EP4211019A1

EP4211019A1 - Teaching trajectories in a motor vehicle

Info

Publication number: EP4211019A1
Application number: EP21758395.4A
Authority: EP
Inventors: Jana Seeland; Christian Hopp; Eugen Dick
Original assignee: Volkswagen AG
Current assignee: Volkswagen AG
Priority date: 2020-09-08
Filing date: 2021-08-19
Publication date: 2023-07-19
Also published as: CN116457265A; US20230331249A1; DE102020211260A1; WO2022053284A1

Abstract

The invention relates to a method for operating a motor vehicle (10), said method comprising: - operating the motor vehicle (10) in a teaching mode in which a driver can travel along and thereby teach a trajectory (T); - operating the motor vehicle (10) in a following mode in which the motor vehicle (10) at least semi-autonomously follows the trajectory (T) or an adapted variant (TA) thereof; wherein, in the teaching mode, a maximum adjustable steering angle can be limited and this limitation of the steering angle can be variably specified depending on a driving situation. The invention also relates to a control unit (12) for carrying out such a method.

Description

description

Teaching in trajectories in a motor vehicle

The invention relates to a control unit and a method for operating a motor vehicle. The motor vehicle can in particular be a passenger car or a truck. In general, the invention is aimed at the field of at least partially autonomous support for drivers of a motor vehicle when carrying out desired driving maneuvers.

A solution is known from DE 10 2014 220 114 A1, in which a driver can teach in a desired trajectory (that is, a target movement path along which the vehicle is to move). For this purpose, he can follow the corresponding trajectory in a teach-in mode and thus specify it. The motor vehicle is then to travel this trajectory autonomously. The background is that the driver can have parking processes that are to be carried out frequently carried out autonomously by the motor vehicle.

DE 10 2014 220 144 A1 already teaches that it is important for autonomous driving or following the trajectory that a comparatively large steering angle is then available to the vehicle. This is required, for example, to compensate for control differences. However, if the driver has already set a correspondingly large and, in particular, maximum steering angle during learning, the scope for compensating for the control differences or for setting sufficiently large additional steering angles when following the trajectory autonomously is correspondingly reduced.

DE 10 2014 220 144 A1 therefore proposes limiting the adjustable steering angle during the learning process, for example by means of a counter-torque generated in the steering system in good time.

It has been shown that such a solution can lead to a loss of comfort from the driver's point of view and, more precisely, to a steering behavior that is perceived as unnatural during the learning process.

One object of the present invention is therefore to improve the learning of trajectories from the driver's point of view. This object is solved by the subject matter of the appended independent claims.

Advantageous developments are specified in the dependent claims.

It has been recognized that the driver can perceive the steering behavior of the motor vehicle as unnatural due to the previous fixed limitation of the adjustable steering angle in the learning mode. In other words, because of the artificially limited steering angle, he can perceive the steering behavior in the learning mode as an unnatural deviation from the steering behavior of the motor vehicle in normal driving mode, in which corresponding limitations typically do not exist.

In contrast to the prior art, the invention therefore proposes checking the actual requirement for a steering angle limitation in the learning mode (also referred to below as learning mode) and/or dynamically adapting a steering angle limitation to a current driving situation. In other words, the steering angle limitation should preferably only be provided when this is actually necessary. Similar to the prior art, a corresponding requirement can result from the fact that when following the learned trajectory compared to the learning mode, additional steering angle reserves should be available, e.g. to be able to compensate for control differences when following the trajectory. However, embodiments of the invention provide for investigating whether a taught-in trajectory can be subsequently adapted and/or changed with knowledge of a current driving situation in such a way that the at least partially autonomous following of the trajectory is then possible even without corresponding steering angle reserves or with at least reduced steering angle reserves. It is then not absolutely necessary to limit the steering angle directly in the learning mode, or such a limit can be reduced.

In particular, a method for operating a motor vehicle is proposed, with: operating the motor vehicle in a learning mode (or also learning mode) in which a driver can drive along a trajectory and thereby learn;

Operating the motor vehicle in a follow-up mode in which the motor vehicle follows the trajectory or a variant thereof which has been adapted (eg subsequently by smoothing), at least partially autonomously; a maximum steering angle that can be set can be limited in the learning mode, and this steering angle limitation can be variably defined as a function of a driving situation. All of the approaches that are common in the prior art can be used to teach the trajectory, but also to follow it. For example, a starting position and/or a target position can be stored for the purpose of teaching. Furthermore, speed profiles, steering angles and/or time-dependent location information for describing the trajectory can be stored, in particular in a section between the starting position and the target position.

