METHOD, DEVICE FOR DATA PROCESSING AND SYSTEM

Description

FIELD OF THE INVENTION

[0001] The present invention relates generally and in particular to a method, a device for data processing and a system for detecting manipulation of travel data of a control device of a vehicle. BACKGROUND OF THE INVENTION

[0002] Control devices for vehicles are generally known, particularly for commercial vehicles and trucks. Such control devices are also known as electronic tachographs. By law, such electronic tachographs must be provided in commercial freight transport to monitor driving and break times in trucks with a total weight of over 7.5 t (tonnes). Typically, the recorded journey data can be read out via a communication interface on the electronic tachograph and transferred to a storage medium (e.g., USB stick).

[0003] Common electronic tachographs contain identification cards (driver cards) on which personalized card usage data is stored. The data stored on the driver cards represents, for example, information about the distance traveled during the journey and information about the driver's current status. The driver's status describes, for example, whether the driver is currently driving the truck, performing other work, or taking a break.

[0004] The electronic tachograph is typically connected to an odometer sensor ("KITAS" - Kienzle Sensor). The odometer sensor, for example, uses a Hall sensor to determine the rotations of the drive axle. Furthermore, it can calculate the driving speed and, from this, the distance traveled based on the number of pulses (based on a unit of time).

[0005] It is known that electronic tachographs are additionally connected to the truck's engine control unit via a communication bus (e.g., CAN bus; CAN stands for "Controller Area Network") for further verification. The electronic tachograph receives additional movement information about the truck's current driving speed from the engine control unit. This movement information can be compared with the movement information determined by the distance sensor in a subsequent analysis.

[0006] Various software programs are available for evaluating the driving data from the tachograph and the inserted driver card.

[0007] The journey data from the tachograph and the inserted ticket are typically evaluated with regard to the distance traveled, the driving speed of the last few days, any position data obtained via a connected GNS system (GNS stands for "Global Navigation System"), as well as the entered movement states (e.g. driving, break, work and break) that a driver enters, for example, via a control panel on the tachograph.

[0008] Based on this driving data, evaluations are carried out, for example, regarding the driving and break times that the driver has complied with or not complied with during the last few days.

[0009] Depending on the results of the evaluations, violations of the maximum permitted speed or of the driving and break times are identified by the control authorities (in Germany the BAG ("Federal Office for Goods Transport") and the police) and, if necessary, prosecuted.

[0010] Since the basis for subsequent evaluation regarding compliance with maximum permitted speeds and driving and break times is typically the pulses from the distance sensor (distance pulses; distance signal), which are used to calculate the driving speed and thus the distance traveled, it is known that distance pulses can be manipulated to manipulate driving data and thus complicate investigations into violations of the maximum permitted speed or driving and break time regulations. It is also known that the inserted tickets can be manipulated by making unauthorized copies.

[0011] For several years, various methods have been used to manipulate the distance sensors, such as adding electronic circuits within the distance sensor, blocking the distance signal by short circuiting and / or other electrical means, or manipulating the distance signal by attaching magnets.

[0012] The movement information coming from the engine control unit may also be manipulated; for example, by making changes to the engine control unit itself, using emulators to simulate a speed signal from the engine control unit.

[0013] It is known that the operating software of the control device is capable of detecting such manipulations and storing corresponding error messages in the journey data, which are subsequently read from the tachograph by the control authorities.

[0014] Likewise, if there are discrepancies between the distance signals from the distance sensor and the engine control unit, error messages are stored in the engine control unit and can be read out later.

[0015] However, it is known that, for example, error messages in the engine control unit can be deleted by external devices that are connected via the onboard diagnostic connector (OBD connector) in the vehicle.

[0016] If the engine control unit no longer has any error messages, but the control unit may still have error messages stored, the manipulations can be detected.

[0017] However, the situation is different if the distance signal has been successfully manipulated without the error detection routines in the control unit detecting this manipulation. In this case, the distance traveled / speed is stored in such a way that entries with manipulated movement states (driving and break times) may not be detected during any inspections.

[0018] Logically, the total mileage of the vehicle, i.e. the distance actually travelled, should no longer be correct if the driver enters into the electronic tachograph that he is currently on a break, but the vehicle is actually being moved, contrary to this.

[0019] In such cases, during an inspection by the control authorities, the discrepancy between the mileage on the truck's speedometer ("tacho") and the mileage displayed in the tachograph is noticed.

[0020] However, it is known that system messages about the mileage coming from the engine control unit are intercepted and instead the mileage (distance traveled) of the control unit is sent to the speedometer, so that in the end the control unit and the speedometer show the same values.

[0021] It is also known that the firmware (operating software) of the tachograph and / or the firmware of the truck's control units may be directly tampered with. Once the firmware of the electronic tachograph has been tampered with, any (simulated or fictitious) movement states of the truck can be recorded, without these movement states necessarily having any connection to the actual movement state. Therefore, an inspection and analysis of the downloaded trip data may no longer reveal any tampering.

