WO2018087321A1 - Method and system for operating a motor vehicle - Google Patents

Abstract

The invention relates to a method for operating at least one motor vehicle (20), said method involving the steps of providing (S1) a data record (10) characterizing a vehicle environment, a driving behavior, and the movement of the motor vehicle (20); ascertaining (S2) at least two of the following types on the basis of the data record (10) provided: the type of road (S2a) on which the motor vehicle (20) travels; the type of maneuver (S2b) performed by the motor vehicle (20); the type of driver (S2c) steering the motor vehicle (20); forming (S3) a group (38) comprising at least two of the types ascertained; providing (S4) an initial adjustable artificial neural network (22), which is assigned to the group (38) and in which a specific driving behavior is predefined for the group (38), in a computing unit (18) of the motor vehicle (20); and comparing the driver's actual driving behavior with the driving behavior predefined in the adjustable artificial neural network (22) and adjusting the adjustable artificial neural network (22) by means of the computing unit (18) of the motor vehicle (20) according to the comparison. The invention further relates to a system for operating at least one motor vehicle (20), said system being configured to carry out the disclosed method.

Description

 Patent application

Method and system for operating a motor vehicle

DESCRIPTION:

 The present invention relates to a method and a system for operating a motor vehicle.

Modern motor vehicles achieve an ever higher degree of automation. More and more activities that a driver performs to control a motor vehicle can be at least partially taken over by means of suitable devices of modern motor vehicles, so that the driver is always relieved. Examples of such developments, which include, for example, a lane departure warning system or a distance control cruise control, are already known from the general state of the art, in particular from mass-produced vehicles.

In order to enable motor vehicles to take on more and more tasks from human drivers in the future, so that safe and compliant driving fully autonomous or driverless vehicles can be provided on further development, data on actions or activities that are performed by human drivers in controlling a motor vehicle are already stored today, analyzed and finally made usable in motor vehicles.

Approaches are already known from the prior art. Thus, document US 2016 0026182 A1 proposes how the driving behavior of individual drivers is learned by means of machine learning and can be used in autonomous vehicles. can be made so that driverless vehicles have a personalized driving style that corresponds to a driving style of a human occupant. For this it is shown that for each occupant at least one own driving profile is created, which is stored in a special server arrangement, and is downloaded into a vehicle in which the occupant is located. The driving profile can be adapted and updated in the server arrangement by means of machine learning during a manual drive in which the occupant assumes the role of the driver. Thus, the driving profile of a driver is gradually developed to be used in a vehicle capable of driving without a driver. This ultimately has the purpose that the occupant of the driverless vehicle feels particularly comfortable when using different vehicles while driving, as the driverless vehicle accelerates in a way the occupant own way, for example, brakes, drives in a curve, distance to vehicles in front stops etc.

The document US Pat. No. 6,879,969 B2 describes a system which recognizes driving patterns automatically and in real time by means of a single artificial neural network for statistical pattern recognition in a motor vehicle in order, for example, to recognize different driving environments (city, highway, etc.). Based on a recognized driving environment, a suitable controller makes changes to vehicle parameters, for example, to reduce fuel consumption. For example, a suspension setting can be changed as soon as the system recognizes that the vehicle is being driven on a motorway. Furthermore, the system can be used to evaluate the driving pattern, for example to be able to tailor driver warning systems, such as fatigue detection or distraction detection, particularly well to the needs of a current driver, or to recognize different driving styles.

A method for programming a computer based on an artificial neural network, which is used in cognitive safety systems of vehicles, is described in the document EP 2 884424 A1. In this case, an observation computer based on an artificial neural network is used in a large number of vehicles, which records data of the current environment of the vehicle. and data about driver reactions. Further, the observation computer is arranged to recognize visual patterns and correlate them with the recorded data so that a data set is generated. The data sets from the plurality of vehicles are combined and used as a basis for programming the computer used in vehicles in cognitive security systems. In general, such cognitive security systems include, for example, traffic sign recognition, pedestrian recognition or recognition of preceding traffic.

The document DE 10 2013 003 042 A1 concentrates on obtaining and using rule sets which are suitable for automatically activating a vehicle function as soon as certain conditions are met. These conditions result from an instantaneous situation of the vehicle, which is detected by means of a sensor system of the vehicle, such as a temperature sensor, rain sensor, etc. In a variety of vehicles, activation of a vehicle function, such as a driver's rear window heating, is related to a current situation of the vehicle, such as outside temperature less than 5 ° C and rain, and a corresponding record is generated. The data record is sent to a central server, where a set of rules is formed for each vehicle function to be automated, which is then transmitted to the motor vehicle to be automated. Thus, it is possible that in another vehicle, under similar, in particular equal conditions, the rear window heating associated with the other vehicle is activated, without the intervention of a driver or a learning phase of the other vehicle, since it has the set of rules.

Object of the present invention is to make the behavior of human drivers in motor vehicles with a particularly high level of detail usable.

This object is achieved by a method and by a system having the features of the independent claims. Advantageous embodiments with expedient developments of the invention are specified in the other claims. In the method according to the invention for operating at least one motor vehicle, a data record is initially provided which characterizes a vehicle environment, a driver behavior and the movement of the motor vehicle. This means that environmental data of the vehicle is stored in the data record. For example, the environmental data may include data about a development of the environment or the environment of the motor vehicle, whether there are other road users in the vicinity of the vehicle and what kind of these are (eg other motor vehicles, pedestrians, cyclists, etc.), whether markings are present on the roadway and what type these are (eg, guidelines, warning lines, side boundary lines, etc.), and / or what environmental conditions prevail (eg, brightness, temperature, precipitation, etc.).

Furthermore, driver behavior data of the driver can be stored in the data record. The driver behavior data includes data that characterizes, among other things, at least one driver action. For example, the driver behavior data may include data about an acceleration and / or speed with which the driver respectively moves the steering wheel (steering wheel acceleration or speed), the brake pedal, and / or the accelerator pedal (accelerator acceleration) ), how fast the driver changes between the individual pedals, whether the driver activates or deactivates a vehicle function (eg turn indicator, headlight etc.), which seat system setting the driver has selected and / or if the driver Gear) changes and, if necessary, which driving speed the driver chooses. The driver behavior data may also include other data, for example, whether a driver has fatigue symptoms or whether the driver is distracted from controlling the motor vehicle, eg. B. by operating a vehicle infotainment system or a mobile phone.

Furthermore, movement data of the motor vehicle can be stored in the data record. For example, the movement data may include data on speeds and / or accelerations, wherein one of the speeds and one of the accelerations are each assigned to one of three main axes of the vehicle, namely a longitudinal axis, a transverse axis or a vertical axis. As a result, in each case a translational speed along the longitudinal axis (longitudinal speed), along the transverse axis (transverse speed) and along the vertical axis Axis (vertical speed) and each one rotational speed about the longitudinal axis (rolling), to the transverse axis (pitch) and about the vertical axis (yaw) are stored in the movement data. Furthermore, in the movement data, data about respective accelerations associated with the speeds can be stored, for example, longitudinal, lateral, vertical, roll, pitch and / or yaw acceleration.

