Method for generating a longitudinal control model for a vehicle for determining at least one longitudinal control variable for the longitudinal control of the vehicle, and longitudinal control profile determining device

Abstract

Proposed is a method for generating a longitudinal control model for a vehicle for determining, in particular within the vehicle, at least one longitudinal control variable for the longitudinal control of the vehicle (1) along an intended route (S), wherein: the longitudinal control model is generated on the basis of swarm data determined by means of a vehicle swarm (3) and stored in an external memory device (100); the swarm data comprise driving course data, which are characteristic of at least section-specific longitudinal control, and preferably lateral control, of the particular swarm vehicle (1) along a route section, and route data, which are respectively associated with the driving course data and which specify the course of the particular route section and preferably the position thereof, comprising: - providing a trainable machine-learning longitudinal control model which comprises a set of parameters for mapping at least one input variable of the longitudinal control model onto at least one output variable, said parameters being settable in the course of a training process to values learned as a result of the training process; and - training the longitudinal control model with the swarm data received from the external memory device, by virtue of the route data being supplied as input variables and by virtue of the at least one output variable being matched against at least one longitudinal control variable which is determined on the basis of the driving course data associated with the route data and which is characteristic of the longitudinal control.

Description

[0001] Description

[0002] Method for generating a longitudinal guidance model for a vehicle for determining at least one longitudinal guidance variable for the longitudinal guidance of the vehicle and longitudinal guidance profile determination device

[0003] The present invention relates to a method and a longitudinal guidance profile determination device for generating a longitudinal guidance model for a vehicle for determining, in particular for real-time determination, at least one longitudinal guidance variable for longitudinal guidance of the vehicle along an intended route.

[0004] Today's production vehicles are increasingly integrating more and more complex driver assistance functions. In addition to safety-related assistance functions, more and more comfort-related driving functions are being implemented and further developed. These driving functions not only provide lateral guidance within the roadway but also longitudinal guidance of the vehicle. Longitudinal guidance maintains predefined target speeds, which can be determined, for example, by factors such as preceding vehicles, speed limits, road geometry, or aggregated speed profiles along the planned trajectory.

[0005] The aggregated speed profiles can, for example, be created from the vehicle fleet's data on a backend, which can then be made available to the vehicle fleet. The use of aggregated speed profiles makes it possible to achieve more human-like driving behavior without the need for complex and computationally intensive algorithms. However, this requires an online connection to a backend system and in-vehicle data localization and further processing.

[0006] Path planning algorithms that can calculate speed profiles online in the vehicle are known from the state of the art. The disadvantage of these path planning algorithms is that they require a large amount of resources.

[0007] DE 102017213 071 A1 discloses a device for reducing the speed of a motor vehicle to a predetermined value at a predetermined position. For this purpose, a position detection device accesses map data and compares it with a planned route to determine the positions where a traffic light, pedestrian crossing, intersection, stop sign, or other traffic sign is located.

[0008] DE 102015 005 902 A1 discloses a method for operating a motor vehicle, in which a relative position of the motor vehicle with respect to a traffic light located in front of the motor vehicle in the direction of travel is determined using a provided navigation map. Furthermore, a traffic flow is determined based on a traffic flow determined by a detection device of the motor vehicle. A traffic light sequence is determined based on the determined traffic flow, and the motor vehicle is determined based on the determined traffic light sequence.

[0009] DE 102014215673 A1 discloses a driver assistance system in a motor vehicle. It proposes a detection system for detecting upcoming events that require a reduction of the current speed to a reduced target speed, and a functional unit for determining a deceleration strategy for reducing the speed from the current speed to the target speed at the location of the upcoming event. The driver can create multiple speed profiles for a specific route section, which are then activated according to predefined rules.

[0010] The present invention is based on the object of overcoming the disadvantages known from the prior art and of providing a method and a longitudinal guidance profile determination device for generating a longitudinal guidance model for a vehicle for determining, in particular for real-time determination, at least one longitudinal guidance variable for the longitudinal guidance of the vehicle along an intended route, by means of which a reliable and rapid implementation of a longitudinal guidance in a vehicle can be achieved.

[0011] The object is achieved according to the invention by the subject matter of the independent claims. Advantageous embodiments and further developments of the invention are the subject matter of the dependent claims.