[0022] The manipulations by changing the firmware in the control device are typically realized by the fact that the movement states (driving and break times) of the truck are no longer entered by the driver, but by the manipulated firmware of the control device, for example by storing corresponding data in the control device.

[0023] The aforementioned manipulation of control units, control devices, and route signals, and the resulting manipulation of driving data, creates a major safety problem, which is why regulatory authorities are closely monitoring compliance with driving and break times for trucks. Driver fatigue can potentially lead to serious accidents.

[0024] The detection of manipulated control devices is therefore based on the public's strong need for security.

[0025] From the published patent application DE 10 2018 201064 A1 a method for monitoring the total distance travelled is known.

[0026] The published patent application DE 10 2019 119784 A1 discloses a method and system for detecting manipulation of a vehicle.

[0027] A method for protecting a vehicle’s mileage against manipulation is known from the published patent application DE 10 2017 209817 A1.

[0028] The object of the present invention is therefore to provide an improved method, an improved device for data processing and an improved system for detecting manipulation of travel data of a control device of a vehicle, as well as a device for transmitting travel data of a control device of a vehicle to detect manipulation of the travel data, in order, for example, to improve the detection of manipulation. SUMMARY OF THE INVENTION

[0029] According to a first aspect, the present invention provides a method according to claim 1.

[0030] According to a second aspect, the present invention provides a data processing apparatus according to claim 8.

[0031] In a third aspect, the present invention provides a system according to claim 9.

[0032] Further aspects and features of the present invention emerge from the dependent claims, the accompanying drawings and the following description of embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Embodiments of the invention will now be described by way of example and with reference to the accompanying drawings in which: Fig. 1 schematically illustrates an embodiment of a system for detecting manipulation of travel data of a control device of a vehicle; Fig. 2 schematically in graphs in Fig. 2A bis Fig. 2D Illustrates embodiments of route sections and movement profiles; Fig. 3 schematically illustrates in a block diagram an embodiment of a training method for a machine learning algorithm for detecting manipulation of travel data of a control device of a vehicle; Fig. 4 schematically illustrates in a block diagram an embodiment of a general-purpose computer; and Fig. 5 schematically illustrates in a flowchart an embodiment of a method for detecting manipulation of travel data of a control device of a vehicle. DESCRIPTION OF EMBODIMENTS

[0034] In Fig. 1 An embodiment in accordance with the present invention is illustrated. Before a detailed description, general explanations of the embodiments follow.

[0035] As mentioned at the beginning in the background of the invention, manipulation of control devices, control units, and distance signals in vehicles is known, particularly in trucks. This involves manipulating the recorded driving data to complicate the investigation of violations of the maximum permissible speed or of driving and break time regulations. This poses a major safety problem, as serious accidents, particularly due to driver fatigue, can potentially occur. Therefore, there is a need for security measures to detect manipulation.

[0036] It was recognized that a plausibility check of the tachograph's trip data based on pattern recognition in various data sources (environmental data, control unit data, etc.) using comparative data from other trips on the same route could reveal tampering with the trip data. In particular, it could be determined whether the trip data has been tampered with, regardless of the underlying type of tampering (e.g., distance signal, tachograph firmware).

[0037] It was also recognized that, based on anonymized case data from a large number of control devices, manipulation of security certificates and manipulation of the mileage display could be revealed.

[0038] Therefore, some embodiments relate to a method for detecting manipulation of travel data of a control device of a vehicle, the method having the features of claim 1.

[0039] According to the invention, the method is applied to driving data from a control device of a truck.

[0040] The procedure can be used on-site, for example during a check (e.g. by stopping) by control authorities, or during a later analysis of stored driving data.

[0041] The method aims to determine whether manipulation of the trip data is likely or has occurred, whereby in some embodiments, determining the exact cause or the exact type of manipulation is not important. In particular, when the method is applied during on-site inspections, some embodiments provide for a classification result to be displayed to the inspection authorities, which is indicative of, for example, existing manipulation, probable manipulation, or no manipulation of the trip data, so that possible further steps can then be initiated by the inspection authorities to determine the type and extent of the manipulation.

[0042] The method is executed in particular on a computer, server, or in the cloud (which includes one or more remote computers). Therefore, some embodiments relate to a data processing device, wherein the data processing device includes an electronic circuit configured to execute the method described herein.

[0043] The electronic circuit of the data processing device can contain one or more processors (e.g., CPU, application processor, graphics processor, etc.), one or more memory elements (e.g., hard disk, RAM, ROM, semiconductor memory, etc.), one or more FPGAs ("Field Programmable Gate Array"), one or more application-specific circuits (ASICs - "Application Specific Integrated Circuit"), and / or typical electronic components configured to carry out the method. The method can be based on computer programs containing a sequence of instructions that, upon execution of the instructions, cause a computer / processor to carry out the method described herein. The method can be based in part on computer programs and in part on electronic circuits.

[0044] The electronic circuit of the data processing device contains a communication interface for exchanging data with other computers, devices, etc., via a network. Data communication can be wired or wireless. The network can be a mobile network, a computer network (e.g., the Internet), etc., and the electronic circuit of the data processing device then contains corresponding hardware interfaces and implements corresponding communication protocols for data exchange.