In addition, at least two of the following types are determined based on the provided data set: road type of a road traveled by the motor vehicle, driving maneuver type of a driving maneuver performed by the motor vehicle, driver type of a driver controlling the motor vehicle. In other words, the provided record is evaluated as to what type a road is, whereupon the motor vehicle is driven. For example, a first evaluation result a road type highway, a road type rural road, a street type city street, etc. result. A further subdivision of the types of road is conceivable, it may prove advantageous, for example, to distinguish whether a highway runs inside or outside a local border or if a special traffic zone, eg. B. a tempo-30 zone, a traffic-calmed area o. Ä., Is present.

The data set is also evaluated as to what type of driving maneuver is performed by the motor vehicle. For example, a second evaluation result can result in a type of overtaking maneuver, a type of driving maneuver subsequent driving, a driving maneuver type turning, a driving maneuver type reverse parking etc.

Further, the record is evaluated as to what type is a driver who controls the motor vehicle. For example, a third evaluation result may yield a driver type of fast-moving driver, a driver type of slow-moving driver, a driver type of economically driving driver, a driver type of unsafe driving driver, etc.

A driver of the type of fast-moving driver can be understood as meaning a driver who exhibits a driver's behavior that is usually referred to as sporty. Furthermore, a fast-moving driver can be understood to mean that The driver would like to spend as little time as possible in order to get from the start to the finish by means of the vehicle. The driver type of the fast-moving driver can thus represent a driver who shows a sporty driving behavior, and / or a driver who wants to achieve the fastest possible achievement of objectives.

This is followed by forming a grouping comprising at least two of the determined types. For example, the grouping may include a road type and a driving maneuver type, a driving maneuver type and a driver type or a driver type and a road type. Or the grouping includes a compilation of the three types. This results in a large number of possible groupings, for example a first grouping, which is assigned to the road type motorway, the overtype maneuver type and the driver type to fast-moving drivers, a second grouping, which is assigned to the street type city street and the driver type of fast-moving drivers, etc ,

Furthermore, provision is made of an initial, adaptable artificial neural network assigned to the grouping in a computing unit of the motor vehicle, wherein a specific driver behavior for the grouping is predefined in the initial, adaptable artificial neural network. That is, in a suitable computing unit of the motor vehicle, which may be embodied, for example, as a controller system, an artificial neural network-based computer program is implemented, which is usually referred to as an artificial neural network. For example, the artificial neural network manufacturer can be integrated into the motor vehicle, in particular in the manufacture of the motor vehicle or transmitted by means of a server device on the manufactured motor vehicle. This artificial neural network is adaptable and initially has an initial configuration or programming, whereby the adaptable artificial neural network is able to an actual driving situation, which results from the evaluation of the road type, the driving maneuver type and / or the driver type recognize and expect a driver action corresponding to the actual driving situation. For example, in the motor vehicle there is an initial, adaptable artificial neural network that belongs to a driving situation, which is assigned a grouping with the road type highway, the driving maneuver type overtaking and the driver type fast driving driver. For this driving For example, the initial, customizable artificial neural network may expect, among other things, that the driver first activates the turn signal and then deactivates it after a certain amount of time.

Further, comparing the driver's actual driver behavior with the driver behavior predefined in the adaptable artificial neural network and adapting the adaptive artificial neural network based on the comparison performed by the computing unit of the motor vehicle. In other words, by means of the arithmetic unit, an adaptation process is carried out in which the initially initial, adaptable artificial neural network is programmed, or in which the initially initial, adaptable artificial neural network programs itself by a comparison between a real driver action and one of the initial, customizable artificial neural network expected driver action takes place. Based on the result of the comparison, the expected driver action is adjusted so that the expected driver action is as similar as possible to the actual driver action when a particularly similar, in particular a same driving situation occurs again. The artificial neural network changed or adapted for the first time by the adaptation process is no longer initial. By repeating the adjustment process several times, the adaptive artificial neural network learns to expect driver behavior for a particular driving situation that progressively deviates less and less from the actual driver behavior of a human driver for the particular driving situation.

The adaptation process may, for example, proceed as follows: in an actual driving situation in which the road type motorway, the driving maneuver type overtaking and the driver type of fast driving driver have been detected, it is determined that the driver is moving the accelerator pedal of the motor vehicle with an actual accelerator pedal acceleration, which deviates from the expected accelerator pedal acceleration. The adaptive artificial neural network can thus detect a difference between the expected and the actual accelerator pedal acceleration. The adaptable artificial neural network corrects the expected accelerator pedal acceleration according to the difference, so that an adjusted expected Accelerator acceleration is stored in the adaptable artificial neural network, after which the adjustment process is terminated.

For a possible following the adjustment process driving situation in which also the road type highway, the driving maneuver type overtaking and the driver type fast driving driver has been detected, the adjustment process is started again. Thus, another difference is determined from a new, actual accelerator pedal acceleration and the now adjusted, expected accelerator pedal acceleration. Based on the further difference, analogously to the adaptation process described above, the adaptable artificial neural network is adapted again.

The adaptation process can be repeated so often, that is, the corresponding artificial neural network learns until the corresponding artificial neural network has learned out. This may mean that a sufficiently small deviation between the expected and the actual driver behavior z. B. is detected by the adaptable artificial neural network or that the customizable artificial neural network receives an instruction to discontinue.

An essential aspect of the invention is that a grouping is formed which comprises at least two of the determined types. This makes the behavior of human drivers in vehicles with a particularly high level of detail usable.

In another embodiment of the present invention, as long as an actual driving situation corresponds to the types of grouping, comparing the actual driver behavior of the driver with the driver behavior defined in the artificial neural network and adjusting the artificial neural network depending on the comparison performed respectively. In other words, the adaptable artificial neural network present in the motor vehicle continuously evaluates the data record, which may be designed in particular as a data stream, during the journey. This means that the adaptation process for the adaptable artificial neural network described above takes place with a low, preferably without a time delay, in particular already during the driving operation. This allows the adaptable artificial neural network to adapt very quickly to the expected driver behavior.

Alternatively or additionally, in the present invention it is also possible for the data record to be stored and to subsequently select at least a portion of the data record, wherein the proportion corresponds to an actual driving situation of the types of the grouping, and to comparing the actual driver behavior based on the proportion the driver with the driver behavior defined in the artificial neural network and adjusting the artificial neural network in dependence on the comparison performed. In other words, the data set or data stream acquired during the drive of the motor vehicle is held in the motor vehicle during the drive of the motor vehicle at least until at least one adaptation process has taken place. For this purpose, for example, after completion of the journey, the record is examined as to whether there are any of the driving situations stored in the record, which correspond to the grouping. Thus, at least one driving situation is identified in which, for example, the road type highway, the driving maneuver type overtaking and the driver type of fast driving driver have been recognized, if the grouping of the road type highway, the driving maneuver type overtaking and the driver type are assigned to fast driving drivers. Then, the identified driving situation is compared with the grouping, that is, the driver action stored in the record by means of the adaptive artificial neural network with the expected driver action stored in the grouping is compared, so that a fitting process is carried out as described above.

