DE102018006747A1 - Method for evaluating data from a vehicle - Google Patents

Abstract

The invention relates to a method for evaluating data (7) by means of an external data processing device (2), wherein in the vehicle (1) data (7) is detected, encrypted and transmitted to the external data processing device (2) via at least one sensor (4) become. The inventive method is characterized in that the acquired data (7) are homomorphically encrypted, after which the encrypted data (7) in the external data processing device (2) are evaluated in the encrypted state to an encrypted evaluation result, after which the encrypted evaluation result to the vehicle ( 1) is transmitted back and decrypted in the vehicle (1).

Description

The invention relates to a method for evaluating data from a vehicle by means of an external data processing device, the data being recorded, encrypted and transmitted to the external data processing device in the vehicle via at least one sensor.

The sensitivity of data collected in the vehicle is known in principle from the general state of the art. Therefore, an attempt is made to evaluate the data primarily within the vehicle, which, depending on the data, however, may be expensive. An evaluation in an external data processing device, for example a cloud, would often be much simpler, in particular if neuronal networks could be used for the evaluation.

In order to protect the sensitive data at least during the transmission accordingly, the US 9,779,557 B2 a system and method in which the data is collected in the vehicle, encrypted there and transmitted to a cloud or the like. The data is then decrypted in the cloud in order to be able to evaluate it, after which a re-encryption and retransmission of the data to the vehicle, if necessary, takes place.

The case is relatively complex, since a multiple encryption and decryption is necessary. Moreover, in the cloud, the data is unencrypted, allowing for potential misuse, and allowing the cloud operator access to potentially sensitive data of the vehicle, vehicle owner, or the like, so there is a special privacy-preserving agreement between the operator of the cloud and each individual driver must be closed, which is relatively complex in the matter and in particular in the documentation.

Furthermore, situations may arise in which the cloud has knowledge of data which, for example, indicate a safety-critical situation or the like. Then the operator of the cloud would be obliged accordingly to inform the driver here, in order to exclude potential liability claims. Again, this is extremely expensive or hardly possible in practice.

The object of the present invention is now to further improve a process as mentioned in the cited US-A-publication as a generic prior art and in particular to avoid or at least minimize the disadvantages described above.

This object is achieved by a method having the features in claim 1, and in particular in the characterizing part of claim 1. Advantageous embodiments and further developments emerge from the subclaims dependent thereon.

The method according to the invention acquires the data in the vehicle via at least one sensor as a technical device and then encrypts these data within the vehicle in an encryption device before they are transmitted to the external data processing device, comparable to the generic state of the art. The difference according to the invention consists in the fact that the data recorded via the sensors are homomorphically encrypted in the encryption device before they are evaluated in the external data processing device in a further encrypted state to form an encrypted result. Homomorphic encryption allows just that. The data is encrypted securely and reliably. Nevertheless, they can be evaluated via suitable procedures in the encrypted state to a then also encrypted result. The entire processing of the data in the external data processing device is thus encrypted. Thus, neither the operator of the external data processing device nor the vehicle manufacturer or anyone else has access to the encrypted data, neither the raw data, that is, the data captured by the sensors and transmitted encrypted from the vehicle, nor to the data during the Evaluation, and also not to the evaluation result, since all this is encrypted and can only be decrypted with the appropriate homomorphic key of the vehicle. The encrypted evaluation result is then transmitted back to the vehicle and decrypted in the vehicle in a technical decryption device that may be integrated with the encryption device.

This ensures that there is no unencrypted data outside the vehicle at any time. Thus, the data and the privacy of the person using the vehicle are ideally protected and the right to informational self-determination by the person using the vehicle is ensured safely and reliably. Nevertheless, it is possible to have optimized evaluation processes carried out by external service providers in high-performance external data processing devices, namely, as already mentioned, without having to go through expensive and costly data protection processes and without the risk of transmission of data or evaluation results by the operator of the external Data processing device exists. Also in If unauthorized access to this, for example, a so-called hacker attack, the data would be encrypted accordingly and would reveal to the attacker no sensitive information, as long as he does not have the same time in the vehicle counter key homomorphic encryption available.

An advantageous development may provide that the sensor system for capturing the data and the encryption devices for homomorphic encryption of the data and homomorphic decrypting the transmitted encrypted evaluation results within the vehicle is securely encapsulated, so as to access within the vehicle, for example, with the Vehicle connected external devices, such as a smartphone integrated via Bluetooth, which is coupled to a telephone or media system of the vehicle or the like, to avoid or at least significantly complicate.