In order to follow or, in other words, follow this trajectory, the motor vehicle can adjust the corresponding steering angles and/or speeds, at least partially autonomously and depending on the location. For this purpose, a control device can access or control corresponding actuators in the vehicle by evaluating the information learned and in particular stored in the learning mode.

A steering angle can be understood to mean any angular variable that describes a steering behavior of the motor vehicle. For example, it can be a steering wheel angle or a wheel steering angle measured on the vehicle wheels. A rack position or a general operating variable of a steering system, from which conclusions can be drawn about the corresponding steering behavior of the motor vehicle (hereinafter also simply referred to as vehicle), can be used as the steering angle.

Approaches that are common in the prior art can be used to limit the steering angle. For example, an electromotive counter-torque can be applied, which acts against a steering movement specified by the driver (e.g. acts on a toothed rack in an electromechanical steering system). In the case of a steer-by-wire system, a reaction force actuator can generate a corresponding counter-torque and/or an actuator coupled to the rack cannot convert a steering wheel angle into a correspondingly effective steering movement on the vehicle wheels.

Correspondingly, in order to limit the steering angle, at least one control device of the motor vehicle can be activated and/or output control signals in order to bring about the limitation.

This control device can also detect a driving situation according to any of the approaches described below. Depending on the driving situation detected, the steering angle limitation can be changed and thus dynamically adapted to the current driving situation. The steering angle limitation can have at least two states, namely an unlimited state and a limited state. In the unlimited state, the maximum adjustable steering angle can correspond to a technically and/or mechanically and/or system-related maximum possible steering angle. In the limited state, the steering angle can assume a value below the maximum possible steering angle mentioned. In other words, a limitation of the maximum possible steering angle can be selectively activated or deactivated, which corresponds to a variable definition of the steering angle limitation. In particular, it is possible to switch between these states, depending on which driving situation is currently present.

In general, if there is essentially unlimited maneuverability of the vehicle as a driving situation, for example because there are no obstacles in the vehicle surroundings, a steering angle limitation can be dispensed with and/or this can generally be small. In this case, it can be assumed that the trajectory for tracking can be adapted in such a way that it can be traversed with a steering angle below the maximum possible steering angle. The remaining difference between the steering angle required to follow the adapted trajectory and the maximum possible steering angle then remains as a still available scope for varying the steering angle in order to compensate for any control differences that may occur.

If, on the other hand, the maneuverability of the vehicle is impaired, for example due to obstacles in the vehicle environment and/or the presence of so-called bottlenecks, the steering angle limitation can be activated. The maximum adjustable steering angle is then preferably below a maximum possible steering angle. This takes into account the fact that the trajectory can subsequently only be adjusted to a limited extent or not at all due to the small remaining maneuvering room. Consequently, it is also not possible to ensure that the steering angles required for following the trajectory can be reduced by adapting the trajectory and that the described steering angle margin for compensating for any control differences remains.

The solution presented here is generally advantageous in that the steering angle is limited in a more needs-based manner. In particular, it can only be activated when it is actually necessary to ensure that a taught-in trajectory is followed with sufficient precision. This improves this from the driver's point of view Steering behavior of the vehicle in learning mode, as this is perceived as less restricted.

A further development provides that the driving situation is recorded (and/or evaluated) using at least one of the following:

- An environment sensor system of the motor vehicle, for example comprising at least one radar sensor, a lidar sensor, a distance sensor, an ultrasonic sensor or a camera;

- A communication system of the motor vehicle (in particular a so-called Car2X communication system), which is set up to communicate with at least one vehicle-external unit, for example with another vehicle, with a mobile radio subscriber, with a server device or with an Internet service;

Map information (e.g. stored in a vehicle navigation system or retrievable from an external server) that describes a vehicle environment;

- A computer-implemented model, in particular a machine learning model, which is set up to determine a definition of the steering angle limit;

- A computer-implemented method for evaluating and/or classifying the driving situation.

Any of the above variants can be used to assess the free maneuverability of the motor vehicle. In other words, the driving situation can be decisively defined and/or evaluated by the extent to which the vehicle can maneuver freely, ie to what extent there are obstacles in the vicinity of the vehicle that restrict (free) maneuverability.