[0012] In a method according to the invention, in particular a computer-implemented method, for generating a longitudinal guidance model for a vehicle for (in particular vehicle-internal) determination, preferably for real-time determination, of at least one longitudinal guidance variable for the longitudinal guidance of the vehicle along an intended route, the longitudinal guidance model is generated on the basis of swarm data determined by means of a swarm of vehicles and stored on an external storage device.

[0013] The swarm data includes driving path data that is characteristic of at least partial longitudinal, and preferably transverse, guidance of the respective swarm vehicle along a route section. Furthermore, the swarm data includes route data associated with the driving path data, which indicate the course of the respective route section and preferably its position.

[0014] Preferably, the longitudinal guidance model is generated by a particularly processor-based longitudinal guidance profile generation device (which is described in more detail below). This device may include the external storage device. However, it is also conceivable that the longitudinal guidance profile generation device receives the swarm data from the external storage device on which the swarm data is stored, upon submission of a request, and / or retrieves it from this device. It is conceivable that only a temporary communication connection exists between the longitudinal guidance profile generation device and the external storage device for data exchange.

[0015] It is conceivable that the longitudinal guidance profile generation device is (geographically) spaced from the external storage device and / or from the vehicle to which the longitudinal guidance model is to be provided. It is conceivable that the longitudinal guidance profile generation device is a backend of an OEM or vehicle manufacturer or a service provider.

[0016] Preferably, swarm data are retrieved from the external storage device to generate a longitudinal guidance model.

[0017] Preferably, swarm data (for generating a longitudinal guidance model), in particular transmitted in response to a request, are received from the external storage device.

[0018] The method comprises providing a trainable machine-learning longitudinal guidance model, which comprises a set of (in particular trainable) parameters for mapping at least one input variable of the longitudinal guidance model to at least one output variable. The parameters can be set during a training process to values ​​learned as a result of the training process. In particular, the parameters are fixed after completion of the training process. The machine-learning longitudinal guidance model is preferably based on an (artificial) neural network.

[0019] Preferably, the data (to be processed), in particular the route data (or data derived therefrom), are fed to the longitudinal guidance model or the (artificial) neural network as input variables. Preferably, the longitudinal guidance model or the artificial neural network maps the input variables to output variables depending on a parameterizable processing chain, wherein the output variable is preferably the at least one longitudinal guidance variable or, preferably, a plurality of longitudinal guidance variables are selected as output variables.

[0020] Preferably, the neural network is designed as a deep neural network (DNN), in which the parameterizable processing chain has a plurality of processing layers, and / or a so-called convolutional neural network (CNN) and / or a recurrent neural network (RNN).

[0021] The method comprises training the longitudinal guidance model with the swarm data (received from the external storage device or based on the data stored on the external storage device) by supplying route data (to the longitudinal guidance model) as input variable(s) and comparing the at least one output variable (preferably the plurality of output variables) (output by the longitudinal guidance model) with at least one longitudinal guidance variable determined based on the driving path data associated with the route data and characteristic of the longitudinal guidance. In particular, the driving path data, preferably the driving path data characteristic of the longitudinal guidance, are used as ground truth data during the training process.

[0022] Preferably, the longitudinal guidance model of machine learning or the artificial neural network is trained using the swarm data as training data, whereby the parameterizable processing chain is parameterized by the training.

[0023] Training is preferably carried out using supervised learning. However, it would also be possible to train the longitudinal guidance model or the artificial neural network using unsupervised learning, reinforcement learning, or stochastic learning. The longitudinal guidance model preferably serves to generate speed profiles for longitudinal dynamics control. The longitudinal guidance model preferably outputs at least one longitudinal guidance variable and preferably a plurality of longitudinal guidance variables as output variables, by means of which a speed profile and / or an acceleration profile and / or a longitudinal guidance along the route section corresponding to the route data supplied as input variables can be derived and / or determined.

[0024] In other words, a (trained) longitudinal guidance model is generated which, when route data characteristic of a route section (of the vehicle) being traveled and / or intended is supplied as input variable(s), outputs longitudinal guidance variable(s) characteristic of a longitudinal guidance (such as a speed and / or acceleration curve) of the vehicle as output variable(s).

[0025] The (trained and / or trainable) longitudinal guidance model is a (computer-implemented and / or machine-readable) model that can be executed by a processor device. In particular, the (trained) longitudinal guidance model maps the input variable(s) to the output variable(s) (in particular, unambiguously).