1. [1] Xie et al (2014), Crypto-Nets: Neural Networks over Encrypted Data, arXiv:1412.6181v2 [cs.LG] https://arxiv.org/abs/1412.6181
2. [2] Dowlin et al. (2016) CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy, Proceedings of the 33rd International Conference on Machine Learning, New York, NY, USA, 2016. JMLR: W&CP volume 48. http://proceedings.mlr.press/v48/gilad-bachrach16.pdf
3. [3] Hesamifard et al. (2017) CryptoDL: Deep Neural Networks over Encrypted Data, arXiv: 1711.05189v1 [cs.CR] https://arxiv.org/abs/1711.05189.

An extraordinarily favorable development of the method according to the invention provides that the evaluation in the external data processing device takes place at least partially by models which are created or optimized on the basis of machine learning, in particular by means of neural networks. The combination of homomorphic encryption and machine learning approaches, for example through so-called Convolutional Neural Networks, is currently a current focus of research, in particular of providers of so-called clouds. In this context, reference may be made to the following articles:

1. [1] Xie et al (2014), Crypto-Nets: Neural Networks on Encrypted Data, arXiv: 1412.6181v2 [cs.LG] https://arxiv.org/abs/1412.6181
2. [2] Dowlin et al. (2016) CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy, Proceedings of the 33rd International Conference on Machine Learning, New York, NY, USA, 2016. JMLR: W & CP volume 48. http: //proceedings.mlr. press / v48 / gilad-bachrach16.pdf
3. [3] Hesamifard et al. (2017) CryptoDL: Deep Neural Networks on Encrypted Data, arXiv: 1711.05189v1 [cs.CR] https://arxiv.org/abs/1711.05189.

Such a machine learning model, e.g. The use of neural networks to evaluate the homomorphically coded technical data is a very expedient approach to making the evaluation of the data efficient without having to decrypt the data itself or even the evaluation result. The models are trained in advance and in particular also continuously with unencrypted pattern data, as set out in particular in [1].

• - die Fahrzeuggeschwindigkeit;
• - die Fahrzeugbeschleunigung;
• - Verkehrszeichen oder andere Objekte in der Fahrzeugumgebung;
• - die geographische Position des Fahrzeugs;
• - die aktuelle Zeit; und/oder
• - Spracheingaben von das Fahrzeug nutzenden Personen und/oder daraus abgeleitete Befehle;

umfassen. Solche Daten, welche zur Steuerung des Fahrzeugs, beispielsweise zur Steuerung von Komfortfunktionen oder auch zur Steuerung eines autonomen oder teilautonomen Fahrbetriebs, zur Steuerung von Fahrerassistenzsystemen und dergleichen, dienen können, lassen sich nun also in dem Fahrzeug entsprechend erfassen und gemeinsam mit der Sensorik und der Erfassung gekapselt innerhalb des Fahrzeugs verschlüsseln. Sie können dann verschlüsselt an die externe Datenverarbeitungseinrichtung übertragen und dort verschlüsselt ausgewertet werden. Die Auswertung kann, insbesondere beim Einsatz von Modellen des maschinellen Lernens, wie z.B. neuronalen Netzen, sehr effizient und qualitativ hochwertig in sehr kurzer Zeit erfolgen, da über die externe Datenverarbeitungseinrichtung eine weitaus höhere Rechenleistung bereitgestellt werden kann, als in dem Fahrzeug selbst. Die Daten können dann immer noch verschlüsselt an das Fahrzeug zurück übertragen werden, und können dort entsprechend eingesetzt werden, um Funktionen des Fahrzeugs zu steuern, sei es Fahrfunktionen im Rahmen eines autonomen Fahrens, Funktionen zur Navigation des Fahrzeugs oder auch Einstellungen zur Erhöhung des Komforts der Insassen des Fahrzeugs oder dergleichen.A very favorable development of the idea, it provides that the data from internal and / or external sensors of the vehicle are detected and among others

• the vehicle speed;
• - the vehicle acceleration;
• - traffic signs or other objects in the vehicle environment;
• - the geographical position of the vehicle;
• - the current time; and or
• - voice input from persons using the vehicle and / or commands derived therefrom;

include. Such data, which can be used to control the vehicle, for example to control comfort functions or to control autonomous or semi-autonomous driving, for controlling driver assistance systems and the like, can now be detected accordingly in the vehicle and together with the sensors and the Encrypt capture encapsulated within the vehicle. They can then be transmitted encrypted to the external data processing device and evaluated there encrypted. The evaluation can be very efficient and of high quality in a very short time, especially when using models of machine learning, such as neural networks, since a much higher computing power can be provided via the external data processing device than in the vehicle itself. The data can then still be transmitted encrypted back to the vehicle, and can be used there accordingly to control functions of the vehicle, be it driving functions in the context of autonomous driving, functions for navigating the vehicle or also settings to increase the comfort of the occupants of Vehicle or the like.

