WO2019016114A1

WO2019016114A1 - Method for operating a driver assistance system for a motor vehicle with renewed storage of training data, driver assistance system as well as motor vehicle

Info

Publication number: WO2019016114A1
Application number: PCT/EP2018/069198
Authority: WO
Inventors: Ciaran Hughes; Catherine Enright; Jonathan Horgan; Gustavo Pelaez; Olivia Donnellan; German Feijoo
Original assignee: Connaught Electronics Ltd.
Priority date: 2017-07-17
Filing date: 2018-07-16
Publication date: 2019-01-24
Also published as: DE102017115991A1

Abstract

The invention relates to a method for operating a driver assistance system (2) for a motor vehicle (1), wherein in a training phase of the driver assistance system (2), during which the motor vehicle (1) is maneuvered manually by a driver along a trajectory (11), the trajectory (11) is stored, and on the basis of at least one image, which is provided by means of a camera (4) of the motor vehicle (1), a plurality of object features (18), which describe objects in an environmental region (8) of the motor vehicle (1), is stored, and in an operating phase of the driver assistance system (2) the motor vehicle (1) is maneuvered at least semi-autonomously along the stored trajectory (11) on the basis of the stored trajectory (11) and the stored object features (18), wherein a plurality of object features (19) are detected in the operating mode, the detected object features (19) are assigned to the stored object features (18), and a decision is made on the basis of the assignment whether renewed storage of the object features (18) and/or the trajectory (11) is required.

Description

Method for operating a driver assistance system for a motor vehicle with renewed storage of training data, driver assistance system as well as motor vehicle

The present invention relates to a method for operating a driver assistance system for a motor vehicle, wherein in a training phase of the driver assistance system, during which the motor vehicle is maneuvered manually by a driver along a trajectory, the trajectory is stored, and on the basis of at least one image, which is provided by means of a camera of the motor vehicle, a plurality of object features, which describe objects in an

environmental region of the motor vehicle, is stored, and in an operating phase of the driver assistance system the motor vehicle is maneuvered at least semi-autonomously along the stored trajectory on the basis of the stored trajectory and the stored object features. In addition, the present invention relates to a driver assistance system for a motor vehicle. Finally, the present invention relates to a motor vehicle.

From the prior art, driver assistance systems are known by means of which the motor vehicle can be maneuvered along a previously recorded trajectory. To record the trajectory, the driver assistance system can first be operated in a training phase. In the training phase, the motor vehicle is manually maneuvered by a driver of the motor vehicle. The trajectory, which the motor vehicle travels during manual maneuvering, is recorded by the driver assistance system. Furthermore, it is provided that during manual maneuvering, objects in the environment of the motor vehicle are detected and object features which describe these objects are stored.

In a subsequent operating phase of the driver assistance system, the motor vehicle can then be maneuvered semi-autonomously or autonomously along the recorded trajectory. In the driver assistance systems known from the prior art, the motor vehicle is

maneuvered between a starting position and a target position along the trajectory. For example, the motor vehicle can be maneuvered along the stored trajectory if it is detected that the motor vehicle has reached the starting position of the trajectory. In the case of the at least semi-autonomous maneuvering of the motor vehicle, the previously stored object features can be used as landmarks. In this way, the driver of the motor vehicle can be supported, for example, in parking maneuvers in a home zone. For example, the driver can be assisted in the daily parking of the motor vehicle into a garage of his home or in a parking lot at his workplace. It may be the case that the environment in which the trajectory has been recorded changes depending on time. For example, the environment may vary depending on the time of the year. When the training phase has been carried out in the summer, leaves are present on the trees, which are recorded as objects. These leaves are no longer present in winter, which significantly changes the detected objects and their object features. In this case, a new training phase or a renewed recording of the trajectory and the object features is necessary. According to the prior art, it is provided that this renewed recording of the object features and the trajectory is predetermined by the driver by means of an operator input.