This results, for example, in the advantage that a multiplicity of driver actions carried out by the driver can flow into the adaptation process, as a result of which a particularly low amount of computation is to be operated until a sufficiently small deviation between the expected and the actual driver behavior is ascertained. Furthermore, filtering can take place, as a result of which the stored before the adjustment process, whether or not they should be included in the adjustment process.

In a further advantageous embodiment of the invention, it is provided that the provision of the data set is carried out by data from at least one of the following devices of the motor vehicle are detected: a navigation system, a camera system, a radar sensor system, a sensor system for detecting a vehicle movement, a sensor system for detection the driver's behavior. This means that devices installed in the motor vehicle are used to acquire the data on the basis of which the data record is created. For example, data provided by the navigation system can be used to determine the longitudinal speed of the motor vehicle, as well as for route planning.

The camera system, which may include a plurality of cameras, which may be oriented towards occupants of the motor vehicle or to the surroundings of the motor vehicle, may be used in addition to the data acquisition for driver assistance systems for data acquisition of the driver behavior. A radar sensor system used for example for the adaptive cruise control can also be used in addition to the detection of surrounding the motor vehicle traffic to capture data on a development of the vehicle environment.

The sensor system for detecting a vehicle movement may be firstly applied to detecting an automobile-related accident and secondly applied to the detection of the driving maneuver.

For a detection of fatigue, the sensor system for detecting the driver behavior is usually used, which alternatively or additionally can be used when recording speeds and / or accelerations which are applied to the steering wheel or the pedals.

Thus, the facilities have multiple functions, which makes them particularly cost-effective to use. According to a further advantageous embodiment of the present invention, it is provided that the determination of the road type is carried out by the provided data set is evaluated at least in terms of a location from map data of the navigation system. For example, map data that is part of the navigation system of the motor vehicle can be used for route planning, as well as for detecting on which type of road the motor vehicle is located. This offers the advantage that a determination of the type of road on which the motor vehicle is located or on which the motor vehicle is driven can be verified.

In a further embodiment of the present invention, it is provided that the determination of the driver type is carried out by evaluating the provided data record at least with regard to one of the following characteristics: proportion of the respective roads of the respective road types, a speed and acceleration profile with respect to three main axes of the motor vehicle, a steering wheel angle and a steering wheel angle curve, a respective pedal position of a brake and accelerator pedal. That is, determining the driver type is based on one or a combination of two or three of the mentioned characteristics. For example, the driver type can be determined by evaluating what type of roads the driver is traveling mainly by means of the motor vehicle. Thus, for example, a driver type highway driver, a driver type urban transport driver, a driver type highway driver, etc. can be identified. Also conceivable is a mixed type, z. B. a driver type, which drives proportionally on highways and highways. In addition, the driver types of fast-moving drivers, slow-moving drivers, economically driving drivers, etc. can be determined on the basis of the speed and acceleration profiles, for example. Typically, for example, a high-speed driver often experiences high lateral accelerations compared to a slow-moving driver. By means of an evaluation of the steering wheel angle or Lenkradwinkelverlaufs can be detected whether the driver quickly or jerkily or gently executes steering movements, whereby z. B. an assignment to the type of fast-driving driver or slow-moving driver can be done. An assignment to the type of fast-moving driver or slow-moving driver can be made in an analogous manner by the respective pedal position of the accelerator or brake pedal and the accelerations and speed. speeds with which the respective pedals are actuated in each case.

A driver, in addition to the types of drivers mentioned, for example, the driver type unsafe driving drivers are assigned, if the evaluation of the record shows that the driver behavior particularly often leads to a situation to be corrected, in particular emergency action of the driver. Such an emergency action may include, for example, particularly fast, jerky steering movements, an abrupt termination of an overtaking process or a particularly strong braking, in particular emergency braking. In particular, the driver can be classified as unsafe if the driver behavior leads to an accident, in particular to a traffic accident.

The fact that the method uses a particularly high level of detail in the utilization of the data, that is, takes into account that the type of driver can change during a journey, it is provided that the driver type continuously, for example, in a certain time rhythm or a change in the Road type etc. again determined and / or verified.

An assignment of the driver to a particular type of driver is thus possible, with the omission of a query of a self-assessment of the driver in question, whereby an influence of misjudgment of the driver with respect to his own driving skills is omitted in the present process.

Furthermore, it has been shown to be advantageous that a logic provided in the motor vehicle does not recognize any suitable driver behavior and excludes this from the adaptation of the artificial neural network. Thus, if a driver action leads to a situation to be corrected, in particular emergency action by the driver, or to an accident, in particular a traffic accident, this is recognized by the logic in order to prevent the driver action preceding the corrective situation or the traffic accident from entering the adaptation process. In other words, it is provided that driver actions, which are linked, for example, to the driver type of unsafe drivers, are not used for an adaptation process of the adaptable artificial neural network. This is just behavior used by human drivers in motor vehicles, which is considered safe for the driver, the motor vehicle and / or other people.

A further embodiment of the invention provides that different groupings are predefined in the motor vehicle, each of which has different combinations of types, wherein for the given groupings respective initial artificial neuronal networks to be adapted are provided, in each of which a grouping-specific initial driving behavior for the respective group Groupings is predefined. In other words, a plurality of customizable artificial neural networks may be provided, each associated with a respective grouping and initially having an initial configuration, respectively, whereby the respective customizable artificial neural networks each expect a driver behavior that is grouping-specific. This means that the adaptable artificial neural networks are able to recognize an actual driving situation which results from the evaluation of the road type, the driving maneuver type and / or the driver type and expect a corresponding actual driver action. The respective groupings are assigned different combinations of types from each other. Thus, it is therefore possible to specify which driving situations are to be included in the artificial neural network assigned to the associated grouping and to be adapted. For example, it can be specified that only the artificial neural network to be adapted is adapted so that the behavior of human drivers learns during an urban overtaking process. As a result, a desired, particularly high level of detail is achieved with which the behavior of human drivers in vehicles is harnessed.

An advantageous embodiment of the invention provides that by means of the at least one motor vehicle, a transmission of the respective existing, each one grouping associated artificial neural networks to a server device takes place after they have been adapted. In this case, transmission to the server device or uploading from the motor vehicle to the server device can take place, for example, when the motor vehicle is connected to a vehicle-produced data exchange system, such as a vehicle diagnostic system wirelessly or by cable. Particularly suitable is an upload while the motor vehicle is in a workshop. In this case, learned artificial neural networks, which are for example optimally adapted, can be uploaded from the vehicle to the server device, as well as artificial neural networks that have not yet learned. When uploading to the server device of the respective artificial neural networks, a copy thereof may remain in the motor vehicle. If such a transmission of the artificial neural networks respectively present in a motor vehicle and assigned in each case to a grouping occurs more frequently and / or by means of a large number of motor vehicles, a collection arises in the server device which consists of a multiplicity of adapted artificial neural networks. This collection may be in a desired manner by personnel z. As the manufacturer or by suitable computer-based routines continue to be processed.