• - Herzfrequenz;
• - den Blutdruck, Hirnströme;
• - Schwitzen, die Stimmlage, den Sprachfluss und/oder daraus abgeleitete Daten über den physischen und emotionalen Zustand der in dem Fahrzeug befindlichen Personen;

beinhalten können. Insbesondere diese Art von Daten ist hinsichtlich des Datenschutzes relativ brisant, sodass hier die Ausgestaltung der Erfindung, bei welcher die Daten für Dritte eben nicht unverschlüsselt zugänglich werden, von besonderem Vorteil sind. Anhand der medizinischen Daten lässt sich das Befinden von in dem Fahrzeug befindlichen Personen, insbesondere von einer das Fahrzeug fahrenden Person, sehr gut erfassen. Dies kann beispielsweise dabei helfen, sicherheitskritische Situationen frühzeitig zu erkennen und durch einen gezielten Eingriff in technische Funktionen des Fahrzeugs, beispielsweise ein Abbremsen oder dergleichen, frühzeitig entgegen zu wirken. Gleichzeitig ist es dabei so, dass diese Art von Daten nicht nur hinsichtlich des Datenschutzes hochrelevant ist, sondern auch hinsichtlich der Verantwortung des Betreibers der externen Datenverarbeitungseinrichtung. Würde dieser beispielsweise erkennen, dass eine das Fahrzeug fahrende Person von akuten gesundheitlichen Problemen betroffen ist, müsste er von sich aus eingreifen, um eventuellen Haftungsansprüchen entgegen zu wirken. Dadurch, dass er die Inhalte der auf seiner Datenverarbeitungseinrichtung ausgewerteten Daten jedoch nicht kennt, weil diese verschlüsselt dort anlangen, verschlüsselt ausgewertet und die Auswerteergebnisse verschlüsselt zurückgesandt werden, ohne dass eine Verschlüsselung in der externen Datenverarbeitungseinrichtung erfolgt, entfallen hier Haftungsprobleme, was die sinnvolle Nutzanwendung von derartigen medizinisch relevanten Daten von Personen in dem Fahrzeug deutlich erleichtert.A further very advantageous embodiment of the method provides, as an alternative or in particular, in addition thereto, that the data, which are recorded by the internal and external sensors of the vehicle, comprise medical data of persons in the vehicle, in particular of a person driving the vehicle, which among other things the

• - Heart rate;
• - the blood pressure, brain waves;
• - sweating, the voice, the speech flow and / or derived data on the physical and emotional state of the persons in the vehicle;

may include. In particular, this type of data is relatively explosive in terms of data protection, so that here the embodiment of the invention in which the data for third parties just not accessible unencrypted, are of particular advantage. On the basis of the medical data, the condition of persons in the vehicle, in particular of a person driving the vehicle, can be detected very well. This can help, for example, to recognize safety-critical situations at an early stage and counteract early by a targeted intervention in technical functions of the vehicle, such as a slowdown or the like. At the same time, this type of data is highly relevant not only in terms of data protection, but also in terms of the responsibility of the operator of the external data processing device. For example, if this person recognized that a person driving the vehicle was affected by acute health problems, he would have to intervene on his own initiative in order to counteract possible liability claims. Due to the fact that he does not know the contents of the data evaluated on his data processing device, because these arrive there encrypted, encrypted evaluated and the evaluation results are encrypted sent back, without encryption in the external data processing device, this accounts for liability problems, resulting in the useful application of significantly facilitates such medically relevant data of persons in the vehicle.