In this context, DE 10 2014 018 192 A1 describes a method for controlling a vehicle in which a target trajectory is determined from driving data of the vehicle, which is used as the basis for the vehicle during automatic travel. In a modified environment of the vehicle, a trajectory determined from current vehicle travel data is compared with a target trajectory, and a deviation between the new trajectory and the target trajectory is determined. A new target trajectory is set depending on the deviation.

It is an object of the present invention to provide a solution, how an operation of a driver assistance system in which a trajectory is recorded in a learning phase and an at least semi-autonomous maneuvering along the trajectory is performed in an operating phase can be carried out more efficiently and more user-friendly.

According to the invention this object is solved by a method, by a driver assistance system as well as by a motor vehicle having the features according to the respective independent claims. Advantageous embodiments of the invention are the subject matter of the dependent claims.

According to one embodiment of a method for operating a driver assistance system for a motor vehicle, in a training phase of the driver assistance system, during which the motor vehicle is preferably maneuvered manually by a driver along a trajectory, the trajectory is stored, and in particular on the basis of at least one image, which in particular is provided by means of a camera of the motor vehicle, a plurality of object features, which describe in particular objects in an environmental region of the motor vehicle, is stored. In an operating phase of the driver assistance system, the motor vehicle is maneuvered along the stored trajectory at least semi-autonomously, in particular on the basis of the stored trajectory and the stored object features. In this case, it is in particular provided that a plurality of object features is detected in the operating mode. In addition, the detected object features are assigned in particular to the stored object features. In particular, it is decided by means of the assignment whether renewed storage of the object features and/or the trajectory is required.

A method according to the invention serves to operate a driver assistance system for a motor vehicle. In a training phase of the driver assistance system, during which the motor vehicle is manually maneuvered by a driver along a trajectory, the trajectory is recorded and, on the basis of at least one image which is provided by means of a camera of the motor vehicle, a plurality of object features which describe objects in an environmental region of the motor vehicle, are stored. In an operating phase of the driver assistance system, the motor vehicle is maneuvered along the stored trajectory at least semi- autonomously on the basis of the stored trajectory and the stored object features. In this case, it is also provided that a plurality of object features are detected in the operating mode, the detected object features are assigned to the stored object features, and a decision is made on the basis of the assignment whether renewed storage of the object features and/or the trajectory is required.

By means of the method, the driver assistance system of the motor vehicle is first to be trained in the training phase or a learning phase. In the subsequent operating phase or the subsequent operating phases, the motor vehicle can then be maneuvered on the basis of the training using the driver assistance system. For example, the method can be used to drive the motor vehicle into a garage or a parking space of the driver. The method can also be used to drive the motor vehicle to a parking space at the driver's workplace. In this case, it is provided that, during the training phase, the driver specifies the trajectory which extends from a starting position to a target position. Along this trajectory, the motor vehicle can then be maneuvered at least semi-autonomously at a later stage in the operating phase. In the training phase, the motor vehicle is operated exclusively manually by the driver. In the training phase, the trajectory, which the driver specifies through manual maneuvering, is recorded using the driver assistance system. For this purpose, for example, the set steering angle and/or the revolutions of at least one wheel of the motor vehicle can be recorded continuously. In addition, the start position and the target position of the trajectory can be stored. It can also be provided that the start position, the target position and/or the trajectory are determined by means of a satellite-based position determination system.

In addition, at least one object in the environmental region of the motor vehicle is detected during the training phase. Preferably, several objects are detected in the environmental region. During manual maneuvering, the environmental region of the motor vehicle can be detected with the camera of the motor vehicle or the driver assistance system. In particular, a plurality of cameras can be used to detect the environmental region. With the at least one camera images describing the environmental region are provided. By means of a corresponding object detection algorithm, the object features which describe the objects in the environmental area can then be recognized in the images. In addition, further environmental sensors of the driver assistance system can also be used to detect the objects. Several object features can also be determined for an object. These object features and their position can then be stored in a digital environmental map, which describes the environmental region of the motor vehicle. The trajectory and/or points of the trajectory can also be stored in this environment map. Thus, there is information in the environmental map which describes the relative position of the object features relative to the stored trajectory. Thus, the stored object features can be used as landmarks for orientation in the subsequent operating phase of the driver assistance system. In the operating phase it can be provided that the motor vehicle is maneuvered semi- autonomously by means of the driver assistance system. In this case, the driver assistance system engages in the steering and the driver of the motor vehicle continues to actuate the accelerator pedal and the brake pedal. Preferably, the motor vehicle is maneuvered fully automatically along the trajectory. In this case, the driver assistance system also takes over the intervention in a drive motor and a braking system of the motor vehicle.