In a further advantageous embodiment of the invention, it is provided that transmitting the at least one adapted artificial neural network from the server device to at least one target vehicle and applying the at least one adapted artificial neural network in the target vehicle. In other words, at least one learned artificial neural network can be downloaded to a motor vehicle or to a plurality of motor vehicles. In this case, the transmission to the respective motor vehicle or download from the server device into the respective motor vehicle can take place, for example, if the respective motor vehicle is in each case connected wirelessly or by cable to a vehicle manufacturer's own data exchange system, such as a vehicle diagnostic system. Particularly suitable is a transmission to the respective motor vehicle, while the respective motor vehicle is located in a workshop.

Furthermore, the learned artificial neural network in the motor vehicle is used, which is set up to be controlled at least partially by the learned artificial neural network. That is to say, the artificial neural network downloaded into the respective motor vehicle is able to undertake, at least partially, activities which the human driver performs to control a motor vehicle, with the aid of suitable devices of the motor vehicle. For example, the activities may include activating and / or deactivating vehicle functions, such as a direction of travel. pointer, a heating device, a cruise control system, etc. and / or functions for controlling a vehicle engine include. The advantage here is that the driver is particularly relieved.

According to a further advantageous embodiment of the invention, it is provided that in the server device a best selection is performed by filtering out a certain percentage of the matched artificial neural networks per grouping and applying at least one of the filtered-out artificial neural networks in the target vehicle. This means, for example, that the collection in the server is handled in a manner by a staff z. B. the manufacturer or by suitable computer-assisted routines is further processed, so that a certain number of group-specific artificial neural networks, which certain criteria, such as low fuel consumption, fastest possible achievement of goals, most economical driving style, etc., can be selected. For example, a best artificial neural network can be identified and downloaded into a motor vehicle or a plurality of motor vehicles for the lowest possible fuel consumption, activities that the human driver performs to control a motor vehicle At least partially take over devices of the motor vehicle. Similarly, for example, a best artificial neural network can be identified and downloaded to a motor vehicle or a plurality of motor vehicles, wherein the best artificial neural network for the fastest possible achievement of goals, the activities performed by the human driver to control a motor vehicle, using appropriate facilities of the motor vehicle to take at least partially. The advantage here is that the driver of the motor vehicle benefits from the activities that take over the adapted artificial neural networks that have been adapted by particularly capable drivers.

Alternatively or additionally, it is also possible for the application of at least one adapted artificial neural network to take place in the target vehicle, in which the adapted artificial neural network at least partially autonomously controls the target vehicle controls. That is, the adapted artificial neural network, which has been downloaded for example in the motor vehicle, takes over activities for controlling the motor vehicle, which are suitable to drive the motor vehicle in particular on the road. Such activities may include, for example, acceleration, braking, steering, etc. This is referred to as teilautonomes or autonomous driving, depending on the nature and amount of the assumed for the driver activities for controlling the motor vehicle. The motor vehicle, which is thus teilautonom or autonomously controlled, is thus able to be operated in a semi-autonomous or autonomous operation.

In this case, it is possible for the driver of the motor vehicle to select a driver type, for example by means of a selection device, so that the motor vehicle exhibits a driving behavior corresponding to the selected driver type during the partially autonomous or autonomous operation of the motor vehicle. Further, an action taken by the driver of the motor vehicle may be selecting one of a plurality of alternative routes suggested by the navigation system. This may mean, for example, that if the driver has selected the driver type of fast-moving driver or if the driver type of fast-moving driver is active in the motor vehicle, one of the proposed routes is selected by means of the artificial neural network, thus realizing a particularly timely arrival time can be. Is as a driver type economically driving driver in the motor vehicle active or selected, z. For example, an alternative of the suggested routes is selected or an alternative route is calculated / proposed so that as little fuel as possible is used to reach the destination.

The fact that the adapted artificial neural network assumes these activities, the driver is particularly relieved. The driver can be partially relieved, for example, by the driver continues to monitor and control the steering of the motor vehicle, the adapted artificial neural network takes over a speed control of the motor vehicle. The driver can also be relieved to a greater extent, so be taken out of the process of controlling the motor vehicle. In particular, the driver can be completely removed from the process of controlling the motor vehicle, so that the motor vehicle ultimately travels in the traffic without the intervention of the driver, wherein the driver can turn to other, in particular non-driving activities, while the motor vehicle fully autonomously drives to a predetermined by the driver target.

Furthermore, it has proven to be advantageous that the application of at least one adapted artificial neural network in the target vehicle takes place by the adapted artificial neural network intervening in a route guidance of the navigation system. This means that the adapted artificial neural network, which has been downloaded, for example, into the motor vehicle, recognizes that a traffic obstruction, for example a traffic obstruction, for example, on a route of the route or route planning predetermined by the driver by means of the navigation system of the motor vehicle. As a traffic jam, a construction site, etc. is present. Such recognition can, for. B. done based on data provided by the navigation system of the motor vehicle. The recognition can also be done by being evaluated by the artificial neural network when the driver, deviating from the recommended route of the navigation system, selects a different route, eg. B. earlier than expected leaves the route, z. B. turns. The artificial neural network then proposes an alternative route based on the adjustments of drivers who have previously used a similar, in particular the same route to avoid the traffic obstruction.

An advantageous embodiment of the invention provides that in the target vehicle per grouping a plurality of adapted artificial neural networks each having at least one grouping-specific control decision is used, wherein, if an actual driving situation corresponds to a grouping, a majority decision maker from the individual group-specific control decisions of the individual, adapted artificial neural networks, an overall control decision for the target vehicle falls. In other words, the motor vehicle per grouping comprises a plurality of adapted artificial neural networks, each of which is capable of taking over activities of the driver. That is to say, the group-specific adapted artificial neural networks used in the motor vehicle propose a driver action in each case in an actual driving situation which corresponds to the grouping, so that a large number of possible control decisions is present for the actual driving situation. Individual tax decisions may differ from each other be lent, so that a majority decision is used to find a total tax decision, which compares the individual tax decisions with each other and z. For example, the control decision most frequently suggested is identified as the overall control decision. The identified overall control decision is then implemented using the appropriate means so that the motor vehicle behaves according to the overall control decision. For example, with ten artificial neural networks, six may suggest activating the turn signal for five seconds in the actual driving situation, with the four other artificial neural networks suggesting to activate the winker for three seconds. Then the turn signal is activated according to the result of the majority decision maker for five seconds.

In such an embodiment of the invention, the degree of deviation between the expected and the actual driver action of the individual artificial neural networks being learned may be high, since the majority decision maker is used, which prevents any wrong decisions of individual artificial neural networks from being incorporated in the overall control decision. This results in the advantage that a computational effort that would have to be performed, for example, in the adaptation process or in the server device can at least partially be dispensed with, which would have to be expended for the generation of an artificial neural network that achieves an overall control decision of a similar, in particular equal quality.