An advantageous development of the idea provides that the encrypted data for creating a driving profile and for determining an encrypted classification of the driving profile from the data on the external data processing device are used .. On this basis, for example, individual insurance rates based on the individual driving behavior and a technically record detailed risk assessment. This is in principle already possible today and is typically realized by the installation of a telematics box or by means of the sensors contained in a mobile phone. The embodiment of the method according to the invention described would now make it possible to operate without such devices and without disclosing detailed data on driving behavior. For this purpose, the data or information collected by the sensors can be transmitted in encrypted form to an external data processing device, for example the backend server of a vehicle manufacturer. On the basis of the evaluated data, a classification can then be made for the respective driver, who, for example, fundamentally classifies the driver profile, but without knowing the data. Thus, a corresponding service to insurance providers is possible. The person driving the vehicle still retains complete control over the data collected by the sensors and the evaluation result. Only one classification is determined on the basis of the evaluation of the homomorphically encrypted data, for example, by adapted neural networks and can accordingly be made available as a driver profile. The driver, who then can decrypt this data in the vehicle and see it as the only one, can then pass on the result of the service thus provided, for example, to his insurance company as the basis for an individual insurance tariff. Encrypted test data make it possible to independently check the evaluation model without having to disclose the actual evaluation algorithm. It is also the case that a corresponding method for the vehicle manufacturer as a service provider in an advantageous manner is easy and inexpensive to implement, since he has no access to personal data, because they remain completely encrypted, and so the requirements for data security and associated costs can be kept low.

A further advantageous embodiment of the idea provides that the encrypted data is used to determine emotions of persons using the vehicle. This can be realized according to a very advantageous development of the idea, especially via an evaluation by neural networks. The data from one or more sensors, such as an indoor camera, vital sensors or microphones, may be evaluated by a machine learning method, preferably in neural networks, to capture the emotions of a person using the vehicle, such as a person driving the vehicle determine. The technical solution described above with the encrypted transmission and evaluation of the data and the evaluation result makes it possible to do all this without the data itself being accessible to third parties. The evaluation process, which can process the homomorphically encrypted data, for example by means of adapted neural networks, then takes place in the external data processing device on the encrypted data without access to the raw data. Also the result of the evaluation is present in the external data processing device only in encrypted form and is transmitted back to the vehicle. Only there in the ideally encapsulated system, the evaluation result can be decoded. The advantages for the person using the vehicle are that no personal data collected by the sensors, such as the facial expression, the heart rate, a conversation and none of the evaluation results, for example emotions or conversation content, are shared with the operator of the external data processing device and Nevertheless, the evaluation results obtained can nevertheless be used for controlling, for example, comfort functions and / or voice inputs in the vehicle or else for the selective transmission of evaluation results, for example, into social networks. For the operator of the external data processing device, the method also has advantages. The evaluation model of machine learning is located in the data center of the operator of the external data processing device and not in the vehicle itself. It is thus very well protected against theft. Since there is no access to personal data, the implementation is, as already mentioned several times, simpler and cheaper possible, as if appropriate regulations of data protection must be observed. Furthermore, here again there is the advantage for the operator of the external data processing device that he has no access to the raw data, such as an irregular heartbeat or the evaluation results, for example, a detected fatigue. In this way, although the driver can warn accordingly, liability claims, if the warning is ignored to the operator of the external data processing device omitted, which in terms of the cost of the operator of the external data processing device, and the willingness to offer such evaluation as a service of decisive advantage.

Another very advantageous embodiment provides that the encrypted data for learning vehicle functions by machine learning and evaluating situations are used for recurring occurrence. About this type of solution according to the method, it is possible to carry out an individualization of the vehicle. In particular, the methods of machine learning can help to individualize the vehicle very much. For example, it is possible that the vehicle responds to certain stimuli, for example the neighbor's cat and / or the neighbor herself, with corresponding reactions. For example, in the case of the appearance of the cat, mop water can be sprayed or honed, for example, in the case of the neighbor's appearance, a window can be shut down. Although the underlying model can not be trained with the encrypted example images, this can be done on the basis of special, e.g. non-parametric models, e.g. based on a "nearest neighbor" classification. About such methods of machine learning, e.g. Nonparametric models, then encrypted example images can be compared with new encrypted images, so as to train the "behavior" of the vehicle and then implement the "behavior" in case of a recurring same situation as an additional comfort function, so that the vehicle reacts accordingly, if comes in a known him by the sensors detected situation.

In this case, it may be necessary for corresponding sensor data to be made available for example as training data or as data for machine learning and transmitted to the external data processing device. There, a model can then be trained with the help of a machine learning method. In the example described, the sensor data can be transmitted in encrypted form and processed in encrypted form in the external data processing device without the operator of the external data processing device having access to the unencrypted data or raw data of the sensors. The advantage with this method for the person using the vehicle is that no personal sensor data or, for example, training images or evaluation results on which the vehicle was trained must be shared with the operator of the external data processing device. For this, again, it has the advantage that it has no access to the personal data of the person using the vehicle and therefore its services easier and cheaper, without complex data protection regulations must be observed, can offer.