[0026] It is conceivable that, for example, acceleration values ​​and / or speed values ​​and / or (relative) position information of the vehicle for at least three, preferably at least five, and preferably at least ten future points in time, or variables characteristic thereof, are specified (as output variables). Preferably, acceleration values ​​and / or speed values ​​and / or (relative) position information of the vehicle for a maximum of 20, preferably a maximum of 10, preferably a maximum of five future points in time, or variables characteristic thereof, are specified (as output variables).

[0027] Preferably, the learned or trained longitudinal guidance model (i.e. after completion of the training process) determines a speed profile and / or an acceleration profile and / or a (planned) temporal longitudinal guidance course of the vehicle (for the route data supplied to the longitudinal guidance model) on the basis of the (output) output variable(s) or the (output) longitudinal guidance variable(s).

[0028] Preferably, the learned or trained longitudinal guidance model is stored in a storage device of a longitudinally guiding vehicle, so that a control device of the vehicle for longitudinal guidance of the vehicle can carry out the longitudinal guidance of the vehicle by means of the (learned or trained) longitudinal guidance model.

[0029] In order to offer a more cost-effective alternative, the proposed method advantageously learns speed profiles using trajectories and a ground truth and calculates them online in the vehicle.

[0030] Compared to the invention, the known path planning algorithms are significantly more complex to develop and require greater resources overall. Advantageously, the proposed method makes it possible to implement complex longitudinal vehicle guidance based on a large amount of swarm data without simplifying the underlying model assumptions or optimizing the implementation on the target control unit.

[0031] Learning speed profiles based on maps aggregated from the vehicle fleet using artificial intelligence algorithms differs significantly from conventional solutions.

[0032] The driving history data preferably consists of driving dynamics variables, for example movement data and / or yaw rate of the vehicle.

[0033] It is conceivable that the swarm data could be used for training in sections. It is conceivable that the swarm data could be further subdivided (into data sets) depending on geometric parameters of the respective route section and used to train the longitudinal guidance model. Geometric parameters could include the length of a route section, a curve curvature, a turning point of a curve curvature, a road junction and / or a road junction, and the like, as well as combinations thereof.

[0034] It is also conceivable that the swarm data could be preprocessed using a pattern recognition algorithm, and the presence of patterns of longitudinal vehicle movement could be determined for different route sections depending on the course of the route sections. If patterns are recognized, the swarm data could be further subdivided depending on the recognized patterns (and a training process could be carried out based on these). It is also conceivable that the swarm data could be preprocessed before being fed into the system, and averages could be calculated from the swarm data for similar route sections, or that the swarm data could be checked for a meaningful pattern and possibly excluded (for example, in the event of an accident).

[0035] In a preferred method, the route data is map data, in particular navigation maps, which were retrieved, for example, from an (external) database.

[0036] In a further preferred method, the route data are route data determined by the respective swarm vehicles of the vehicle swarm. The route data are preferably acquired and / or generated by at least one (vehicle-mounted) sensor device of the swarm vehicle. For example, it is conceivable that the route data are acquired and / or generated by measuring the steering angle and / or detecting the lateral guidance of the vehicle and / or measuring the (relative) position (and / or detecting the longitudinal guidance) of the vehicle.

[0037] In other words, the route data preferably consists of the trajectories of the swarm vehicles.

[0038] In a further preferred method, the swarm data is map data that, depending on a position on the map, indicates at least one longitudinal guidance variable characteristic of a longitudinal guidance profile. Preferably, a map is accessed on which a speed and / or acceleration of the vehicle swarm or vehicle fleet is assigned to one (in particular each) position on the map.

[0039] In a further preferred method, the longitudinal guidance model comprises at least one environmental variable as at least one input variable. The environmental variable can be, for example, a variable characteristic of a weather condition, slipperiness, temperature, or wind. This can be retrieved (online) from a database or measured using a (vehicle-mounted) sensor device (of the vehicle). This can provide weather-dependent longitudinal guidance. In a further preferred method, the longitudinal guidance model comprises environmental variables characteristic of the vehicle's surroundings as input variables. Preferably, the at least one and preferably the plurality of environmental variables are detected by an environment detection device for detecting the vehicle's surroundings of the (swarm) vehicle.