The present invention also provides a system for operating at least one motor vehicle, which is set up to carry out the method according to the invention or an advantageous embodiment of the method according to the invention. Advantageous embodiments of the method according to the invention are to be regarded as advantageous embodiments of the system according to the invention, wherein the system comprises in particular means for carrying out the method steps.

Further advantages, features and details of the invention will become apparent from the following description of preferred embodiments and from the drawings. The features and feature combinations mentioned above in the description as well as the features mentioned below in the description of the figures and feature combinations are useful not only alone but also in combination with each other, without departing from the scope of the invention.

The drawing shows in:

1 shows a flowchart in which the steps of a method for operating at least one motor vehicle are shown;

FIG. 2 shows a further embodiment of the method, wherein an adaptation process takes place with a time delay; FIG.

Fig. 3 shows the motor vehicle of another embodiment of the method, wherein

 Facilities of the motor vehicle are shown;

4 shows the motor vehicle of a further embodiment of the method, wherein artificial neural networks are respectively associated with associated groupings; another embodiment of the method, wherein the artificial neural networks between a server device and the motor vehicle are transmitted;

FIG. 6 shows the server device in which a best selection of the artificial neural networks takes place; FIG.

7 shows a road on which the motor vehicle is at least partially autonomously controlled according to a further embodiment of the method;

Fig. 8 shows a process in which the artificial neural network active in a

 Route planning intervenes;

9 shows a further embodiment of the method in which a majority decision maker is used; and 10 shows a system for operating the motor vehicle.

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

The method for operating at least one motor vehicle 20 shown for the first time in FIG. 2 is shown in FIG. 1 in a flow chart. The method includes a step S1 that performs providing a data set 10 that characterizes a vehicle environment, driver behavior, and the motion of the motor vehicle 20. To perform this characterization, input data sets such as a vehicle environment data set 12, a driver behavior record 14, and / or a vehicle motion data set 1 6 are provided.

In this case, the vehicle surroundings data record 12 contains, for example, data on a development of the environment or the environment of the motor vehicle 20, so that it is possible to determine whether, for example, a pedestrian path, a cycle path or building structures (eg, buildings in addition to a carriageway used by the motor vehicle 20) , Tunnel walls, passive protective structures such as protective barriers or concrete walls). Furthermore, data on conditions in the environment of the motor vehicle 20 may be stored in the vehicle surroundings data record 12, for example information about a brightness, an outside temperature a precipitate, etc. By means of an evaluation of the data stored in the vehicle surroundings data record 12, the vehicle environment can be characterized.

The driver behavior record 14 includes data describing behavior of a driver located in the motor vehicle 20, and the driver may perform activities for controlling the motor vehicle 20. A driver's behavior includes an action that the driver performs (driver action), how the action is carried out, whether the driver shows signs of tiredness, whether the driver is distracted, whether the driver is concentrated and / or further information, the disconfirmation about give the driver behavior. By means of an evaluation of the data stored in the driver behavior data record 14, the driver behavior can be characterized. The vehicle movement data record 16 contains data by means of which a total movement of the motor vehicle 20 can be described completely in particular. The total movement of the motor vehicle 20 is usually divided into a system of three main axes, namely a longitudinal, a transverse and a vertical axis, each extending through the motor vehicle 20 and at a point, for. B. Mass center of the motor vehicle 20 intersect. This means that the total movement can comprise a longitudinal velocity or acceleration along the longitudinal axis, a transverse velocity or acceleration along the transverse axis, and / or a vertical velocity or acceleration along the vertical axis. Furthermore, the total movement about the longitudinal axis may include a roll acceleration or acceleration, the pitch axis may have a pitching speed and / or a yawing speed may be around the vertical axis. Furthermore, data about possibly existing vibrations along or about the respective main axis can also be stored in the vehicle movement data record 16. By means of an evaluation of the data stored in the vehicle movement data record 14, the vehicle movement can be characterized.

By combining the input data sets, a data set is provided which is provided for the following steps of the method.

FIG. 1 also shows sub-step S2a, sub-step S2b, and sub-step S2c, which each form a sub-step of step S2 of the method for operating at least one motor vehicle 20, which is also depicted in FIG. In the step S2, a road type, in the sub-step S2b a driving maneuver type and / or in the sub-step S2c a driver type are determined in particular simultaneously in the sub-step S2a. In particular, at least two of the mentioned types are determined.

When determining the road type in sub-step S2a, the data record provided in step S1 is evaluated as to which type of road is the road on which the motor vehicle 20 is driven. For example, the road type of a motorway is possible, which is characterized, inter alia, by high longitudinal speeds of the vehicles driving thereon, a multi-lane formation of roadways, a structural separation of the roadways and a motorway-specific signage. can draw. Furthermore, the road type of a highway is possible, which can be characterized inter alia by a longitudinal speed of the vehicles driving on it up to about 100 kilometers per hour, a curvy road course and the presence of road intersections. In addition, the road type of a city street is possible, which can be characterized, inter alia, by low longitudinal speeds of the vehicles driving thereon, a variety of buildings next to the road and by the presence of other people on and / or next to the road. Other types of roads are conceivable.

When determining the driving maneuver type in sub-step S2b, the data record provided in step S1 is evaluated as to what type of driving maneuver is performed by means of the motor vehicle 20. For example, the driving maneuver type of a follow-on journey is possible, in which the motor vehicle 20 is driven behind another road user, wherein the motor vehicle 20 uses the same lane as the other road users. Furthermore, another driving maneuver type of overtaking is possible, in which the motor vehicle 20 drives past another road user, wherein the motor vehicle 20 uses a different lane than the other road users. In addition, another driving maneuver type of turning is possible, in which by means of the motor vehicle 20, a change of direction at intersections or junctions is performed. Other driving maneuvers are conceivable.

When determining the driver type in sub-step S2c, the data record provided in step S1 is evaluated as to which type of driver controls the motor vehicle 20. For example, the driver type of a fast-moving driver is possible, which can be characterized among other things by a frequent Ausreizen a maximum speed limit, fast turning on bends and frequent changes from one to another lane. Furthermore, the driver type of a slow-moving driver is possible, which can be characterized, inter alia, by a slow driving, a rare change from one to another lane and a slow turning in curves. Other driver types are conceivable. A step S3 is also shown in FIG. In the step S3, a grouping 38 (not shown in FIG. 1, first shown in FIG. 4) is formed from the types determined in step S2, ie road type, driving maneuver type and driver type, the grouping 38 having at least two different types each having. For example, a group 38 may be formed from a road type and a driver type, a driver type and a driving maneuver type, and / or a driving maneuver type and a road type, respectively. Further, the grouping 38 may also be formed by including a road, a driving maneuver and a driver type.