Obstacles in the area surrounding the vehicle can generally be detected and/or free spaces can be detected by means of the environment sensor system. In general, the surroundings sensor system can map the surroundings of the vehicle in a manner known per se.

With the communication system, the motor vehicle can receive information that preferably defines a current maneuverability or generally degrees of freedom of movement of the motor vehicle. For example, it can from a control device of a car park information relating to an occupancy of parking spaces, the presence of bottlenecks within the car park or generally of lane boundaries within the receive parking garage. The same is also possible with regard to individual parking areas or parking spaces that are not multi-storey car parks.

In an analogous manner, the map information can relate to information that defines the degrees of freedom of movement of the motor vehicle, e.g. due to a traffic infrastructure and/or general traffic routes at the location of the motor vehicle. The vehicle environment can be described using the map information in particular as to whether there are turning possibilities, lanes fork, lane narrowings or generally what space conditions exist along a lane, e.g. due to structural restrictions.

The model and in particular the machine learning model can be implemented in a control device of the motor vehicle, the control device being a computer. However, it can also be implemented in a vehicle-external computer device, but then preferably transmit determination results relating to the steering angle limitations to a control device of the motor vehicle (for example via mobile radio). The model and in particular the machine learning model can receive information relating to the driving situation of the vehicle. In particular, the model and in particular the machine learning model can receive any of the information described herein, by means of which the vehicle environment and/or the driving situation can be described. In particular, the model and in particular the machine learning model can receive information from the environment sensors mentioned, the communication system and/or map information. The model, and in particular the machine learning model, can use such information as an input. It can output a definition of the steering angle limit as an output variable.

In particular, it can output whether the steering angle limitation is to be activated or not (e.g. whether the limited or unlimited state described above should be activated). As an alternative or in addition, it can output an extent of the steering angle limitation, for example what value the maximum adjustable steering angle should specifically assume.

Instead of a computer-based model, general computer-implemented methods and/or methods for detecting and in particular evaluating and/or classifying the driving situation can also be used. Steering angle limitations can generally be determined using a characteristic curve, map or table values.

The optional machine learning model can be trained and/or created on the basis of training data sets and as part of a machine learning process. These Training datasets can describe driving situations as well as existing steering angle limitations. For example, the driving situations can be objective and preferably verified, in which steering angle limitations have been rated as appropriate or not. On the basis of such training data sets, the machine learning model can learn a connection between a driving situation or variables describing the driving situation and steering angle limitations that are perceived as objectively appropriate.

In general, the model and in particular the machine learning model can mathematically model a relationship between input variables and output variables. As part of the learning or training process, links between these variables and/or weightings of the variables and/or so-called intermediate layers can be defined, fixed or, in other words, learned. In a manner known per se, the machine learning model can link the input variables and output variables via different layers, the layers each having nodes that are linked via links to nodes of an adjacent layer. In particular, the machine learning model can be an artificial neural network or be based on one.

In general, at least one driving situation parameter that describes the current driving situation can be obtained. As part of the method, it can be provided to check whether the driving situation parameter satisfies a predetermined steering angle limitation criterion. If this is the case, the steering angle limitation can be activated and/or limited to a predetermined value (in particular a value below a maximum possible steering angle). If this is not the case, a corresponding limitation can be omitted. The driving situation parameter can be, for example, a minimum distance between the motor vehicle and the environment and/or obstacles in the environment.

Alternatively or additionally, the driving situation parameter can be or include a classifier stored in a map and/or generally location-dependent, for example. This can, for example, describe a degree of difficulty of a current location, a driving situation there and/or a locally required driving maneuver, for example at a bottleneck. Such parameters and in particular classifiers can be transmitted to a control unit executing the method and/or determined or read out by the latter. For this purpose, software interfaces to map information and/or a map server external to the vehicle can be used, eg map/Car2X. For example, the need for a steering angle limitation can be recognized or learned by the method and/or control unit, and the information or parameters required to determine the extent of the steering angle limitation are preferably obtained via the above software interface. This information can optionally be linked to the size of the vehicle (eg no steering angle restrictions or, in comparison, smaller ones may be necessary for small vehicles).