[0045] In the method described herein, journey data from the tachograph is obtained. Journey data includes, for example, data stored on the tachograph and the driver card(s), such as a driving speed profile (driving speed over time), mileage (distance traveled), and the recorded driving and break times. These journey data therefore represent a movement profile of the vehicle within the recorded period and, therefore, a movement profile of the vehicle on a route traveled by the vehicle. According to the invention, the movement profile is characterized by the driving speed profile and the driving and break times.

[0046] The data processing device can load the travel data from a storage medium or receive it via a communication interface.

[0047] It was recognized that if at least one position of the vehicle is known, which is indicative of at least one section of the route traveled by the vehicle, the vehicle's movement profile can be compared with the movement profile of other vehicles that have traveled the same route section. Based on a plurality of movement profiles of vehicles (of a vehicle class such as trucks) on the route section, a movement profile, in particular a driving speed profile, characteristic of the route section can be determined.

[0048] In some embodiments, a position of the vehicle is indicative of a route section if it is determined within the time period of the recorded trip data. In some embodiments, a position of the vehicle is indicative of a route section if the time of the position determination is shortly before or after the time period of the recorded trip data. From the presence of a position of the vehicle, in some embodiments, on the one hand, a part of the trip data can be associated with the position of the vehicle and, on the other hand, an area surrounding the position can be determined, which is then indicative of a route section of the route traveled by the vehicle (e.g., the position can indicate a gas station on a highway, so that a route section is a section of the highway).

[0049] For example, the position can be determined from toll data or from position data from a vehicle's GPS system, if connected. However, this data is not always available.

[0050] In some embodiments, it is provided to provide the enforcement authorities with a (portable) device that can be used for on-site inspections and features a GPS system for determining their position. The device reads the travel data from the on-site inspection device via a data bus and then transmits its own determined position and the travel data to the data processing device. Since in some embodiments the position is also determined directly on-site during the inspection, it is also indicative of a section of the route traveled by the vehicle, which then corresponds to a section of the route in the vicinity of the checkpoint.

[0051] Therefore, the disclosure relates to a device for transmitting travel data from a control device of a vehicle to detect manipulation of the travel data, the device containing an electronic circuit which is configured to: to read the journey data from the control device, wherein the journey data represent a movement profile of the vehicle on a route travelled; to determine one's own position; and to transmit the read journey data and one's own position to a data processing device via a network, wherein one's own position is indicative of a position of the vehicle and the position of the vehicle is indicative of at least one section of the route travelled.

[0052] The electronic circuit of the device may contain one or more processors (e.g., CPU, application processor, graphics processor, etc.), one or more memory elements (e.g., hard disk, RAM, ROM, semiconductor memory, etc.), one or more FPGAs ("Field Programmable Gate Array"), one or more application-specific circuits (ASICs - "Application Specific Integrated Circuit"), and / or typical electronic components configured to execute the method. The method may be based on computer programs containing a sequence of instructions that, upon execution of the instructions, cause a computer / processor to execute the method described herein. The method may be based in part on computer programs and in part on electronic circuits.The device's electronic circuit contains a communication interface for exchanging data with other computers, devices, mobile communication devices, etc. over a network. Data communication can be wired or wireless. The network can be a cellular network, a computer network (e.g., the Internet), etc., and the device's electronic circuit then contains corresponding hardware interfaces (e.g., an LTE module ("Long Term Evolution")) and implements corresponding communication protocols for data exchange. The device's electronic circuit can support Wi-Fi®, Bluetooth®, etc. for communication with mobile communication devices.

[0053] The device's electronic circuit contains a GNS module for determining position, e.g. via GPS ("Global Positioning System") or Galileo.

[0054] The device's electronic circuit contains interfaces / data buses for reading the trip data and control unit data and then implements corresponding communication protocols. The obtained trip data and the obtained at least one position of the vehicle are input into a machine learning algorithm, wherein the machine learning algorithm is configured (i.e., trained) to determine whether the obtained trip data has been tampered with based on the vehicle's movement profile and the obtained at least one position of the vehicle.

[0055] According to the invention, the machine learning algorithm is based on a neural network. The machine learning algorithm can also be based on an SVM ("Support Vector Machine"), a logistic regression, a decision tree, or the like, which is not claimed.

[0056] The machine learning algorithm is configured, i.e. it is trained, to determine a classification result for the received trip data, which indicates whether the received trip data has been manipulated.

[0057] Therefore, in some embodiments, a classification result is output for the obtained trip data, wherein the classification result is indicative of a probability of whether the trip data is tampered with.

[0058] As mentioned above, it is intended to display the journey data and the classification result to the control authorities in some embodiments so that possible further steps could then be initiated by the control authorities to determine the nature and extent of the manipulation.

[0059] Therefore, in some embodiments, the electronic circuit of the device is further configured to communicate with a mobile communication device and to transmit its own position and / or the read-out travel data to the mobile communication device.

[0060] Furthermore, in some embodiments, the electronic circuit of the device is further configured to receive a classification result for the transmitted travel data from the data processing device and to transmit the classification result to the mobile communication device.