[0040] Preferably, the environment detection device for detecting a vehicle environment (of a (swarm) vehicle) is selected from a group of sensors, which includes a color camera, a front camera, a rear camera, an infrared camera, LIDAR (abbreviation for light detection and ranging or light imaging, detection and ranging), radar, ultrasonic sensors and the like, as well as combinations thereof. Preferably, the environment detection device for detecting a vehicle environment generates spatially resolved (in particular 2D and / or 3D) sensor data (from a vehicle environment of the respective (swarm) vehicle).

[0041] In a further preferred method, the surroundings variable and / or the environmental variable are received (from the swarm vehicle and / or the vehicle) via a digital communication connection and / or via a wireless (in particular private and / or public) network, preferably via Car2Car communication, C2C communication, and / or Car2x communication, and / or Car2l communication. The network can be a local network, such as a WLAN network.

[0042] The environmental variable can, for example, be a variable that is characteristic of the traffic and / or surrounding vehicles and / or construction site and / or route-related features.

[0043] In a further preferred method, swarm data selected depending on a given vehicle, preferably depending on the vehicle type, and / or user-specifically, is received. It is conceivable that only swarm data from a specific vehicle type is used to generate the longitudinal guidance model. This advantageously allows for the generation of longitudinal guidance variables or speed profiles specifically adapted to the vehicle type.

[0044] It is also conceivable that, during the training of the longitudinal guidance model, swarm data generated and / or collected by the user (of the vehicle for which the longitudinal guidance model is to be generated)—for example, using another vehicle—may be taken into account (in a special way, e.g., through weighting). This allows for user-specific driving habits to be advantageously taken into account.

[0045] The present invention is further directed to a method for the vehicle-internal determination, in particular real-time determination, of at least one longitudinal control variable for the longitudinal guidance of a vehicle along an intended route of the vehicle as a function of route data characteristic of the course of the intended route.

[0046] According to the invention, the at least one longitudinal control variable is determined by means of the longitudinal control model generated according to one of the methods described above according to one embodiment.

[0047] By providing the trajectory from the lateral guidance, speed profiles can then be calculated online in the vehicle using the trained algorithm.

[0048] The advantages are that the invention can generate speed profiles without a connection to the backend. This eliminates the need for complex and expensive hardware and software, replacing them with a software module.

[0049] In a preferred method, the external storage device and / or the longitudinal guidance profile determination device is a particularly cloud-based and / or an external server. An external server is understood to mean, in particular, a server external to the (swarm) vehicle, in particular a backend server. The external server is, for example, a backend of a vehicle manufacturer or a service provider, which is configured to evaluate and / or store (determined) map data and / or swarm data and / or data derived therefrom (in particular in each case relating to traffic situations with regard to a plurality of (swarm) vehicles). The functions of the backend or the external server can be carried out on (external) server farms. The (external) server can be a distributed system. The external server and / or the backend can be cloud-based.

[0050] The present invention is further directed to a, in particular processor-based, longitudinal guidance profile determination device for vehicles for generating a longitudinal guidance model for a vehicle for determining, in particular internally within the vehicle, at least one longitudinal guidance variable for the longitudinal guidance of the vehicle along an intended route.

[0051] The longitudinal guidance profile determination device is suitable and intended to generate or generates the longitudinal guidance model on the basis of swarm data determined by means of a swarm of vehicles and stored on an external storage device.

[0052] The swarm data comprise driving course data which are characteristic of at least partial longitudinal guidance, and preferably transverse guidance, of the respective swarm vehicle along a route section, and route data respectively assigned to the driving course data which indicate the course of the respective route section and preferably its (relative and / or absolute) position.

[0053] According to the invention, the longitudinal guidance profile determination device is suitable and intended to train a provided, trainable longitudinal guidance model of machine learning, which comprises a set of parameters for mapping at least one input variable of the longitudinal guidance model to at least one output variable, which can be set to values ​​learned as a result of the training process within the scope of a training process, with the swarm data received from the external storage device by supplying the route data as input variables and comparing the at least one output variable with at least one longitudinal guidance variable determined on the basis of the driving course data associated with the route data and characteristic of the longitudinal guidance.

[0054] It is therefore also proposed within the scope of the invention that a longitudinal guidance model is provided by means of which speed profiles can be calculated online in the vehicle by providing the trajectory, for example from the lateral guidance and / or navigation data, using the trained algorithm.