In particular, one development provides that the maximum steering angle that can be set is set to a reduced value or is kept at a predetermined value below a (system-related and/or technically possible) maximum possible steering angle if the driving situation involves maneuvering the motor vehicle without maintaining a minimum distance corresponds to interference contours in the area. An interference contour in the area can generally be understood to mean a collision-relevant structure and/or a collision-relevant obstacle or object in the area. This can in particular be structural structures such as walls or pillars. Alternatively or additionally, the interfering contour can be classified, i.e. the steering angle can be limited depending on the type or class of the interfering contour to which a minimum distance was fallen below. For example, solid obstacles such as walls or pillars could lead to a greater restriction of the steering angle than, for example, vehicles stopping on a roadway, which will probably no longer be standing there when driving off in follow-up mode in the future.

If the minimum distance to such a structure is not reached, the possibility of subsequently changing a taught-in trajectory is reduced. Such a change can lead to the vehicle running the risk of colliding with the interfering contour when following the adapted trajectory. Accordingly, the taught-in trajectory should be traversed as precisely as possible, which is why, analogous to the prior art and to compensate for subsequent control differences, limiting the steering angle is preferred.

Alternatively or additionally, it can be provided that the maximum steering angle that can be set is set to an increased value or kept at a predetermined value if the driving situation corresponds to maneuvering at a minimum distance from interfering contours in the area.

The increased and decreased values mentioned above relate to changes in the maximum adjustable steering angle in comparison to driving situations that existed before a driving situation currently being considered. In particular, initially without Steering angle limitation can be started and the maximum adjustable steering angle can then be reduced. However, it is also possible to switch from a previously limited, maximum adjustable steering angle to a correspondingly increased value with less or no limitation.

If the minimum distance to interfering contours in the area is maintained, the maximum steering angle that can be set can be increased to the (e.g. system-related and/or technical) maximum possible steering angle. This corresponds to a deactivation of the steering angle limitation since the maximum possible or available steering angle is then available. The maximum steering angle that can be set (i.e. steering angle that can be set or specified by the driver) then corresponds to the maximum possible steering angle that is system-related or technically possible.

Compliance with minimum distances to interference contours can in turn be based on the

Environment sensors are evaluated, but also based on map information if this is compared with a current position of the vehicle. However, the presence and location of interfering contours in the area can also be transmitted to the motor vehicle via the communication system of the motor vehicle, if present, whereupon the minimum distance from these interfering contours can then be checked.

As described, it can be provided that a taught-in trajectory is adapted at least in sections before following it. In particular, this can take place when no steering angle limitation has taken place in the corresponding section in the learning mode and/or an increased value of the steering angle has been specified. For the reasons described above, to ensure the possibility of compensating for control differences, the trajectory can then be adapted in such a way that it can be followed with reduced steering angles. In other words, the trajectory can be smoothed.

One possibility for this is that at least one curve radius (or, in other words, radius of curvature) of the trajectory is increased. Additionally or alternatively, a deflection point (in particular in front of a corresponding curve or curvature) can be moved forward along the trajectory, for example, or a deflection point (e.g. from a curve and/or curvature) can be moved backwards along the trajectory. The directional designations front and rear refer to a direction of travel from a starting position of the trajectory to a target position (with positions further forward being shifted in the direction of the target position). The point of inflection and the point of deflection may differ as a result distinguish that a steering angle is changed there beyond a predetermined threshold value. A criterion that can preferably also be defined is that this change or the steering angle set hereby is maintained over a certain driving distance and/or at least not reduced beyond a predetermined extent. For example, the turning point can be characterized in that a larger steering angle is set compared to driving straight ahead or driving with a small steering angle. The deflection point can be characterized in that the steering angle is reduced compared to driving with a larger steering angle or, in general, cornering, and in particular driving with a smaller cornering radius follows.

The corresponding shifting of inflection points or deflection points represents a simple possibility to smooth the trajectory. In particular, this reduces the risk that the driver perceives the adaptation of the trajectory as unnatural or as an inappropriate deviation from the trajectory actually learned.

In general, it can be provided in this connection that the trajectory is adapted as a function of environmental information recorded in the learning mode. In particular, the surroundings can be mapped in the learning mode, for example by detecting the surroundings using the surroundings sensors. In this way, the adjustment can take place taking into account relevant and/or current environmental conditions. The environmental information can, for example, be taken into account in such a way that it is checked whether displacements of the steering point and/or the deflection point or, in general, a changed curve radius leads to a risk of collision with the vehicle environment. As an alternative or in addition to the learning mode, corresponding environmental information can also be obtained by any other variants described herein. For example, these can be obtained via a communication system of the vehicle or using map information.