According to an essential aspect of the present invention, it is provided that a plurality of object features is detected in the operating phase. These detected object features are then assigned to the stored object features. This means that object features are also detected during the subsequent operating phase or the subsequent operating phase by means of the at least one image which is provided with the camera. These detected object features are then assigned to the object features previously stored in the training phase. Based on the assignment, the motor vehicle can then be maneuvered along the stored trajectory. In the present case, the assignment of the detected object features to the stored object features is used to verify whether a renewed storage of the object features and/or a renewed recording of the trajectory is required. On the basis of the assignment of the detected object features to the stored object features, it is possible to automatically recognize whether the environmental region or the region in which the object features have been detected has changed. In other words, by means of the assignment it is verified whether it is necessary to perform a new training phase. In particular, it is determined whether the stored object features and/or the trajectory, which have been determined as training data, must be determined again. Thus, it can be recognized automatically whether a renewed determination of the training data or a renewed execution of a training phase is required without requiring a manual input from the user or the driver. In this way, the driver assistance system can be automatically adjusted to changing environment conditions. These changes in the environment can be caused by changes in the objects due to the season, lighting and/or weather conditions. Thus, the driver assistance system can be operated more efficiently and more user- friendly overall.

Preferably, the object features and/or the trajectory are stored anew during a subsequent operating phase if it is decided that a renewed storage of the object features and/or the trajectory is required. If the assignment of the detected object features to the stored object features shows that a new recording of the training data is required, this can be carried out automatically in the subsequent operating phase. Thus, the driver assistance system can automatically adapt to the changing environmental conditions, whereby reliable operation can be guaranteed. This automatic renewed storage of the object features and/or the trajectory can thus be carried out continuously without the user or the driver perceiving this. In this way, transitions in ambient conditions can also be

considered. This applies, for example, to autumn, when trees have only shed part of their leaves. Thus, the driver assistance system can be operated more user-friendly.

In one embodiment, the assignment of the detected object features to the stored object features is carried out by means of a visual method of simultaneous localization and mapping. In other words, a so-called VSLAM algorithm (VSLAM - Visual Simultaneous Localization and Mapping) is used. In addition or as an alternative, algorithms related to VSLAM may be used, for example Structure From Motion and/or Bundle Adjustment. With the aid of the algorithm, the object features can be determined during the training phase and be entered into the environment map. In addition, points describing the trajectory can be entered into the environment map. In the operating phase, the algorithm is used to assign the detected object features to the stored object features determined in the training phase. Based on the assignment of the detected object features to the stored object features, the algorithm can determine the current position of the motor vehicle relative to the recorded trajectory. Thus, the assignment of the object features can be carried out in a reliable manner.

In a further embodiment, a visual descriptor describing the respective object feature in the image is determined for the stored object features and/or the detected object features, and the assignment is determined by means of the visual descriptor. One or more visual descriptors can be determined and stored for the stored object features. In the same way, visual descriptors can be determined for the detected object features in the operating phase. In the assignment of the detected object features to the stored object features, the visual descriptors of the detected object features can in particular be assigned to the visual descriptors of the stored object features. The respective visual descriptor is, in particular, an algorithmic representation of what the object feature "looks" like in the image. The respective visual descriptor may describe a color, shape, texture or the like of the features. This makes a reliable assignment of the detected object features to the stored object features possible.