In a step S4, which in FIG. 1. is shown, an artificial neural network 22 (not shown in Fig. 1, first shown in Fig. 2) in the motor vehicle 20, in particular in a computing unit 18, for. B. provided in a controller system. The control unit system can be designed, for example, to monitor and / or trigger functions and / or states of the motor vehicle 20. The artificial neural network 22 is provided with the data set 10 or the artificial neural network 22 has access to the data set 10. Furthermore, the artificial neural network 22 is adaptable and associated with a specific grouping 38. That is, there is an adaptive artificial neural network 22 in the motor vehicle 20 which is associated, for example, with a grouping 38 comprising the road type highway, the driving maneuver type following drive and the driver type of slow driving driver. In addition, the adaptive artificial neural network 22 is integrated with the motor vehicle 20 such that it is capable of detecting an actual driving situation 24 in which the motor vehicle is located. The actual driving situation 24 can be classified, based on the types mentioned. For example, the customizable artificial neural network 22 may determine that the actual driving situation 24 (not shown in FIG. 1, first shown in FIG. 2) includes the highway type of highway, the driving maneuver type of following travel, and the driver type of slow-moving drivers. In an initial configuration of the adaptive artificial neural network 22, an expected driver behavior associated with the grouping 38 is predefined, that is, the adaptive artificial neural network 22 may expect, among other things, in the actual driving situation 24 For example, the driver keeps a driving speed and a distance to a preceding road user the same.

FIG. 1 also shows a step S5 in which a comparison takes place between a real driver action and an expected, ie predefined, driver action from the initially initial, adaptable artificial neural network 22. That is, the record 10 is evaluated by the customizable artificial neural network 22 so that the customizable artificial neural network 22 can determine what actual driver action the driver is performing. This comparison can be performed for example by means of the computing unit 18 of the motor vehicle 20, so z. In the control system. Further, in the step S5, an adjustment process takes place in which the adaptable artificial neural network 22 is changed based on a comparison result. That is, the initial initial configuration of the adaptive artificial neural network 22 is changed in the fitting process so that the customizable artificial neural network 22 has customized programming. By means of any further adaptation processes, the adapted programming is further adapted. In the adaptation process, in the custom artificial neural network 22 programming, the expected driver action is changed so that the expected driver action is particularly well-matched to actual driver action. For this purpose, the adaptable artificial neural network 22 determines the actual driving situation 24, that is to say, for example, that in reality the road type highway, the driving maneuver type follow-on driving and the driver type are slow-moving drivers. Then, it is determined by the adaptable artificial neural network 22 whether the actual driving situation 24 corresponds to the grouping 38. If so, the customizable artificial neural network 22 checks which driver action is deposited for that grouping 38, that is expected, and compares the expected driver action with the actual driver action that the driver is performing. As a result of comparison, a difference between the expected and the actual driver action is generated, for example, a difference between an expected and an actual acceleration of the vehicle speed. Then, the adaptive artificial neural network 22, in programming, appropriately changes the expected acceleration of the vehicle speed. This is now in the programming one adjusted, expected or predefined vehicle acceleration, which is assigned to the group 38. The adjusted, predefined vehicle acceleration is used for a possible later actual driving situation 24, which corresponds to the grouping 38, to generate the comparison result.

In Fig. 2, another embodiment is shown, wherein the adjustment process takes place with a small time delay. In particular, the adjustment process can be carried out with a particularly low, ideally without a time delay. That is, the data set 10 is evaluated by the tunable artificial neural network 22 during operation, particularly during driving of the motor vehicle 20, so that the tunable artificial neural network 22 can determine during driving that an actual driving situation 24 corresponds to the grouping 38 and what actual driver action the driver performs. Further, during the driving operation, the adjustment process takes place in which the adaptive artificial neural network 22 is changed based on the comparison result as described above. In other words, ideally immediately after it has been determined that the actual driving situation 24 of the grouping 38 corresponds to the adaptation process of the adaptable artificial neural network 22 is started, so already in another actual driving situation 24, which again the grouping 38 corresponds to the customized, predefined driver action is changed to a lesser extent than at the first occurrence of the actual driving situation 24.

In a further embodiment, the adjustment process takes place at a high time delay. In particular, the adaptation process can take place during a different operation from the driving operation of the motor vehicle 20, that is, for example, after a termination of the driving operation, for. B. while the motor vehicle 20 is parked. That is, the data set 10 generated, for example, in the driving operation is evaluated by the adaptable artificial neural network 22 after completion of the driving operation. Thereby, the adaptive artificial neural network 22 may determine after completion of the driving operation whether at least a portion of the data set 10 contains data of an actual driving situation 24 that has occurred during the past driving operation and corresponds to the grouping 38. Furthermore, the adaptable artificial neural network factory 22 after the end of the driving operation determine which driver action the driver has then executed. The adjustment process in which the adaptive artificial neural network 22 is changed based on the comparison result as described above may thus also be done after the completion of the driving operation.

Alternatively or additionally, it is possible that the adaptation process begins as soon as it is recognized that the motor vehicle 20 has changed to a road of a different road type, for example, was driven off a highway on a country road.

In Fig. 3, the motor vehicle 20 is shown a further embodiment of the method. The motor vehicle 20 has devices for providing the data record 10. Shown in detail are a navigation system 26, a camera system 28, a radar sensor system 30, a sensor system 32 for detecting a vehicle movement and a sensor system 34 for detecting a driver behavior. The devices can have a multifunctionality, so that, for example, the navigation system 26, which can be designed, for example, as a G PS satellite navigation system, can be used for route planning and for detecting or verifying the vehicle movement. The camera system 28, which may comprise a plurality of cameras, which may be oriented, for example, to the driver, to further occupants or to an outside area of the motor vehicle 20, may be used for imaging for pedestrian recognition and for recording the development next to the carriageway , Further, the radar sensor system 30, which may include, for example, a front-facing and a rear-facing radar sensor, may be used to detect other vehicles and detect a precipitate. The vehicle motion sensing sensor system 32 may be used to provide triggering of active vehicle safety systems for control and data for the record 10. The sensor system 34, which provides driver behavior data for the record 10, may alternatively or additionally also be used to detect if the driver is experiencing fatigue symptoms. In a further embodiment of the method for operating the at least one motor vehicle 20, the navigation system 26 is used to determine or at least to verify which road type the road is on which is driven by means of the motor vehicle 20.

By means of an evaluation of the provided data record 10, a driver type can be determined. That is, the driver who controls the motor vehicle 20 can be classified in terms of driver behavior and assigned to a driver type. This can be done by evaluating, based on the data set 10, how often the driver has driven on a road of a certain road type by means of the motor vehicle 20. For example, a driver type of fast-moving driver can be identified when it is clear that the driver prefers the highway-type highway, although target achievement would have been possible on roads of a different type. If the driver particularly frequently excites a maximum permissible maximum speed or if he generates high lateral accelerations on the motor vehicle 20 by fast cornering, the driver can be assigned to the driver type of fast-moving driver. The driver can also be assigned to the driver type of fast-moving driver, if it is determined that the driver particularly fast, a steering wheel of the motor vehicle 20 moves, that is, if often a particularly high steering wheel speed and a particularly high steering wheel acceleration. The driver may also be assigned to the driver type of fast-moving driver when it is determined that the driver is accelerating, particularly jerkily, by depressing an accelerator pedal or a brake pedal, or by using the accelerator pedal for a long period of time. In this way, a driver can be at least temporarily assigned, for example, to one of the following driver types: fast-moving driver, slow-moving driver, economically driving driver, unsafe driver, etc. A driver can be assigned to the driver type of unsafe driver, for example, when his driver actions have led to an accident or when his driver actions often lead to a corrective action. Such a corrective action may include driver actions which are suitable for preventing a self-inflicted accident, for example an emergency termination of a wrongly initiated overtaking operation. Furthermore, in FIG. 3, a logic 36 is depicted, by means of which unlearnable driver behavior is recognized, so that it is possible to exclude the unlearnable driver behavior from the adaptation process of the adaptable artificial neural network 22. For example, driver actions performed by a driver of the unsafe driver type may be excluded from the adjustment process. But it can also be driver actions, which are performed by a Fah rer of another driver type, be excluded from the adjustment process, for example, leading to an accident driver actions.