In summary, maximum permissible adjustments to the trajectory can be defined using the environmental information. Additionally or alternatively, planned adjustments to the trajectory can be checked on the basis of the environmental information with regard to collision risks resulting from the adjustments. If there are no collision risks, the adjustment can be assessed as permissible. If, on the other hand, a risk of collision is determined, the adjustment can be evaluated as invalid and not implemented. The invention also relates to a control unit for a motor vehicle that is set up to carry out a method according to one of the preceding claims. The control unit can be set up to receive information relating to the driving situation from all of the units described herein. For example, it can be connected to the surroundings sensors of the motor vehicle, to any communication system, or to a unit that provides map information. Likewise, the control device can execute the described machine learning model or be connected to a device that executes this machine learning model. Furthermore, the control device can be set up to output parameters and in particular control signals relating to the steering angle limitation. For example, it can control an actuator of the steering system in order to define and/or implement a steering angle limitation. In particular, it can control an actuator for generating the counter-torques described herein, in order thereby to enable the steering angle to be limited.

In general, the control device can include a processor device and/or a memory device. Program instructions can be stored on the memory device which, when executed by the processor device, cause the control unit to carry out a method according to any aspect described herein.

Exemplary embodiments of the invention are explained below with reference to the accompanying schematic figures:

1 shows a vehicle in a schematic plan view, which is operated in the learning mode and has a control device according to an exemplary embodiment of the invention.

FIG. 2 is a schematic representation of the subsequent adaptation of the trajectory learned according to FIG. 1 .

2A is a schematic representation of the post-fitting of an alternative trajectory.

FIG. 3 is a flowchart of a method performed by the controller of FIG. 1 according to the present invention. 1 shows a vehicle 10 (a passenger car) in a schematically greatly simplified plan view. The vehicle 10 is located in an environment 100 which is delimited by structural structures 102 on the side from the vehicle's point of view. These structural structures 102 form interfering contours in the vehicle environment. The walls of a multi-storey car park are an example. In the starting position S shown, the vehicle 10 is at a comparatively large distance A from these interfering contours. A single distance A is only shown as an example, but it could also be defined as a distance circle around vehicle 10 with radius A, or it can assume any other contour that defines distance A between vehicle 10 and the points relevant to the driving situation structural structure represents.

The vehicle 10 includes a control unit 12. It also includes an environment sensor system 14 which is connected to the control unit 12 in a data-transmitting manner. In the example shown (only as an example), this has two environment sensors 16, which are distance sensors, for example. A large number of corresponding environment sensors 16 are preferably provided in order to detect distances to the environment on all sides of the vehicle and thus to reliably map the vehicle environment with regard to interference contours. In a manner known per se, the information (environmental information) recorded by the surroundings sensor system 14 can be used, for example, to determine the distance ratios to the interference contours 102, as entered by the distance A in FIG.

Control unit 12 is preferably also connected to a vehicle-external server 104 via a data connection D indicated by dashed lines. It can use this to obtain map information from which environmental information can be derived. In this way, for example, a position, extension and/or classification of interfering contours in the surrounding area can be inferred, for example a position of walls 102. Knowing the vehicle's own position, distance information from the surrounding area or to local interference contours can be determined.

Alternatively or additionally, the server 104 may include or execute a machine learning model. To do this, it can receive input variables from control unit 12, for example a current vehicle position, or any environmental information recorded by the surroundings sensors. Based on this, the necessity and/or the extent of a steering angle limitation can be determined by the machine learning model knowing these input variables and can be output to control unit 12 . It is not shown separately that control unit 12 is connected to an actuator for limiting the steering angle. This can be any actuator mentioned in the general part of the description, with which a counter-torque can preferably be generated against a manual torque exerted by the driver on the steering wheel (not shown).