[0061] Furthermore, in some versions, the electronic circuit of the device is designed to enable control of the device by means of the mobile communication device (e.g. notebook, smartphone, tablet, etc.) which is, for example, in the possession of the control authorities at the time of the control.

[0062] As mentioned above, in some embodiments, the obtained at least one position of the vehicle enables filtering of possible route sections of the route traveled by the vehicle, since within a limited period of time, only route sections in the vicinity of the vehicle's position are considered. In some embodiments, each of these possible route sections has a characteristic movement profile, particularly a driving speed profile for trucks. For example, the driving speed profile is different in a city or on a country road than on a highway.

[0063] The machine learning algorithm is therefore trained, based on a large amount of comparison data, to classify movement profiles of a vehicle on a driven route / route section into manipulated and non-manipulated movement profiles.

[0064] The comparison data can, for example, be recorded driving data from other real vehicles, which can then be classified accordingly and used for training. The comparison data can, for example, be driving data from training vehicles that have driven on a variety of routes and were manipulated and / or not manipulated. The comparison data can, for example, be based on traffic simulations or other known simulation methods.

[0065] It was further recognized that environmental data of the route section can improve the accuracy of the classification result of the machine learning algorithm.

[0066] Therefore, in some embodiments, environmental data of the route section is obtained and the obtained environmental data of the route section is input into the machine learning algorithm, wherein the machine learning algorithm further determines whether the obtained trip data is tampered with based on the obtained environmental data of the route section.

[0067] The environmental data can be determined from digital maps based on the at least one position and / or loaded from a memory.

[0068] In some embodiments, the environmental data of the route section represents positions of parking lots, rest areas, gas stations and / or toll booths.

[0069] In general, environmental data can be retrieved wired and / or wirelessly. Environmental data can also be received wirelessly via any radio equipment, such as those located at parking lots, rest areas, gas stations, toll booths, and the like, as well as via radio equipment of other vehicles. Environmental data can, for example, be retrieved (wirelessly) by enforcement officers from other vehicles that are in the vicinity of the checkpoint or, for example, are driving past the checkpoint. This allows, for example, position data and / or movement profiles of other vehicles to be used to detect tampering.

[0070] If, in some embodiments, the vehicle's movement profile includes break times, for example, the driving speed during the break time is practically zero, and the break time can only have occurred at certain designated locations, e.g., in parking lots, at rest areas, at gas stations, and / or at toll booths. Based on the at least one position and the driving speed profile, the distance to the position at which the break was taken can be determined in some embodiments. If the distance deviates from the actual distance to the parking lots, rest areas, gas stations, and / or toll booths on the route section, this could also be indicative of manipulation of the travel data. In some embodiments, such pattern recognition can be trained into the machine learning algorithm.

[0071] It was further recognized that, in principle, route data of the route section can improve the accuracy of the classification result of the machine learning algorithm.

[0072] Therefore, in some embodiments, route data of the route section is obtained and the obtained route data of the route section is input to the machine learning algorithm, wherein the machine learning algorithm further determines whether the obtained trip data is tampered with based on the obtained route data of the route section.

[0073] In some embodiments, the route data of the route section represents a maximum speed profile, an elevation profile, past traffic jams and / or past traffic reports.

[0074] The road data can, for example, be determined based on the at least one position from digital maps and can be determined from (official) traffic data platforms and / or loaded from a memory.

[0075] Each route section is fundamentally characterized by a predefined maximum speed profile and a predefined elevation profile, whereby, in some embodiments, characteristic patterns occur in the driving speed profile of a large number of vehicles. Starting from, for example, a checkpoint, and provided the permitted maximum speeds are observed, in some embodiments, time periods with the corresponding permitted maximum speeds arise in the travel data or the driving speed profile, whereby the time interval between the time periods is also characteristic. Due to the elevation profile, in some embodiments, characteristic patterns occur in the driving speed profile, e.g., a lower driving speed on steep inclines, whereby the time interval between the patterns is also characteristic.For example, in some embodiments, characteristic patterns occur due to deceleration and acceleration, so that the machine learning algorithm can be trained to recognize the presence or absence of these patterns in the driving speed profile of the obtained driving data. The same also applies in some embodiments to movement profiles in traffic jams or other traffic situations that can be determined based on traffic reports (e.g., road closures, slippery roads, etc.).

[0076] It was further recognized that vehicle ECU data can improve the accuracy of the classification result of the machine learning algorithm.

[0077] Therefore, in some embodiments, control unit data of the vehicle, which was recorded at least partially within the time period of the trip data, is obtained and input into the machine learning algorithm, wherein the machine learning algorithm further determines, based on the obtained control unit data, whether the obtained trip data is tampered with.

[0078] The control unit data can be received / read from an engine control unit, an ABS control unit (anti-lock braking system), an airbag control unit, a transmission control unit or the like.

[0079] In some embodiments, the electronic circuit of the device is configured to read the vehicle's control unit data and transmit it to the device for data processing via the network.

[0080] In some embodiments, the control unit data includes error data, wherein the error data represents error messages.