Furthermore, it is advantageous if, on the basis of the assignment, a quality measure is determined which describes a matching between the detected object features and the stored object features, and it is decided on the basis of the quality measure whether the renewed storage of the object features and/or the trajectory is necessary. In particular, it is provided that the quality measure is determined by means of the matching of the visual descriptors of the detected object features with the visual descriptors of the stored features. This quality measure describes, in particular, how good the matching of the detected feature with the stored features is. On the basis of this quality measure, it can then be decided whether a renewed storage of the training data is necessary. For this purpose, the quality measure can, for example, be compared with a predetermined limit value.

In a further embodiment, a number of the detected object features, which are assigned to the stored object features, is determined, and the number is used to decide whether the object features and/or the trajectory need to be stored anew. A further criterion, which determines whether the renewed storing of the training data is required, is the number of valid matches in the assignment of the detected object features to the stored object features. If, in the operating phase, the object features are extracted from the image, a detected object feature cannot be assigned to each of the stored object features. If the environmental conditions change, the number will decrease. If, for example, this number falls below a predefined threshold value, it can be decided that the renewed determination of the object features and/or the trajectory is necessary.

In a further embodiment, one of the stored object features is deleted if no object feature is assigned to this object feature in a predetermined number of operating phases and/or for a predetermined time duration in the operating phases. It can be provided in particular that an information is assigned to each of the stored object features, which describes the time since this object feature was last detected in an operating phase. This time can be a physical time, for example, a date or time, or the number of operating phases carried out. If the time exceeds a predetermined limit, this stored object feature can be deleted. Here, for example, it can be assumed that this stored object feature describes an object which is no longer present, or which has been stored incorrectly.

In a further embodiment, a further object feature is stored if it is detected in a

predetermined number of operating phases. In each operating phase object features are recognized which cannot be assigned to any of the stored object features. If these object features are detected in several successive operating phases, it can be assumed that they originate from a new object or that they originate from a change in the environmental conditions. If these object features are detected in several operating phases or a predetermined number of operating phases, they can be stored. The selection of the newly stored object features can be determined on the basis of criteria which describe the quality of the object feature. Here, for example, the optical saliency and/or the noise in reconstruction can be taken into account. In addition, it can be verified whether the object feature can be assigned to further object features or whether it is an outlier. The number of newly stored features can be limited. For example, the number of newly stored object features can correspond to the previously deleted object features. Thus, the memory space can be limited and the computational effort can be reduced.

In a further embodiment, a recording information for the respective object features is stored in the training phase and/or in the operating phase, and the assignment is performed using the respective recording information. In the training phase, additional recording information can be determined, which is then assigned to the stored object features. This recording information can describe the recording of the image by means of the camera and/or the ambient conditions. In the same way, recording information for the detected object features can be determined during the operating phase. When comparing the detected object features with the stored object features, this recording information can then be taken into account. For example, it can be verified whether the recording information is similar or comparable. In this way, the robustness in the assignment can be improved.

In particular, it is provided that the recording information describes a lighting situation in the environmental region, a time of day and/or a season. It is thus possible, for example, to take account of the fact that different illumination situations are given during the storage phase of the object features during the training phase and during the detection of the object features during the operating phase. If, for example, object features have been stored under bright ambient conditions, it may be the case that these are not detected during dark operating conditions in the operating phase. If the stored object features were determined at a different time of the year than the object features are detected, differences may also occur. For example, it is possible to store object features which describe leaves in trees or plants, which are not present during an operating phase which takes place in winter. This enables a reliable allocation of the object features during the operating phase.

A driver assistance system for a motor vehicle according to the invention is adapted for performing a method according to the invention and the advantageous embodiments thereof. The driver assistance system can comprise the camera by means of which the objects can be detected in the environment of the motor vehicle. In addition, the driver assistance system can have a motion sensor with which the movement of the motor vehicle can be recorded in the training phase. In addition, the driver assistance system can comprise an electronic control device by means of which the trajectory can be recorded and the object features and/or the associated optical descriptors can be stored. In addition, a steering, a drive motor and/or a braking system can be controlled during the at least semi-autonomous maneuvering by means of the control device. Furthermore, the driver assistance system may have a satellite-based position determination system.