In the state shown, the driver has activated a learning mode, for example via a corresponding input into a vehicle control system. In a manner known per se, a traveled trajectory T, which is indicated by dashed lines in FIG get saved. In the example shown, the vehicle 10 first enters an area B1. The control unit 12 uses any of the information described above and in particular environmental information to determine whether the vehicle 10 is sufficiently manoeuvrable. In particular, it is determined whether the distance A to the interfering contours in the area (i.e. to the walls 102) is above a predetermined minimum value. This is the case in area B1. The control unit 12 therefore determines that no steering angle limitation is required. The maximum possible steering angle that can be set by the driver therefore corresponds to a system-related maximum possible or available steering angle due to the lack of a limitation. In other words, the driver can deflect the vehicle 10 in the area B1 to a maximum and without limitation while learning the trajectory T.

In a region B2 along the trajectory T, the distance from the surroundings or from the walls 102 to the vehicle 10 decreases significantly. This becomes clear from the course of the traversed trajectory T in the area B2 and the significantly reduced distance to the walls 102 compared to the area B1.

This is determined by control unit 12, whereupon a steering angle limitation is activated. The driver can then no longer set the system-related maximum possible/available steering angle as the (manually) maximum adjustable steering angle. Instead, only a reduced maximum possible steering angle is available to him. In the example shown, this continues until the target position Z is reached.

After the learning mode is completed, the trajectory T is preferably adjusted by the control unit 12 or optionally by the external server 104 . This process is shown schematically simplified in FIG. First the original trajectory T is shown, as it was learned by the driver. This is crossed out. Also shown is the course of an adapted trajectory TA, which is shown with a broken line. The representation of FIG. 2 serves to explain the different courses of the trajectory T, TA. The arrangement of these trajectories TA, T shown above or below one another is of no particular importance. In particular, it is provided that these trajectories T, TA overlap as much as possible and are not consistently at a distance from one another, as is only shown for reasons of illustration.

Again, the regions B1 and B2 are shown in which there was no steering angle limitation (B1) or in which a steering angle limitation was activated (B2). In the area B2 with a steering angle limitation, there is preferably no adaptation of the trajectory. It can be assumed here that the vehicle still has a sufficient steering angle reserve available when following the trajectory autonomously, with this reserve corresponding at least to the difference between the limited adjustable steering angle in area B2 and the system-related/technically maximum available steering angle. In addition, the trajectory is not adjusted for safety reasons, so as not to create a risk of collision with the environment.

In contrast, there is no steering angle limitation in area B1. As explained in the general part, this has the advantage that the driver perceives the steering behavior of the vehicle 10 as natural and not restricted. However, this means that in autonomous ferry operation there may not be a steering angle reserve to compensate for any control differences that may occur. This applies in particular to sections of the trajectory in which the driver has actually set the maximum possible steering angle, that is, for example, the steering has been turned to the maximum. In the present case, there is such a full lock at the turn-in point marked with E. The turn-in point E is characterized by a change in the steering angle above a permissible threshold value, e.g. a change of more than 60%. Then, viewed in the direction of travel (i.e. viewed from the starting point S to the destination Z), cornering with the curve radius R along the trajectory T to the deflection point A takes place. From the deflection point A, there is an opposite change in the steering angle, which in turn is preferably above a certain threshold value, so for example again is more than 60%.

The trajectory T is now adapted in such a way that the steering angles required in the area B1 for following the adapted trajectory TA are limited as far as possible. In particular, these are limited in such a way that they are preferably below a maximum possible Steering angle are, so that, for example, no full lock is performed. In this way, it can be ensured that a certain steering angle reserve still remains in order to adjust control differences when driving off the adjusted trajectory TA autonomously.

In the present example, this is achieved by increasing the curve radius R in the curve section between the turning point E and deflection point A (to the adjusted radius RA). In particular, this takes place in that the turning point E is shifted further to the rear (ie closer to the starting point S) along the adapted trajectory TA. This results in the adjusted turning point AE of the adjusted trajectory TA. In addition or as an alternative, the deflection point A can be shifted further forward along the adapted trajectory TA (ie in the direction of the target location Z). This results in the adapted deflection point AA shown.

Despite the absence of a steering angle limitation in area B1, which can be advantageous from the driver's point of view, it is thus ensured that the taught-in trajectory T can be followed with sufficient precision in the form of the adapted trajectory TA.

2A shows a further alternative example of a trajectory adaptation in the case of a traversed S-shaped trajectory or S-shaped curve. The trajectory T and its dashed fitted curve TA are shown superimposed and the trajectory is fitted along its full length. In this case, the adjusted inflection and deflection points EE, AE shift not only along the original trajectory T, but also in directions that are not parallel to it. Correspondingly, it may also be necessary to add connection sections VA to the adapted trajectory TA, which connect the starting point and destination point S, Z with the adapted turning and deflection points AE, EE.