[0081] It is known that control units store error messages, and these messages are stored in the form of so-called—and immutable—"frozen frames." These error messages can contain, for example, the mileage, the time of the error, the vehicle speed, the engine speed, the oil pressure, the engine temperature, the pedal positions, etc., at the time of the error.

[0082] These error messages are generated, for example, when the engine control light has come on, the ABS has been triggered, the airbag has been deployed, the lighting system is faulty, the oil pressure or the engine temperature is critical, etc.

[0083] By comparing the error messages with the vehicle's movement profile, pattern recognition can be further improved in some embodiments. For example, the occurrence of an error message due to ABS activation during a break can be indicative of manipulated driving data.

[0084] In addition to the error messages from the control units, some embodiments read control unit data that has been logged over a period of time in parallel with the driving data. Telematics systems, particularly for trucks, are known that log corresponding control unit data and, if necessary, transmit it to a server (e.g., the freight forwarding company) to enable a review of the technical condition of the vehicle fleet.

[0085] In some embodiments, such control unit data may represent engine speed, vehicle speed, oil pressure, engine temperature, pedal positions, fuel consumption, exhaust emissions, etc.

[0086] By comparing a large number of movement profiles and the corresponding temporal progression of these values ​​on a section of track, the machine learning algorithm is trained in some embodiments to classify the movement profiles as manipulated and non-manipulated based on the control unit data.

[0087] In some embodiments, road data and day of the week and time are obtained, wherein the day of the week and time are associated with the at least one position of the vehicle, wherein the road data represents a road and a direction of travel on the route section, wherein the obtained road data and the day of the week and time are input to the machine learning algorithm, and wherein the machine learning algorithm further determines whether the obtained travel data is tampered with based on the road data, the day of the week and the time.

[0088] For example, due to different traffic situations, the movement profile may be different in the morning than at noon or different during the week than at the weekend, so that this is also taken into account in the pattern recognition in some embodiments in order to further improve the pattern recognition.

[0089] For example, the street and the direction of travel can be transmitted to the device during control via the mobile communication device, which then transmits the corresponding data to the data processing device.

[0090] It was further recognized that emulation driving data could improve the accuracy of the classification result of the machine learning algorithm.

[0091] Therefore, in some embodiments, the electronic circuit of the device is further configured to: To obtain emulation data, wherein the emulation data emulates a movement profile of the vehicle; to generate test signals based on the emulation data; to input the generated test signals into the control device; to read out emulation travel data from the control device, wherein the emulation travel data is based on the input test signals; and to transmit the read out emulation travel data to the device for data processing via the network.

[0092] Therefore, in some embodiments, the device is connected to the vehicle's control unit via a corresponding data bus during an on-site inspection in order to emulate the vehicle's travel while stationary and to read out the travel data recorded by the control unit, which are then transmitted to the device for data processing.

[0093] In some embodiments, the control elements generate emulation data (e.g., via a computer program) via a mobile communication device and transmit these to the device.

[0094] In such embodiments, the device then generates corresponding test signals, which, for example, emulate distance signals from a distance sensor and pause times, which are then input into the control device to emulate a journey with a predefined movement profile. By comparing the predefined movement profile and the emulated movement profile, the classification of the machine learning algorithm can then be further improved, for example, to detect even minor deviations.

[0095] Accordingly, in some embodiments, emulation trip data of the control device of the vehicle and emulation data are obtained, wherein the emulation trip data is based on test signals and the test signals are based on the emulation data, wherein the emulation data emulates a movement profile of the vehicle and the obtained emulation trip data and the obtained emulation data are input into the machine learning algorithm, wherein the machine learning algorithm further determines whether the obtained trip data is tampered with based on the obtained emulation trip data and the obtained emulation data.

[0096] It was further recognized that test drive data could improve the accuracy of the classification result of the machine learning algorithm.

[0097] Therefore, in some embodiments, the electronic circuit of the device is further configured to: to determine one's own positions during a test drive of the vehicle and to transmit them as test position data to the data processing device, wherein the own positions are indicative of the positions of the vehicle during the test drive; and to read test drive data from the control device and to transmit them to the data processing device, wherein the test position data are associated with the test drive data.

[0098] In some embodiments, the device is therefore used during a test drive with the vehicle. In such embodiments, the device is connected to the vehicle's control unit via a corresponding data bus during the test drive (or after the test drive) in order to transmit the recorded test drive data and the positions determined during the test drive to the data processing device. By comparing the movement profile during the test drive and the positions, the classification of the machine learning algorithm could be further improved to detect even minor deviations.

[0099] Accordingly, in some embodiments, test drive data of the control device of the vehicle and test position data are obtained, wherein the test position data are associated with the test drive data and represent positions during a test drive of the vehicle, wherein the test drive data represent a movement profile of the vehicle during the test drive and the obtained test drive data and the obtained test position data are input into the machine learning algorithm and wherein the machine learning algorithm further determines whether the obtained drive data is tampered with based on the obtained test drive data and the obtained test position data.

[0100] Some embodiments relate to a system as described herein for detecting manipulation of travel data of a control device of a vehicle, the system comprising the features of claim 9.