[0055] Preferably, the longitudinal profile determination device is configured, suitable, and / or intended to carry out the method described above as well as all method steps already described above in connection with the method, individually or in combination with one another. Conversely, the method can be equipped with all features described in the context of the longitudinal profile determination device, individually or in combination with one another. The present invention is further directed to a vehicle, in particular a motor vehicle, comprising a longitudinal profile determination device for a vehicle as described above, according to one embodiment. The vehicle can, in particular, be a (motorized) road vehicle.

[0056] A vehicle can be a motor vehicle, in particular a motor vehicle controlled by the driver himself (“driver only”), a semi-autonomous, autonomous (for example, autonomy level 3, 4 or 5 (of the SAE J3016 standard)) or self-driving motor vehicle. Autonomy level 5 refers to fully automated vehicles. The vehicle can also be a driverless transport system. The vehicle can be controlled by a driver or drive autonomously. In addition to a road vehicle, the vehicle can also be an air taxi, an aircraft and another means of transport or another type of vehicle, for example an aircraft, watercraft or rail vehicle.

[0057] The present invention is further directed to a computer program or computer program product comprising program means, in particular a program code, which represents or encodes at least some and preferably all of the method steps of the method according to the invention and preferably one of the described preferred embodiments and is designed to be executed by a processor device.

[0058] The present invention is further directed to a data memory on which at least one embodiment of the computer program according to the invention or a preferred embodiment of the computer program is stored.

[0059] Further advantages and embodiments can be seen from the attached drawings:

[0060] Showing:

[0061] Fig. 1 is a schematic representation of a prior art method for providing aggregated speed profiles;

[0062] Fig. 2 is a schematic representation of a method according to the invention according to a preferred embodiment; Fig. 3a is a schematic representation of a Kl learning phase; and

[0063] Fig. 3b is a schematic representation of a Kl design.

[0064] Fig. 1 shows a schematic representation of a method for providing aggregated speed profiles from the prior art, on which a method according to the invention is to be based according to one embodiment.

[0065] Fig. 1 illustrates that speed profiles and the associated trajectories of the vehicle fleet traveled by vehicles 1 of a vehicle fleet 3 or a vehicle swarm 3 are uploaded to a backend B (and stored in an external storage device 100). There, the data are compiled into a consistent map K, resulting in trajectory profiles with an associated speed profile based on the fleet data. These maps are then made available to the vehicle fleet, allowing them to locate themselves on the map and extract the relevant speed profile.

[0066] Based on map K, a vehicle can be localized within the map and the speed profiles in the vehicle fleet can be controlled longitudinally.

[0067] Fig. 2 shows a preferred embodiment of a method according to the invention based on a further development of the method shown in Fig. 1. Identical reference numerals indicate identical or similarly functioning elements.

[0068] According to a preferred embodiment of the method according to the invention, the collected data used to create the maps K are used to train an algorithm 10 from the field of artificial intelligence such that, upon input of a trajectory, e.g., from the lateral guidance, it outputs a plausible speed profile to the longitudinal guidance. For this purpose, the generated maps K preferably serve as input (trajectories) as well as ground truth (speed profiles) for the algorithm, which could, for example, employ methods of "supervised learning." The trained algorithm can then be integrated back into vehicles. However, these vehicles then do not require a connection to the backend or generated maps.

[0069] In Fig. 2, reference symbol A denotes the integration of the trained algorithm into the vehicles. Fig. 3a shows a schematic representation of a Kl training phase. Reference symbol 2 denotes the speed profiles (as driving history data) (particularly those acquired by a driver or by the shared vehicles 1 of a vehicle fleet 3).

[0070] The reference symbol T denotes the (associated) trajectories. Route data indicating the course of the respective trajectory T are fed as input variables to the (Kl-based) machine learning longitudinal guidance model 10 for learning speed profiles for given trajectories T.

[0071] Here, the longitudinal guidance model 10 outputs a speed profile as an output variable. Reference numeral 4 denotes a comparison of the speed profile output by the longitudinal guidance model 10 with the speed profile 2 (used as ground truth) (and retracted) of the swarm vehicle 1. Any resulting differences are fed to the longitudinal guidance model 10 for learning (e.g., through so-called backpropagation).

[0072] Fig. 3b shows a schematic representation of a Kl version in which the (fully) trained (artificial intelligence-based) longitudinal guidance model outputs a (trained) speed profile 6 as a function of an (intended) trajectory S or route data characteristic of its course.