A further possibility, which has already been mentioned in part above, provides that in the method according to the invention encrypted evaluation results are used internally after decryption in the vehicle or are transmitted outside the vehicle after approval of the person in the vehicle and / or holding the vehicle , The encrypted and then decrypted in the vehicle evaluation results can therefore be used to control internal vehicle functions, as already indicated above, and are transmitted internally within the vehicle, and that unencrypted to be able to address the sensors directly, and based on the evaluation results to achieve a corresponding reaction of actuators. The decryption must be done only once centrally in the vehicle. It is also possible that, after the consent of the persons involved, data will also be transmitted externally, unencrypted or encrypted with the aim of this To decrypt data at the destination. For example, selected results may be shared by the appropriate individuals, for example, to post emotions in a social network, or transmit corresponding results to an insurance provider to benefit from an individual insurance rate or the like.

Further advantageous embodiments of the method according to the invention will become apparent from the embodiment, which is described below with reference to the figures.

• 1 eine schematische Darstellung einer Einrichtung zur Durchführung des erfindungsgemäßen Verfahrens; und
• 2 ein möglicher Ablauf des erfindungsgemäßen Verfahrens.

Showing:

• 1 a schematic representation of a device for carrying out the method according to the invention; and
• 2 a possible sequence of the method according to the invention.

In the presentation of the 1 is schematically indicated a vehicle 1 to recognize or a secure encapsulated system within the vehicle 1 , Next to the vehicle 1 For example, an external computing device is a cloud or the like formed externally of the vehicle 2 indicated in principle. The vehicle 1 communicates with it via a communication designated 3. In the vehicle 1 there is a sensor system 4 which consists of one or in particular of a plurality of sensors. This sensor technology 4 For example, it may include indoor sensors such as an indoor camera, microphones, vitality sensors, and the like. Likewise, via the sensors 4 Vehicle information such as speed, acceleration, direction of travel and the like are detected. All this is over the arrow 5 , which is the internal collection of data by the sensors 4 symbolized, indicated. Also external information, such as information from the environment detection by cameras, lidar systems, radar systems or the like can be detected by the sensors 4 be made. This is by the arrow 6 indicated accordingly. All this collected data 7 are now unencrypted within the vehicle 1 according to the arrow 8th to an encryption device 9 to hand over. In this encryption device 9 a homomorphic encryption of the data takes place 7 , The homomorphic encryption can be realized as partial, ie either additive or multiplicative, homomorphic encryption as well as fully homomorphic, ie additive and multiplicative, encryption.

The encryption took place via an in-vehicle key 10 is by a closed lock 11 symbolizes while transmitting unencrypted or decrypted data 7 through an open lock 12 is symbolized accordingly. The data transfer between the vehicle 1 and the external data processing device 2 , For example, a cloud, so it is homomorphically encrypted. In the cloud 2 Now an evaluation of the encrypted data takes place 7 in such a way that they are not decrypted beforehand, but that the homomorphic encryption is used to encrypt the encrypted data 7 within the external data processing device 2 to be able to process.

The data are evaluated there, preferably with methods of machine learning and / or using neural networks.

The evaluation is carried out using the homomorphically encrypted data 7 , without these, as already mentioned, in the external data processing device 2 be decrypted. Accordingly, there is also an encrypted evaluation result, which is sent to the vehicle 1 encrypted is transmitted back. In a decryption facility 13 of the vehicle 1 can turn with the key 10 this encrypted evaluation result is decrypted accordingly. The decrypted data 7 can then, as it were through the open lock 12 is indicated, for example, be used over the direction indicated by 14 line to actuators 15 in the vehicle 1 to control and thus control vehicle functions. This can be for vehicle functions for autonomous or semi-autonomous driving, for driver assistance systems as well as vehicle functions for increasing the comfort within the vehicle and for individualizing the vehicle 1 be used. The decrypted data 7 may also be directed to a manual release device via the line labeled 16 17 be directed. There can be a the vehicle 1 the person using the data selectively, in the way they want, so that the unencrypted data 7 according to the arrow 18 be transmitted outside the vehicle. For example, data 7 for the design of an individual insurance tariff to an insurance service provider, or it can be appropriate data 7 be sent to a social network, for example, on the mood of the persons in the vehicle, if desired by the person and released from it.