A motor vehicle according to the invention comprises a driver assistance system according to the invention. The motor vehicle is preferably designed as a passenger car.

The preferred embodiments presented with respect to the method according to the invention and the advantages thereof correspondingly apply to the driver assistance system according to the invention as well as to the motor vehicle according to the invention.

Further features of the invention are apparent from the claims, the figures and the description of figures. The features and feature combinations mentioned above in the description as well as the features and feature combinations mentioned below in the description of figures and/or shown in the figures alone are usable not only in the respectively specified combination, but also in other combinations or alone without departing from the scope of the invention. Thus, implementations are also to be considered as encompassed and disclosed by the invention, which are not explicitly shown in the figures and explained, but arise from and can be generated by separated feature combinations from the explained implementations. Implementations and feature combinations are also to be considered as disclosed, which thus do not have all of the features of an originally formulated independent claim. Moreover, implementations and feature combinations are to be considered as disclosed, in particular by the implementations set out above, which extend beyond or deviate from the feature combinations set out in the relations of the claims.

Now, the invention is explained in more detail based on preferred embodiments as well as with reference to the attached drawings.

These show in:

Fig. 1 a motor vehicle according to an embodiment of the invention, which

comprises a driver assistance system;

Fig. 2 the motor vehicle which is manually maneuvered along a trajectory,

wherein the driver assistance system is operated in a training phase;

Fig. 3 the motor vehicle which is maneuvered along a trajectory in an operating phase of the driver assistance system; and

Fig. 4 a digital environmental map describing the trajectory, object features stored in the training phase as well as object features detected in the operating phase.

In the figures, identical or functionally identical elements are provided with the same reference characters.

Fig. 1 shows a motor vehicle 1 according to an embodiment of the present invention in a plan view. In the present case, the motor vehicle 1 is designed as a passenger car. The motor vehicle 1 comprises a driver assistance system 2. The driver assistance system 2 comprises, in turn, an electronic control unit 3. In addition, the driver assistance system 2 comprises at least one camera 4. In the present exemplary embodiment, the driver assistance system 2 comprises four cameras 4 which are arranged distributed on the motor vehicle 1 . In the present case, one of the cameras 4 is arranged in a rear region 5 of the motor vehicle 1 , one of the cameras 4 is arranged in a front region 7 of the motor vehicle 1 , and the other two cameras 4 are arranged in a respective side region 6, in particular in a region of the side mirrors of the motor vehicle 1 . The number and arrangement of the cameras 4 is to be understood as purely exemplary.

With the cameras 4, objects 10 can be detected in an environmental region 8 of the motor vehicle 1 . For this purpose, images or image sequences can be provided with the cameras 4, which describe the objects 10 in the environmental region 8. These images can then be transmitted from the respective cameras 4 to the control unit 3. By means of a corresponding object identification algorithm, the control device 3 can then recognize the objects 10 in the environmental region 8.

In addition, the driver assistance system 2 comprises a motion sensor 9, by means of which a movement of the motor vehicle 1 can be recorded. For example, a steering angle and/or revolutions of at least one wheel of the motor vehicle 1 can be determined continuously with the movement sensor 9. The movement sensor 9 is also connected to the control device 3 for data transmission. In addition, the motion sensor 9 can comprise a receiver for a satellite-based position determination system. Thereby, the position of the motor vehicle 1 can be determined continuously.