[0073] The applicant reserves the right to claim all features disclosed in the application documents as essential to the invention, provided they are novel, individually or in combination, over the prior art. It is further noted that the individual figures also describe features that may be advantageous in themselves. The skilled person will immediately recognize that a particular feature described in a figure may be advantageous even without the adoption of further features from that figure. Furthermore, the skilled person will recognize that advantages may also arise from a combination of several features shown in individual or different figures. List of reference symbols

[0074] 1 swarm vehicle

[0075] 2 Speed ​​profile

[0076] 3 Vehicle swarm or vehicle fleet

[0077] 4 Comparison of output speed profile with speed profile 2

[0078] 6 Learned speed profile 0 External storage device

[0079] K Generated map with trajectories and associated speed profiles

[0080] B Backend

[0081] 5 Route data

[0082] A Integration of the trained algorithm into the vehicles

[0083] TT rajectory

Claims

Patent claims 1. A method for generating a longitudinal guidance model for a vehicle for determining, in particular internally within the vehicle, at least one longitudinal guidance variable for the longitudinal guidance of the vehicle (1) along an intended route (S), wherein the longitudinal guidance model is generated on the basis of swarm data determined by means of a vehicle swarm (3) and stored on an external storage device (100), wherein the swarm data comprise driving course data which are characteristic of at least partial longitudinal guidance, and preferably transverse guidance, of the respective swarm vehicle (1) along a route section, and route data respectively assigned to the driving course data, which indicate the course of the respective route section and preferably its position, comprising: Providing a trainable machine learning longitudinal guidance model comprising a set of parameters for mapping at least one input variable of the longitudinal guidance model to at least one output variable, which parameters can be set during a training process to values ​​learned as a result of the training process; Training the longitudinal guidance model with the swarm data received from the external storage device by supplying the route data as input variables and comparing the at least one output variable with at least one longitudinal guidance variable determined on the basis of the driving course data associated with the route data and characteristic of the longitudinal guidance.

2. Method according to claim 1, characterized in that the route data is map data.

3. Method according to one of the preceding claims, characterized in that the route data are route data determined by the respective swarm vehicles (1) of the vehicle swarm (3).

4. Method according to one of the preceding claims, characterized in that the swarm data is map data which, depending on a position on the map, indicates at least one longitudinal guidance variable characteristic of a longitudinal guidance profile.

5. Method according to one of the preceding claims, characterized in that the longitudinal guidance model comprises at least one environmental variable as at least one input variable.

6. Method according to one of the preceding claims, characterized in that the longitudinal guidance model comprises environmental variables characteristic of the vehicle environment as input variables.

7. Method according to one of the preceding claims, characterized in that swarm data selected as a function of a given vehicle and / or user-specific data are received.

8. Method for the vehicle-internal determination, in particular real-time determination, of at least one longitudinal control variable (6) for the longitudinal guidance of a vehicle (1) along an intended route of the vehicle (1) as a function of route data (S) characteristic of the course of the intended route, characterized in that the at least one longitudinal control variable (6) is determined by means of the longitudinal guidance model generated according to one of the preceding claims.

9. Longitudinal guidance profile determination device (10) for vehicles (1) for generating a longitudinal guidance model for a vehicle for, in particular, vehicle-internal determination of at least one longitudinal guidance variable for the longitudinal guidance of the vehicle (1) along an intended route (S), wherein the longitudinal guidance model is generated on the basis of swarm data determined by means of a vehicle swarm (3) and stored on an external storage device (100), wherein the swarm data comprise driving course data which are characteristic of at least section-wise longitudinal guidance, and preferably transverse guidance, of the respective swarm vehicle (1) along a route section, and route data respectively assigned to the driving course data, which indicate the course of the respective route section and preferably its position, characterized in that the longitudinal guidance profile determination device is suitable and intended to provide atrainable longitudinal guidance model of machine learning, which comprises a set of parameters for mapping at least one input variable of the longitudinal guidance model to at least one output variable, which in, Within the framework of a training process, they can be set to values ​​which are learned as a result of the training process, to train with the swarm data received from the external storage device by supplying the route data as input variables and comparing the at least one output variable with at least one longitudinal control variable determined on the basis of the driving course data assigned to the route data and which is characteristic of the longitudinal control.

10. Vehicle, in particular motor vehicle, comprising a longitudinal guide profile generating device (10) according to the preceding claim.

Images