[0101] By recognizing patterns in vehicle movement profiles using a machine learning algorithm, manipulated driving data could be detected regardless of the type of manipulation, which in some embodiments could solve a major security problem.

[0102] Returning to Fig. 1 This illustrates an embodiment of a system 1 for detecting manipulation of travel data 7 of a control device 3 of a vehicle 2.

[0103] Vehicle 2 here is a lorry (hereinafter: lorry) which was stopped by the control authorities at a checkpoint KP.

[0104] The truck 2 contains the control unit 3 and control units 4, whereby the control units 4 here comprise an engine control unit and an ABS control unit and are summarized below under "the control unit 4" for simplicity.

[0105] The control bodies are in possession of a device 5 and a mobile communication device 6. The device 5 contains data buses 5a and 5b for connecting to the control device 3 and the control device 4, respectively.

[0106] The device 5 reads trip data 7 from the control unit 3 via the data bus 5a and control unit data from the control unit 4 via the data bus 5b.

[0107] The device 5 contains a GPS module 5d and determines its own position 9, which is indicative of the position of the truck 2.

[0108] The device 5 transmits the travel data 7, the control unit data 8 and the position of the device 9 to the mobile communication device 6 via a Bluetooth interface 5c.

[0109] Via a mobile radio interface 5e, the device 5 transmits the travel data 7, the control unit data 8 and the position of the device 9 to a base station 10, which transmits them via a network 11 to a server 12 (data processing device).

[0110] The server 12 contains a memory 13 on which a computer program is stored which implements a (trained) machine learning algorithm 14 and is executed by one or more processors (not shown).

[0111] The server 12 receives, based on the position 9 of the vehicle 2, environmental data 15 and route data 16 from a database 17.

[0112] The obtained trip data 7, the obtained control unit data 8, the position 9 of the vehicle 2, the environmental data 15 and the route data 16 are entered into the machine learning algorithm 14.

[0113] The machine learning algorithm 14 determines a classification result 18 for the obtained trip data 7, which is indicative of a probability whether the trip data 7 has been manipulated.

[0114] The classification result 18 is transmitted via the network to the device 5, which transmits the classification result 18 to the mobile communication device 6 of the control bodies.

[0115] Fig. 2 illustrated schematically in graphs in Fig. 2A bis Fig. 2D Design of route sections and movement profiles.

[0116] In Fig. 2A The control point KP is shown as an example and schematically, at which the vehicle 2 from Fig. 1 is controlled by the supervisory authorities.

[0117] As with reference to Fig. 1 discussed, the device 5 determines a position 9 during the check, which is indicative of the position of the vehicle 2 and thus also of the checkpoint KP.

[0118] Based on position 9 of the checkpoint KP, in this exemplary embodiment, within its perimeter (illustrated by the dashed line) are route A, route B, and route C, with vehicle 2 having traveled one of these routes prior to the check. Position 9 is therefore indicative of at least one section of the route traveled by vehicle 2.

[0119] Route A has a parking lot 20, route B has a rest area 21 and route C has no rest area within the vicinity of the checkpoint KP.

[0120] For the sake of illustration, it is assumed below that vehicle 2 has traveled route B.

[0121] In Fig. 2B Examples of embodiments of movement profiles on the AC routes are illustrated schematically.

[0122] As with reference to Fig. 1 discussed, the device 5 reads the journey data 7 from the control device 3 of the vehicle 2.

[0123] The solid line in Fig. 2B shows the real movement profile 30 of vehicle 2 on route B.

[0124] The dotted line in Fig. 2B shows the movement profile 31 of the vehicle 2 on the route B recorded by the control device 3, which can be extracted from the read-out travel data 7.

[0125] The real movement profile 30 has a consistently higher driving speed of the vehicle 2 than the recorded movement profile 31.

[0126] The recorded movement profile 31 also has, in contrast to the real movement profile 30, a pause time between the times t1 and t2.

[0127] A first comparison movement profile 32 of the route B (short dashed line in Fig. 2B ) was determined from a large number of comparison data from a large number of vehicles and from a large number of journeys with a training vehicle and represents a movement profile of the route B, which was calculated by the (trained) machine learning algorithm 14 from Fig. 1 is classified as non-tampered with.

[0128] The first comparison movement profile 32 of the route B has a typical pause time between the times t3 and t4.

[0129] A second comparison movement profile 33 of the section A (dash-dotted line in Fig. 2B ) was determined analogously to route B and is used by the (trained) machine learning algorithm 14 from Fig. 1 classified as non-manipulated.

[0130] The second comparison movement profile 33 of the route A has a typical pause time between the times t5 and t6.

[0131] A third comparison movement profile 34 of the route C (long dashed line) was determined analogously to route B and is used by the (trained) machine learning algorithm 14 from Fig. 1 classified as non-manipulated.

[0132] The third comparison movement profile 34 of route C has no pause time.

[0133] To illustrate a possible underlying principle of pattern recognition, it is assumed below that at the time of the check at checkpoint KP the control authorities have no knowledge of the section of the route travelled by vehicle 2.