4, the motor vehicle 20 is shown a further advantageous embodiment of the method. It is provided that in the motor vehicle 20 of the grouping 38 an adaptable artificial neural network 22 is assigned, whereby a pairing 40 is formed. By allowing a plurality of arrays 38 and a plurality of customizable artificial neural networks 22, each associated with a grouping 38, to be present in the motor vehicle 20, a plurality of pairings 40 may be formed in the motor vehicle 20. The groupings 38 are each assigned a road type, a driving maneuver type and / or a driver type. If it is detected that there is an actual driving situation 24 corresponding to the grouping 38, the adaptable artificial neural network 22 expects grouping-specific driver behavior. Thus, it is possible, for example, to assign the pairings 40 each certain, different driving situations from each other. That is, the artificial neural network 22 of a mating 40 is adapted only when the actual driving situation 24 corresponds to the grouping 38, that is, for example, when the grouping 38 and the actual driving situation 24 both overtake on the highway by means of a fast-moving Driver, exhibit.

In Fig. 5, the vehicle 20 and a server device 42 is shown, wherein the power vehicle 20 is able to transmit in the motor vehicle 20 existing artificial neural networks 22 to the server device 42 or upload. For this purpose, the motor vehicle 20 and the server device 42 are connected to one another, for example, by means of a transmission cable 44 and / or a wireless data connection 46. The server device 42, in turn, is capable of providing artificial neural networks 22 in the server device 42 to the motor vehicle operator. 20 to transmit or download. The server device 42 may be configured, for example, as a networked vehicle manufacturer-side data exchange system, in particular as a vehicle diagnostic system. Artificial neural networks 22, which z. As in the motor vehicle 20 or in a variety of motor vehicles 20 have been adapted, can be uploaded to the server device 42, so that a plurality of adapted artificial neural networks 22 in the server device 42 comes about. In addition, a plurality of adapted artificial neural networks, which may have been further processed in the server device, may be downloaded into at least one target vehicle or a plurality of target vehicles, where the target vehicle may be the motor vehicle 20. In the respective target vehicles, the downloaded artificial neural networks 22 may be deployed by the artificial neural networks 22 taking actions for controlling the target vehicle 20 from the driver.

FIG. 6 shows the server device 42 in which a multiplicity of adapted artificial neural networks 22 are stored, wherein the adapted artificial neural networks 22 are each assigned to a grouping 38. In the server device 42, in a further embodiment of the method, a best selection is carried out so that for each grouping 38 a certain percentage of the adapted artificial neural networks 22 are filtered out and / or marked so that the percentage of adapted artificial neural networks 22 is used in the at least one target vehicle can. For example, a best artificial neural network 22a may be identified and downloaded to the at least one target vehicle (s) 20, where the best artificial neural network 22a is for activities that the human driver uses to minimize fuel consumption Control of the motor vehicle 20 and the target vehicle performs, at least partially take over using appropriate facilities.

In Fig. 7, the motor vehicle 20 is shown, which is located on a road 48. In accordance with a further embodiment of the method for operating the at least one motor vehicle 20, it is provided that the motor vehicle 20 is at least partly automatic. nom is controlled by means of the artificial neural network 22. This means that the adapted artificial neural network, which has been downloaded for example in the motor vehicle, takes over activities for controlling the motor vehicle, which are suitable for the motor vehicle, in particular in road traffic, for. B. to drive on the road 48 and follow their course 50. Here are valid traffic rules, z. For example, the maximum permissible speed is respected so that safe driving is possible.

Furthermore, as shown in FIG. 8, in a further embodiment of the method, the artificial neural network 22 actively engages in routing from a starting point 54 to a destination point 56. The route planning, which is usually provided by means of the navigation system 26, may include a route 58 and an alternative route 60, wherein the driver may select thereunder. For example, if the driver has selected route 58, then a traffic obstruction 64 is detected by the artificial neural network 22, e.g. As a construction site or a traffic jam, and the alternative route 60 equally a traffic obstacle, the artificial neural network 22 determines another route, eg. B. the alternative route 62, which is proposed to the driver at least for a use. The alternate route 62 may have been created based on experience of drivers who have used a similar route to bypass the traffic obstruction 64, thereby adjusting the corresponding artificial neural network 22. The similar route can be located, for example, in the same city.

FIG. 9 shows an embodiment of the method in which a majority separator 66 is used. In this embodiment, a plurality of artificial neural networks 22 are used in the target vehicle or motor vehicle 20 per grouping 38. If an actual driving situation 24 corresponds to the grouping 38, the individual artificial neural networks 22 present in the motor vehicle 20 each make a control decision 68. The individual control decisions 68 can each be a driver action, whereby the individual control decisions 68 can be different from one another , Preferably, the individual control decisions 68 are similar to each other. In order now to generate an overall control decision 70, based on which the force vehicle 20 or the target vehicle is actually controlled, the individual tax decisions 68 are given in the majority decision 66. In the majority decision 66, the individual control decisions 68 are processed on the basis of, for example, an averaging, so that the majority decision 66 can output the overall control decision 70, that is to say provide the motor vehicle 20 with the motor vehicle 20 or the target vehicle or the plurality of vehicles according to the overall control decision 70 Target vehicles can be controlled.

FIG. 10 shows a system 72 for operating the motor vehicle 20, which is set up to carry out the method for operating the at least one motor vehicle 20 according to one of the embodiments or combination of the embodiments of the method. For this purpose, the motor vehicle 20 comprises a data supply device 74, which is set up to characterize a vehicle environment, a driver behavior and the movement of the motor vehicle. These may include, for example, devices such as the navigation system 26, the camera system 28, the radar sensor system 30, the sensor system 32 for detecting a vehicle movement and / or the sensor system 34 for detecting the driver behavior. Furthermore, this can include a memory system which is designed to store the data record 10 at least temporarily, as well as a provision device by means of which the data record 10 is made available for further use.