FIG. 3 shows a flow chart of an exemplary method that is executed by the control unit 12 from FIG. 1 .

The learning mode is activated in a step S1 and the vehicle 10 begins to follow the trajectory T. In a step S2, which can also take place simultaneously, environmental information is obtained (for example by means of environmental sensors 14) and/or recorded while the trajectory T is being followed. In a step S3, which is preferably carried out continuously while the trajectory T is being driven along, it is determined on the basis of the environmental information whether a steering angle limitation is necessary. For this purpose, the distance A to the surroundings or to the walls 102 is and/or is determined as a criterion other criteria are considered, such as the classification of an interference contour. If this is below a predetermined threshold value, the maneuverability of the vehicle 10 is restricted and a steering angle limitation is therefore activated. If this is not the case, no steering angle limitation is activated. The activation or non-activation of the steering angle limitation can be stored as additional information describing the trajectory T.

In a step S4, the trajectory T was traversed and is then analyzed. In the present case, this includes determining areas B1 and B2 with or without steering angle limitation. In the region B1 with a steering angle limitation, the adaptation of the trajectory T to the adapted trajectory TA then takes place, as explained with reference to FIG. 2 . In a step S5, the autonomous driving mode is activated and this adapted trajectory TA is driven. There can be a significant time difference of several hours or days between steps S1 and S5. In a manner known per se, the driver can thus teach in a parking maneuver that he believes to occur frequently (e.g. in relation to a reserved parking space in a multi-storey car park) in step S1 and follow it autonomously or independently in step S5 with the vehicle 10 (i.e. repeat) permit.

Reference List

10 vehicle

12 control unit

14 environment sensors

16 environment sensor

100 environment

102 wall

104 servers

A distance

B1 Area with steering angle limitation

B2 area without steering angle limitation

S starting point

Z target point

T trajectory

TA adjusted trajectory

E turn-in point

A deflection point

AE adjusted turn-in point

AA adjusted deflection point

D data line

R curve radius

RA adjusted curve radius

VA connection section

Claims

Claims Method for operating a motor vehicle (10), with:

Operating the motor vehicle (10) in a learning mode in which a driver can drive along a trajectory (T) and thereby learn;

Operating the motor vehicle (10) in a follow-up mode in which the motor vehicle

(10) follows the trajectory (T) or an adapted variant (TA) thereof at least partially autonomously; a maximum steering angle that can be set can be limited in the learning mode, and this steering angle limitation can be variably defined as a function of a driving situation. Method according to Claim 1, characterized in that the driving situation is detected using at least one of the following: an environment sensor system (16) of the motor vehicle (10); a communication system of the motor vehicle (10), which is set up to communicate with at least one vehicle-external unit (104);

map information describing a vehicle environment; a computer-implemented model, in particular a machine learning model, which is set up to determine a definition of the steering angle limit; a computer-implemented method for evaluating and/or classifying the driving situation. Method according to Claim 1 or 2, characterized in that the maximum adjustable steering angle is set to a reduced value or is kept at a predetermined value below a maximum possible steering angle if the driving situation involves maneuvering without maintaining a minimum distance from interfering contours (102). corresponds to the neighborhood (100). Method according to one of the preceding claims, characterized in that the maximum adjustable steering angle is set to an increased value or kept at a predetermined value when the Driving situation corresponds to maneuvering at a minimum distance from interfering contours (102) in the area (100). Method according to one of the preceding claims, characterized in that the taught-in trajectory (T) is adapted at least in sections prior to tracking in order to generate the adapted variant (TA). Method according to Claim 5, characterized in that the adaptation comprises increasing at least one curve radius (R) of the trajectory (T). Method according to Claim 5 or 6, characterized in that the adjustment comprises a forward displacement of at least one deflection point (E) and/or a rearward displacement of at least one deflection point (A). Method according to one of Claims 5 to 7, characterized in that the trajectory (T) is adapted taking into account environmental information recorded in the learning mode. Method according to Claim 8, characterized in that maximum permissible adaptations of the trajectory (T) are defined on the basis of the environmental information. Control unit (12) for a motor vehicle (10), which is set up to carry out a method according to one of the preceding claims.