[0134] Based on the recorded movement profile 31, the machine learning algorithm 14 determines that the recorded movement profile 31 for route C is to be classified as manipulated, for example due to the pause time between t1 and t2, the consistently significantly higher driving speed and the different course of the movement profiles 31 and 34.

[0135] Route C also has a maximum speed profile and an altitude profile (route data 16 of route C), which mean that the recorded movement profile 31 cannot have been created on route C.

[0136] Based on the recorded movement profile 31, the machine learning algorithm 14 determines that the recorded movement profile 31 for route A is to be classified as manipulated. For example, due to the pause time between t1 and t2 and not between t5 and t6 and the significantly different course of the movement profiles 31 and 33.

[0137] Route A also has a maximum speed profile and an altitude profile (route data 16 of route A), which mean that the recorded movement profile 31 cannot have been created on route A.

[0138] Based on the recorded movement profile 31, the machine learning algorithm 14 determines that the recorded movement profile 31 for the route B is to be classified as manipulated, for example due to the pause time between t1 and t2 and not between t3 and t4 and the temporally compressed course of the recorded movement profile 31 compared to the first comparison movement profile 32.

[0139] Assuming that the recorded movement profile 31 and the first comparison movement profile 32 do not have any pause times, the machine learning algorithm 14 could still classify the recorded movement profile 31 as manipulated on route B due to the temporally compressed course of the recorded movement profile 31 compared to the first comparison movement profile 32.

[0140] In Fig. 2C is shown schematically and by way of example how environmental data 15 improves the classification of the machine learning algorithm 14.

[0141] On the vertical axis, the distance on the route B to the control point KP is plotted over time (illustrated here as increasing linearly for illustrative purposes only).

[0142] The dotted line illustrates the distance to the checkpoint KP based on the recorded movement profile 31. Point 35 marks the distance at which the supposed break time was taken.

[0143] The short dashed line illustrates the distance to the control point KP based on the first comparison movement profile 32. Point 36 marks the distance at which the typical break time is taken.

[0144] The distance 37 marks the discrepancy between the two distances.

[0145] The environmental data show that in the vicinity of the checkpoint KP on route B there is only rest area 21 where breaks can be taken.

[0146] The distance resulting from the environmental data corresponds to the distance at point 36. Therefore, based on the environmental data, the machine learning algorithm 14 further classifies the recorded movement profile 31 as manipulated on route B.

[0147] In Fig. 2D is shown schematically and by way of example how control unit data 8 could improve the classification of the machine learning algorithm 14.

[0148] The read-out control unit data 8 shows an error message 38 at time t7, which here reports, for example, the triggering of the ABS system.

[0149] However, this error message 38 occurs within the pause time between t1 and t2, so the ABS system's activation at this time is highly unlikely. Error message 38 could also be used to identify mileage readings that appear to be manipulated, since error message 38 stores the mileage at the time the error occurs.

[0150] Overall, the machine learning algorithm 14 outputs a classification result 18, which is indicative of a manipulation of the trip data 7.

[0151] Fig. 3 schematically illustrates in a block diagram an embodiment of a training method for a machine learning algorithm 14-t for determining a manipulation of travel data 7 of a control device 3 of a vehicle 2.

[0152] The machine learning algorithm 14-t is in the training phase here and is based on a neural network in this embodiment.

[0153] The machine learning algorithm 14-t is trained with a training dataset 40.

[0154] The training data set 40 contains a plurality of data sets, wherein each data set of the plurality of data sets contains training trip data 7-t, training control unit data 8-t, at least one position 9-t associated with the training trip data 7-t, training environment data 15-t, training route data 16-t and a classification 41 ("label") that indicates whether the training trip data is manipulated or not.

[0155] The training data set 40 was determined using a large number of comparison data from a large number of vehicles and from a large number of journeys with a training vehicle, with manipulated and non-manipulated data being available.

[0156] The data sets (except for classification 41) are fed into the machine learning algorithm 14-t, which outputs a classification result 18-t for each data set based on this.

[0157] The classification result 18-t and the classification 41 are input into a loss function 42, where the loss function 42 is a cross entropy loss.

[0158] Based on a difference between the classification result 18-t and the classification 41, the weight changes 43 are output and the weights of the machine learning algorithm 14-t are updated accordingly.

[0159] After completion of the training phase, the trained machine learning algorithm 14 with trained weights is available.

[0160] Fig. 4 schematically illustrates in a block diagram an embodiment of a general-purpose computer 130.

[0161] The general-purpose computer 130 represents an electronic circuit with which the data processing apparatus 12 and the device 5 can be implemented as described herein.

[0162] The general purpose computer 130 has components 131 to 135, a GNS module 136 in the case of device 5 and a data bus 137.

[0163] Embodiments that use software, firmware, programs, or the like to perform the methods described herein may be installed on the general-purpose computer 130, which is then configured to be suitable for the particular embodiment.

[0164] The general-purpose computer 130 has a CPU 131 ("Central Processing Unit") that can execute various types of procedures and methods as described herein, e.g., in accordance with programs stored in a read-only memory ("ROM") 132, stored in a memory 134, and loaded into a random access memory ("RAM") 133.