In addition, the motor vehicle 20 has a determination device 76 which is set up to determine a road type, a driving maneuver type and / or a driver type on the basis of the provided data record 10. For this purpose, the data stored in the data record 10 can be examined as to what type the road 48 is, whereupon the motor vehicle 20 is driven. For example, the road type highway, the road type highway, the street type city street, etc. can be determined. Further, the data may be examined as to what type of driving maneuver is performed by the motor vehicle 20. For example, the type of driving maneuver overtaking, the type of driving maneuver following drive, the driving maneuver type turning, the driving maneuver type reverse parking etc. can be ascertained. Alternatively or additionally, the data may be examined to determine which type is a driver controlling the motor vehicle 20. For example, the driver type of fast-moving driver, the driver type slow-moving driver, the driver type economically driving driver, the driver type unsafe driving driver, etc. can be determined.

The motor vehicle 20 also has a grouping device 78 which is designed to form a grouping 38 of at least two of the determined types. In other words, the grouping device 78 is capable of assigning at least two of the determined types to the grouping 38.

Furthermore, the motor vehicle 20 has a network providing device 80, which is set up to provide at least one artificial neural network 22, which has an initial configuration or programming and is adaptable. The artificial neural network is provided in a computing unit 18, not shown, of the motor vehicle 20.

The motor vehicle 20 further comprises a comparison and adaptation device 82, which is adapted to the adaptation process, in which the initially initial, adaptable artificial neural network 22 is programmed or programs itself by comparing a real driver action and a from the initial, customizable artificial neural network 22 expected driver action takes place. Based on the result of the comparison, the expected driver action is adjusted so that the expected driver action is as similar as possible to the actual driver action when a particularly similar, in particular a same driving situation occurs again

LIST OF REFERENCE NUMBERS

10 dataset 66 majority decision-makers

12 Vehicle Environment Record 68 Control Decision

 14 Driver behavior record 70 Overall control decision

16 Vehicle Movement Record 72 System

 18 arithmetic unit 74 Datenbereitstellungseinrich¬

20 motor vehicle processing

 22 Artificial Neural Network76 Investigation Facility

 plant 78 Grouping facility

24 Actual Driving Situation 80 Network Deployment Setup

26 Navigation system

 28 Camera System 82 Comparison and Adjustment.

30 radar sensor system device

 32 sensor system for detection

 the vehicle movement

 34 Sensor system for detection

 a driver's behavior

 36 logic

 38 grouping

 40 mating

 42 server device

 44 transmission cable

 46 Wireless data connection

 48 street

 50 course of the road

 52 Route planning

 54 starting point

 56 target point

 58 route

 60 alternative route

 62 alternative route

 64 traffic obstruction

Claims

 CLAIMS: 1 . Method for operating at least one motor vehicle (20), with the steps:  - providing (S1) a data set (10) which characterizes a vehicle environment, a driver behavior and the movement of the motor vehicle (20); - determining (S2) of at least two of the following types based on the provided data set (10): road type of a road (S2a) traveled with the motor vehicle (20), driving maneuver type of a driving maneuver (S2b) performed by the motor vehicle (20), driver type the motor vehicle (20) controlling the driver (S2c);  - forming (S3) a grouping (38) comprising at least two of the determined types;  Providing (S4) an initial, adaptable artificial neural network (22) associated with the grouping (38) in a computing unit 18 of the motor vehicle (20), wherein in the initial, adaptable artificial neural network (22) a specific driver behavior for grouping (22) 38) is predefined; and Comparing the actual driver behavior of the driver with the driver behavior predefined in the adaptable artificial neural network (22) and adapting the adaptable artificial neural Network (22) in dependence on the comparison performed by means of the computing unit 18 of the motor vehicle (20). 2. The method according to claim 1,  characterized in that  as long as an actual driving situation (24) corresponds to the types of grouping (38), comparing the actual driver behavior of the driver with the driver behavior defined in the artificial neural network (22) and adapting the artificial neural network (22) depending on the comparison made respectively. 3. The method according to claim 1 or 2,  characterized in that  the record (10) is stored and subsequently at least a portion of the record (10) is selected, wherein in the share an actual driving situation (24) corresponds to the types of the grouping (38), and on the basis of comparing the actual driving behavior of the group Driver with the driver behavior defined in the artificial neural network (22) and adjusting the artificial neural network (22) in dependence on the comparison performed. 4. The method according to any one of the preceding claims,  characterized in that  the provision (S1) of the data set (10) takes place by detecting data from at least one of the following devices of the motor vehicle (20):  - a navigation system (26);  a camera system (28);  a radar sensor system (30);  - A sensor system (32) for detecting a vehicle movement; - A sensor system (34) for detecting the driver behavior. 5. The method according to any one of the preceding claims,  characterized in that  the determining (S2a) of the road type takes place in that the provided data record (10) is evaluated at least with respect to a location from map data of the navigation system (26). Method according to one of the preceding claims,  characterized in that  determining the driver type (S2c) by evaluating the provided data set (10) at least with respect to one of the following characteristics:  - Share of each road used for each road type;  a speed and acceleration course with respect to three main axes of the motor vehicle (20);  a steering wheel angle and a steering wheel angle course;  - A respective pedal position of a brake and accelerator pedal. Method according to one of the preceding claims,  characterized in that  a logic provided in the motor vehicle (20) recognizes non-responsive driver behavior and excludes it from matching the artificial neural network (22). Method according to one of the preceding claims,  characterized in that in the motor vehicle (20) different groupings (38) are predefined, which each have different combinations of the types, wherein for the given groupings (38) respective initial artificial neural networks (38) to be adapted are provided, in each of which a grouping-specific initial driving behavior for the respective groupings (38) are predefined. 9. The method according to any one of the preceding claims,  characterized in that  By means of the at least one motor vehicle (20), a transfer of the respective existing artificial neural networks (22), each associated with a grouping (38), to a server device (42) takes place after they have been adapted. 10. The method according to claim 9,  characterized in that  transmitting the at least one adapted artificial neural network (22) from the server device (42) to at least one target vehicle and applying the at least one adapted artificial neural network (22) in the target vehicle. 1 1. Method according to claim 9 or 10,  characterized in that  performing best selection in the server means (42) by filtering out per group (38) a certain percentage of the matched artificial neural networks (22) and applying at least one of the filtered artificial neural networks (22a) in the target vehicle. 12. The method according to any one of the preceding claims,  characterized in that  applying at least one adapted artificial neural network (22) in the target vehicle by the at least partially autonomously controlling the adjusted artificial neural network (22) the target vehicle. 13. The method according to any one of the preceding claims,  characterized in that the application of at least one adapted artificial neural network (22) in the target vehicle takes place by the adapted artificial neural network (22) intervening in a route guidance of the navigation system (26). 14. The method according to any one of the preceding claims,  characterized in that  in the target vehicle per grouping (38) a plurality of adapted artificial neural networks (22) each having at least one grouping-specific control decision (68) is used, wherein if an actual driving situation (24) corresponds to a grouping (38), a majority decision maker (66) from the individual grouping-specific control decisions (68) of the individual adapted artificial neural networks, an overall control decision (70) for the target vehicle falls. 15. A system for operating at least one motor vehicle (20), which is adapted to perform a method according to one of the preceding claims.