Fig. 2 shows the motor vehicle 1 , wherein the driver assistance system 2 is operated in a training phase. In the training phase, the motor vehicle 1 is manually maneuvered along a trajectory 1 1 by a driver of the motor vehicle 1 . This trajectory 1 1 , which comprises a plurality of points 12, extends from a starting point 13 to a target point 14. In the present case, the target point 14 is located within a garage 15. The motor vehicle 1 is

maneuvered past several objects 10 during maneuvering. In the present case, there are 8 objects 10 in the form of plants 16, a pedestrian 17 as well as the garage 15. During the maneuvering of the motor vehicle 1 in the training phase, object features 18 of the objects 10 are recognized and stored on the basis of the images which are provided with the cameras 4. In addition, the trajectory 1 1 or the points 12 of the trajectory 1 1 are stored. The trajectory 1 1 or its points 12 can be determined by means of the motion sensor 9.

Fig. 3 shows the motor vehicle 1 , the driver assistance system 2 being operated in an operating phase which follows the training phase in time. The trajectory 1 1 , which was determined in the preceding training phase, is shown schematically. As soon as it is detected that the motor vehicle 1 is located at the start position 13 of the stored trajectory 1 1 , the motor vehicle 1 can be maneuvered along the trajectory 1 1 to the target point 14 in the garage 15. This applies in the event that there is no obstacle on the trajectory 1 1 . During the operating phase, object features 19 are detected and assigned to the object features 18 stored in the training phase.

Fig. 4 shows a representation of an environmental map 20 which shows the trajectory 1 1 , the points 12 of the trajectory 1 1 , the stored object features 18, and the detected object features 19 which were assigned to the stored object features 18. The assigned object features 19 are assigned to the stored object features 18 by means of a visual algorithm for simultaneous localization and mapping. In the present case, it can be seen that the number of detected object features 19 is smaller than the number of object features 18 stored in the training phase. In the training phase, at least one visual descriptor can be determined for the stored object features 18 and be stored in the digital environmental map 20. In the subsequent operating phase, visual descriptors can also be determined for the detected object features 19. In the operating phase, the visual descriptors of the detected object features 19 can then be assigned to the visual descriptors of the stored object features 18. The motor vehicle 1 can thus be located on the basis of the known points 12 of the trajectory 1 1 .

Due to changing ambient conditions, differences between the detected object features 19 and the stored object features 18 can result. This may, for example, be due to a difference in the season, the time of day and/or the lighting conditions. If the ambient conditions have changed significantly, a renewed recording of the trajectory 1 1 and/or the stored object features 18 is necessary in order to be able to reliably operate the driver assistance system 2 in the operating phases. In the present case, it is intended to automatically recognize when a renewed recording of the stored object features 18 and/or the trajectory 1 1 is required.

In the operating phase, the detected object features 19 are assigned to the stored object features 18 by means of the algorithm. Based on this assignment, a quality measure can be determined which describes the quality of the assignment or the matching. This quality measure can then be used as a criterion as to whether a renewed recording of the stored object features 18 and/or the trajectory 1 1 is necessary. As a further criterion, the number of validly assigned detected object features 19 to the stored object features 18 can be used. When the object features 1 1 are extracted from the image, not all object features 19 can be assigned to the stored object features 18. If the environmental conditions change, the number of valid assignments will decrease significantly. As soon as the number of valid matches is below a threshold value and/or the quality measure falls below a predetermined limit value, it can be decided that a re-storage of the object features 18 and/or the trajectory 1 1 is required. It should be noted that the limit value or the threshold value can be determined in such a way that the operating phase can be performed reliably, but there are indications that the ambient conditions have changed. This renewed storage of the object features 18 and/or of the trajectory 1 1 is then carried out

automatically in the subsequent operating phase.

Furthermore, it is provided that the stored object features 18 and/or the visual descriptors of the stored object features 18 are dynamically adapted during operation of the driver assistance system 2. For example, stored object features 18, which are never recognized in the operating mode, can be deleted. In principle, an initial training phase is required in which the driver explicitly specifies the trajectory 1 1 along which the motor vehicle 1 is to be maneuvered. The object features 18 can also be assigned an information describing the time since this object feature 18 has been found for the last time in the operating phase. If this time exceeds a threshold value, this stored object feature 18 can be deleted from the digital environmental map 20. It may also be provided that during the operating phase new object features 19 are detected which cannot be assigned to any of the stored object features 18. Here, it can be assumed that this object feature 19 is assigned to a new object 10 in the environmental region 8. If this object feature 19 is recognized in a predetermined number of operating phases, this object feature 19 can also be stored or used as a stored object feature 18.