[0165] The CPU 131, the ROM 132, the RAM 133 and the memory are connected to the data bus 137.

[0166] In addition, a communication interface 135 is connected to the data bus 137, which can be configured, for example, for communication via a local area network (LAN), a wireless local area network (WLAN), a mobile telecommunication system (GSM, UMTS, LTE, NR, etc.), Bluetooth, infrared, etc. The communication interface implements corresponding hardware interfaces and communication protocols.

[0167] The GNS module 136 is connected to the data bus 137 and can determine a position in accordance with a global navigation system such as GPS or Galileo.

[0168] Fig. 5 schematically illustrates in a flowchart an embodiment of a method 200 for determining manipulation of travel data of a control device of a vehicle, which in some embodiments runs on the general-purpose computer 130.

[0169] At 201, trip data is obtained, wherein the trip data represents a movement profile of the vehicle along a traveled route, as discussed herein.

[0170] At 202, at least one position of the vehicle is obtained, wherein the at least one position of the vehicle is indicative of at least one route segment of the traveled route as discussed herein.

[0171] At 203, environmental data of the route section is obtained as discussed herein.

[0172] At 204, route data of the route section is obtained as discussed herein.

[0173] At 205, control unit data of the route section is obtained as discussed herein.

[0174] At 206, the obtained travel data, the obtained at least one position of the vehicle, the obtained environmental data, the obtained route data and the obtained control unit data are input into a machine learning algorithm, wherein the machine learning algorithm is configured to determine, based on the movement profile of the vehicle, the obtained at least one position of the vehicle, the obtained environmental data, the obtained route data and the obtained control unit data, whether the obtained travel data has been manipulated.

Claims

1. A method for determining a manipulation of driving data (7) of a control device (3) of a truck (2), comprising: receiving the driving data (7), wherein the driving data (7) represent a movement profile (31) of the truck (2) over a driven route, and wherein the movement profile (31) is characterized by a driving speed profile as well as steering times and pause times, which are represented in the driving data (7); receiving a position (9) of the truck (2), wherein said position is determined at a checkpoint (KP) during a control of the truck (2), and wherein said position (9) of the truck (2) is indicative of at least one section of the driven route; and inputting the received driving data (7) and said received position (9) of the truck (2) into a neural network (14), wherein the neural network (14) is configured to determine, based on the movement profile (31) of the truck (2) and said received position (9) of the truck (2), whether the received driving data (7) are manipulated.

2. The method of claim 1, further comprising: receiving, based on said position (9), environmental data (15) of the section of the route; and inputting the received environmental data (15) of the section of the route into the neural network (14), wherein the neural network (14) determines, further based on the received environmental data (15) of the section of the route, whether the received driving data (7) are manipulated.

3. The method of claim 2, wherein the environmental data (15) of the section of the route represent positions of parking areas (20), rest areas (21), fuel stations, and / or toll stations.

4. The method of any one of the previous claims, further comprising: receiving, based on said position (9), route data (16) of the section of the route; and inputting the received route data (16) of the section of the route into the neural network (14), wherein the neural network (14) determines, further based on the received route data (16) of the section of the route, whether the received driving data (7) are manipulated.

5. The method of claim 4, wherein the route data (16) of the section of the route represent a maximum speed profile and / or an elevation profile of the section of the route.

6. The method of any one of the previous claims, further comprising: receiving control unit data (8) of the truck (2), wherein the control unit data (8) were recorded at least partially within the time period of the driving data (7); and inputting the received control unit data (8) into the neural network (14), wherein the neural network (14) determines, further based on the received control unit data (8), whether the received driving data (7) are manipulated.

7. The method of claim 6, wherein the control unit data (8) include error data, and wherein the error data represent error messages (38).

8. Apparatus for data processing (12), comprising an electronic circuit, wherein the electronic circuit is configured to carry out the method according to any one of the previous claims.

9. A system (1) for determining a manipulation of driving data (7) of a control device (3) of a truck (2), comprising: a device (5) comprising an electronic circuit, wherein the electronic circuit is configured to: read the driving data (7) from the control device (3), wherein the driving data (7) represent a movement profile (31) of the truck (2) over a driven route, and wherein the movement profile (31) is characterized by a driving speed profile as well as steering times and pause times, which are represented in the driving data (7), determine its own position (9) at a checkpoint (KP) during a control of the truck (2) transmit the read driving data (7) and its own position (9) to an apparatus for data processing (12) via a network (11), wherein said own position (9) is indicative of a position (9) of the truck (2), and said position (9) of the truck (2) is indicative of at least one section of the driven route; and the apparatus for data processing (12), comprising an electronic circuit, wherein the electronic circuit is configured to: receive the driving data (7); receive said position (9) of the device (5); and input the received driving data (7) and said received position (9) of the device (5) into a neural network (14), wherein the neural network (14) is configured to determine, based on the movement profile (31) of the truck (2) and said received position (9) of the device (5), whether the received driving data (7) are manipulated.