preferably at least one photograph or optionally a short video clip.

[0051] Data can be uploaded in several ways, for example: Online data transfer can take place in regions with good network coverage. Data packages of a predefined size can be uploaded as soon as they are complete. Data for a specific route can be uploaded in its entirety after the trip is finished.

[0052] Within Cloud 2, step 36 allows for verification of how many vehicles have passed the same location and uploaded the same false positive classification. This enables the evaluation of all uploaded data within an algorithm. In this context, step 37 optionally includes a manual review by a human, for example, based on the submitted photos or video clips. This manual review can be performed online or offline. The results of the review in step 37 can be transmitted as data or signals to the algorithm in step 36. Fig. 5 is a data transmission generally indicated by arrows.

[0053] The results of the evaluation from step 36 are confirmed in step 38, and confirmed false positives are stored in a database along with their exact position and preferably detailed characteristics. Users can download the data from the database, which is preferably stored in the cloud, for example, via mobile devices or devices installed in a vehicle. This is indicated by arrow 39.

[0054] Block 40 describes an improved detection, recognition, and classification of objects within a motor vehicle, achieved through the use of Cloud 2 according to the invention. Following step 32 of Block 31, which corresponds to step 32 of Block 39, an unwanted activation of a driver assistance function (see step 33) can be prevented. For this purpose, in step 41, the signals from the front camera acquired in step 32 are compared with the data on potentially false-positive object detections from the Cloud 2 database. Specifically, the features and the exact position of an object detected by the camera are compared with information stored in the database.If the characteristics and position of the detected object match those of an object stored in the database that has been classified as incorrectly detected, the object detection is ignored in step 42, thus avoiding the activation of a driver assistance function.

[0055] Downloading data from Cloud 39 can be done in various ways: Online data transfer can take place in regions with good network coverage. Regular synchronization can be performed. The download can be customized to a user profile. Furthermore, the download can be incremental, for example, continuing after a defined distance has been covered. Reference symbol list 1 vehicle 2 Cloud 3 Data transmission 4 Data processing 5 lanes 6. Captured image from a great distance 7. Captured image from a short distance 8 detected objects 9 Windscreen 10 Sensor / Camera 11. Capturing the vehicle's surroundings 12. Initial detection and initial classification of at least one specific object relevant for activating the driver assistance function at a distance from the vehicle that is greater than a specified initial threshold. 13. Second detection and second classification of the already detected and classified object at a distance from the vehicle that is less than a specified second threshold. 14 Comparison of the first classification with the second classification 15. Classifying the object as a falsely identified object 16. Send at least one signal with a number of data points about the object to a database. 17. Driver assistance function activated? Adjust 18 already activated driver assistance functions 19 End 20 Driver assistance devices 21 Evaluation facility 31 offline evaluation routines performed within a single motor vehicle 32 objects detected and classified at a great distance 33 Driver assistance function activated 34 Re-detection of the same object at a short distance and classification of the object 35 Data transmission 36 Data analysis 37 manual checks 38 Validation of the data and storage in the database 39 Data transmission 40 improved object detection, recognition, and classification 41 Comparison of the detected object with data from the database 42 Ignore object detection

Claims

Method for operating a driver assistance function of a vehicle (1), comprising at least one sensor (10) for detecting the vehicle environment and an evaluation unit (21) for recognizing objects (8) from signals detected by the at least one sensor (10), characterized in that the method comprises the following steps: - detection of the vehicle environment (11) by means of the at least one sensor (10), - first detection and first classification of at least one specific object relevant for activating the driver assistance function by means of the evaluation unit at a distance from the vehicle that is greater than a defined first threshold (12), - second detection and second classification of the at least one specific object by means of the evaluation unit at a distance from the vehicle that is less than a defined second threshold (13),- if the first classification of the object differs from the second classification (14), classify the object as a falsely identified object (15),- send a signal with a number of data points about the object to a database (16, 35), and- if a driver assistance function was activated as a result of the first classification and if the object was classified as a falsely identified object, adjust the driver assistance function (18)., Method according to claim 1, characterized in that the vehicle environment is detected stepwise or continuously over time by means of the at least one sensor (10). Method according to claim 1 or 2, characterized in that the at least one sensor (10) is designed to optically and / or acoustically detect the vehicle environment in the direction of travel. Method according to one of claims 1 to 3, characterized in that the at least one object (8) is a traffic sign and / or a marking and / or a road user. Method according to one of claims 1 to 4, characterized in that the position of the object and / or at least one feature of the object are sent to a database (16, 35) as data for the object (8). Method according to one of claims 1 to 5, characterized in that after the first classification, data (12) relating to the detected object are compared with data from a database (41) and if the detected object corresponds to an object classified as incorrectly detected in the database, the driver assistance function (42) is adapted before the second detection and second classification of the object (13). Method according to one of claims 1 to 6, characterized in that the signals sent to the database from a plurality of users are evaluated with regard to their reliability in classifying certain objects as incorrectly identified objects (36). Assistance device (20) for a vehicle (1) comprising at least one sensor (10) for detecting features of the vehicle environment and an evaluation device (21) for recognizing objects (8) from signals detected by the at least one sensor (10), which is designed to perform a method according to one of claims 1 to 7. Vehicle (1) comprising an assistance device (20) according to claim 8 and / or designed to perform a method according to any one of claims 1 to 7. A computer-implemented method comprising instructions which, when the program is executed by a computer, cause it to execute a method according to any one of claims 1 to 7. Computer program product comprising instructions which, when the program is executed by a computer, cause it to execute a method according to any one of claims 1 to 7. Computer-readable data carrier on which the computer program product according to claim 11 is stored. Data carrier signal that transmits the computer program product according to claim 11.

Images