In the 2 is shown with reference to a flowchart a possible sequence for the inventive method. Starting from the start, the data first 7 through the at least one sensor 4 in the vehicle 1 detected. Thereafter, the homomorphic encryption of the data takes place 7 in the encryption device 9 , All data indicated within the dot-dashed line 7 or processes are then encrypted accordingly, as it is by the closed lock 11 in the presentation of the 2 is indicated accordingly. The encrypted data 7 are sent to the external data processing device 2 transmitted and the external data processing device 2 calculates an encrypted evaluation result. This encrypted evaluation result becomes the vehicle again 1 transmitted back and there in the decryption device 13 with the vehicle's internal key 10 decrypted. The evaluation result decoded in this way is then checked as to whether it should be made available to the control system in-vehicle. In this case, the forwarding to the vehicle functions takes place so that the in 1 indicated actuators 15 can react accordingly. If this is not desired, the process ends at this point. In parallel, it will be evaluated whether the evaluation result is to be made available outside the vehicle, as already mentioned, for example for insurance service providers, social networks or the like. If this is the case, then it is forwarded from the vehicle 1 after a corresponding release of the data owner. If this is not the case, the procedure is ended and begins for the next recorded data 7 then again.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 9779557 B2 [0003]

Cited non-patent literature

• Xie et al (2014), Crypto-Nets: Neural Networks on Encrypted Data, arXiv: 1412.6181v2 [cs.LG] https://arxiv.org/abs/1412.6181 [0011]
• Dowlin et al. (2016) CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy, Proceedings of the 33rd International Conference on Machine Learning, New York, NY, USA, 2016. JMLR: W & CP volume 48. http: //proceedings.mlr. press / v48 / gilad-bachrach16.pdf [0011]
• Hesamifard et al. (2017) CryptoDL: Deep Neural Networks on Encrypted Data, arXiv: 1711.05189v1 [cs.CR] https://arxiv.org/abs/1711.05189. [0011]

Claims (10)

A method for analysis of data (7) by means of an external information processing device (2), wherein in the vehicle (1) via at least one sensor (4) data (7) is detected, encrypted and transmitted to the external data processing device (2), characterized in that the recorded data (7) are homomorphically encrypted, after which the encrypted data (7) in the external data processing device (2) are evaluated in the encrypted state to an encrypted evaluation result, after which the encrypted evaluation result is transmitted back to the vehicle (1) and in the vehicle (1) is decrypted. Method according to Claim 1 , characterized in that the evaluation in the external data processing device (2) takes place at least partially by models which are created or optimized on the basis of machine learning, in particular by means of neural networks. Method according to Claim 1 or 2 , characterized in that the data (7) are detected by internal and / or external sensors (4) of the vehicle (1), and inter alia - the vehicle speed - the vehicle acceleration - traffic signs or other objects in the vehicle environment - the geographical position of the vehicle Vehicle - the current time; and / or - include voice input from persons using the vehicle and / or commands derived therefrom. Method according to Claim 1 . 2 or 3 , characterized in that the data (7) medical data (7) of persons in the vehicle (1), in particular a person driving the vehicle (1) include, inter alia - heart rate - the blood pressure, brain waves - sweating, the voice , the speech flow and / or data derived therefrom on the physical and emotional state of the persons in the vehicle. Method according to one of Claims 1 to 4 , characterized in that the encrypted data (7) for creating a driving profile and for determining an encrypted classification of the driving profile from the data (7) on the external data processing device (2) are used. Method according to one of Claims 1 to 5 , characterized in that the encrypted data (7) are used for determining emotions of persons using the vehicle (1). Method according to Claim 6 , characterized in that in determining the emotions neural networks in the external data processing device (2) are used. Method according to one of Claims 1 to 7 , characterized in that the encrypted data (7) are used for learning vehicle functions by machine learning and by evaluating situations for recurring occurrence. Method according to one of Claims 1 to 8th , characterized in that the encrypted evaluation result after decryption in the vehicle (1) in-vehicle or after approval of the person in the vehicle (1) and / or persons holding the vehicle (1) is transmitted outside the vehicle. Method according to one of Claims 1 to 9 , characterized in that the at least one sensor (4) and the encryption device (9) and the decoding device (13) are securely encapsulated in the vehicle (1).

Images