Furthermore, it may be provided that additional recording information is added to the object features 18. This recording information may, for example, describe the time of day, the season, the lighting state or the like. These recording information can then be taken into account when the detected object features 19 are assigned to the stored object features 18. Furthermore, these recording information can also be taken into account in the deletion of stored object features 18.

Thus, it is possible to decide whether the stored trajectory 1 1 and/or the stored object features 18 are to be recorded again without user input by the driver. The renewed storage of these training data can then be carried out in the subsequent operating phase. Thus, the training data can be updated continuously and be adapted to the gradually changing environmental conditions. It is not necessary for the driver to manually trigger the renewed recording of the training data. This overall enables a reliable operation of the driver assistance system 2.

Claims

1 . Method for operating a driver assistance system (2) for a motor vehicle (1 ), wherein in a training phase of the driver assistance system (2), during which the motor vehicle (1 ) is maneuvered manually by a driver along a trajectory (1 1 ), the trajectory (1 1 ) is stored, and on the basis of at least one image, which is provided by means of a camera (4) of the motor vehicle (1 ), a plurality of object features (18), which describe objects in an environmental region (8) of the motor vehicle (1 ), is stored, and in an operating phase of the driver assistance system (2) the motor vehicle (1 ) is maneuvered at least semi-autonomously along the stored trajectory (1 1 ) on the basis of the stored trajectory (1 1 ) and the stored object features (18),

characterized in that

a plurality of object features (19) are detected in the operating mode, the detected object features (19) are assigned to the stored object features (18), and a decision is made on the basis of the assignment whether renewed storage of the object features (18) and/or the trajectory (1 1 ) is required.

2. Method according to claim 1 ,

characterized in that

the object features (18) and/or the trajectory (1 1 ) are stored anew during a subsequent operating phase if it is decided that a renewed storage of the object features (18) and/or the trajectory (1 1 ) is required.

3. Method according to claim 1 or 2,

characterized in that

the assignment of the detected object features (19) to the stored object features (18) is carried out by means of a visual method of simultaneous localization and mapping, structure from motion and/or bundle adjustment.

4. Method according to any one of the preceding claims,

characterized in that a visual descriptor describing the respective object features (18, 19) in the image is determined for the stored object features (18) and/or the detected object features (19), and the assignment is determined by means of the visual descriptors.

5. Method according to any one of the preceding claims,

characterized in that

on the basis of the assignment, a quality measure is determined which describes a matching between the detected object features (19) and the stored object features (18), and on the basis of the quality measure it is decided whether the renewed storage of the object features (18) and/or the trajectory (1 1 ) is necessary.

6. Method according to any one of the preceding claims,

characterized in that

a number of the detected object features (19), which are assigned to the stored object features (18), is determined, and the number is used to decide whether the object features (18) and/or the trajectory (1 1 ) need to be stored anew.

7. Method according to any one of the preceding claims,

characterized in that

one of the stored object features (18) is deleted if no detected object feature (19) is assigned to this object feature (18) in a predetermined number of operating phases and/or for a predetermined time duration in the operating phases.

8. Method according to any one of the preceding claims,

characterized in that

a further object feature is stored if it is detected in a predetermined number of operating phases.

9. Method according to any one of the preceding claims,

characterized in that

a recording information for the respective object features (18, 19) is stored in the training phase and/or in the operating phase, and the assignment is performed using the respective recording information.

10. Method according to claim 9,

characterized in that that the recording information describes a lighting situation in the environmental region (8), a time of day and/or a season.

1 1 . Driver assistance system (2) of a motor vehicle (1 ), which is adapted for performing a method according to any one of the preceding claims.

12. Motor vehicle (1 ) comprising a driver assistance system (2) according to claim 1 1 .