Vehicle event prediction device, off-board backend, and method to operate a vehicle event prediction device

Abstract

Operating a vehicle event prediction device (12) comprises receiving and storing input data (16) and vehicle event data (18) and training a prediction model (20), by a data collector module (14) of th

Description

FIELD OF THE INVENTION

[0001] The present invention relates to the field of automobiles. More specifically, the present invention relates to a vehicle event prediction device for a motor vehicle and a corresponding off-board backend. Furthermore, the present invention relates to a corresponding method to operate a vehicle event prediction device. BACKGROUND INFORMATION

[0002] Connected vehicles have unique characteristics that differentiate them from other loT devices. Vehicles operate in an unreliable and hard to predict environment and network space. In contrast to most other loT devices they also have a multitude of different operational modes, such as manual driving, assisted driving, charging, or parking. Many of those operational modes depend heavily on the driver’s personal life, location, circumstances, and behavior. Vehicles are also embedded increasingly more into on a client server architecture. The central systems need to optimize the interaction with the vehicles to decrease latency of actions, such as purchases and activation of services but also to optimize the transmission of large data such as Over the Air (OTA) updates or uploading of collected data from the fleet. The challenge of this wide optimization problem is to generate a predictability without compromising the privacy of the drivers. SUMMARY OF THE INVENTION

[0003] It is an object of the present invention to provide a vehicle event prediction device for a motor vehicle, an off-board backend as well as a corresponding method, which allow a prediction of predefined vehicle events related to the motor vehicle without sharing or compromising Personal Identifiable Information with a system external to the motor vehicle.

[0004] This object is solved by a vehicle event prediction device for a motor vehicle, an off-board backend, as well as a corresponding method to operate a vehicle event prediction device according to the independent claims. Advantageous embodiments are presented in the dependent claims.

[0005] A first aspect of the invention is related to a vehicle event prediction device for a motor vehicle.

[0006] The vehicle event prediction device comprises a data collector module, configured to receive and to store input data and vehicle event data for training a prediction model. The vehicle event data relate to an occurrence of predefined vehicle events. The predefined vehicle events may be configured or available (the events may change depending on software updates or activation of new services).

[0007] The vehicle event prediction device comprises a machine learning training module. The machine learning training module is configured to retrieve the input data and the vehicle event data from the data collector module. The machine learning training module is configured to train a prediction model based on the input data and the vehicle events. The prediction model gives a prediction of an occurrence of a respective one of the predefined vehicle events.

[0008] The vehicle event prediction device comprises a prediction model module, configured to host the prediction model. The prediction model module is configured to generate a prediction calendar based on the prediction model. The prediction calendar describes a probability of the occurrence of the respective one of the predefined vehicle events in predefined future timeslots.

[0009] The vehicle event prediction device comprises a prediction calendar module, configured to host the prediction calendar.

[0010] The vehicle event prediction device further comprises a backend communication module, configured to provide the prediction calendar to an off-board backend. The off-board backend is outside the motor vehicle.

[0011] This invention provides a mechanism to predict vehicle system events using technical vehicle data as well as Personal Identifiable Information from the user to increase the reliability of predictions without sharing or compromising PH with an external system. The system is centered around a quantized prediction calendar, describing the likelihood of the occurrence of a vehicle event at a given future point in time.

[0012] The vehicle event prediction device may be a implemented as a software module installes on an existing device within the vehicle.

[0013] According to an embodiment, the predefined vehicle events comprise a change of a charger plug state, a change of an engine operation state, and / or a mobile connectivity status of the vehicle. These are some examples and not an exhaustive list of vehicle events that are contemplated by this patent application.

[0014] According to an embodiment, the input data comprise vehicle state data, the vehicle state data comprising a current location of the vehicle, a current state of charge of the vehicle. These are some examples and not an exhaustive list of vehicle events that are contemplated by this patent application.

[0015] According to an embodiment, the input data comprise personal data, the personal data comprising a calendar of a vehicle related person, a contact list of the vehicle related person and / or a location list of the vehicle related person.

[0016] According to an embodiment, the data collector module is configured to receive input data from the vehicle.

[0017] According to an embodiment, the data collector module is configured to receive input data from a device outside the vehicle via a mobile communication network including wifi or in-vehicle mobile network.

[0018] According to an embodiment, the data collector module is configured to receive a data collector configuration from the off-board backend, and to collect the input data and / or the events according to the data collector configuration.

[0019] A second aspect of the invention is related to an off-board backend. The off-board backend is configured to provide a data collector configuration to a vehicle event prediction device for a motor vehicle. The data collector configuration describes input data and / or the predefined events to be collected by a data collector module of the vehicle event prediction device. The off-board backend is further configured to receive a prediction calendar from the vehicle event prediction device.

[0020] A third aspect of the invention is related to a method to operate a vehicle event prediction device for a vehicle. The method comprises the following steps.

[0021] Receiving and storing input data and vehicle event data for training a prediction model, by a data collector module of the vehicle event prediction device. The vehicle event data relate to an occurrence of predefined vehicle events.

[0022] Retrieving the input data and the vehicle event data from the data collector module by a machine learning training module.

[0023] Training a prediction model in a prediction model module, based on the input data and the vehicle events. The prediction model gives a prediction of an occurrence of a respective one of the predefined vehicle events.

[0024] Generating a prediction calendar in a prediction calendar module by the prediction model module, based on the prediction model. The prediction calendar describing a probability of an occurrence of the vehicle events in predefined future timeslots.

[0025] Providing the prediction calendar to an off-board backend by a backend communication module.

[0026] Advantageous embodiments of the vehicle event prediction device are to be regarded as advantageous embodiments of the method and the off-board backend. The vehicle event prediction device comprises means for performing the method. The vehicle event prediction device and the off-board backend may comprise computing units.

[0027] A computing unit may in particular be understood as a data processing device, which comprises processing circuitry. The computing unit can therefore in particular process data to perform computing operations. This may also include operations to perform indexed accesses to a data structure, for example a look-up table, LUT.

[0028] In particular, the computing unit may include one or more computers, one or more microcontrollers, and / or one or more integrated circuits, for example, one or more application-specific integrated circuits, ASIC, one or more field-programmable gate arrays, FPGA, and / or one or more systems on a chip, SoC. The computing unit may also include one or more processors, for example one or more microprocessors, one or more central processing units, CPU, one or more graphics processing units, GPU, and / or one or more signal processors, in particular one or more digital signal processors, DSP. The computing unit may also include a physical or a virtual cluster of computers or other of said units.

[0029] In various embodiments, the computing unit includes one or more hardware and / or software interfaces and / or one or more memory units.

[0030] A memory unit may be implemented as a volatile data memory, for example a dynamic random access memory, DRAM, or a static random access memory, SRAM, or as a non-volatile data memory, for example a read-only memory, ROM, a programmable read-only memory, PROM, an erasable programmable read-only memory, EPROM, an electrically erasable programmable read-only memory, EEPROM, a flash memory or flash EEPROM, a ferroelectric random access memory, FRAM, a magnetoresistive random access memory, MRAM, or a phase-change random access memory, PCRAM.

[0031] Further advantages, features, and details of the invention derive from the following description of preferred embodiments as well as from the drawings. The features and feature combinations previously mentioned in the description as well as the features and feature combinations mentioned in the following description of the figures and / or shown in the figures alone can be employed not only in the respectively indicated combination but also in any other combination or taken alone without leaving the scope of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The novel features and characteristic of the disclosure are set forth in the appended claims. The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and together with the description, serve to explain the disclosed principles. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. The same numbers are used throughout the figures to reference like features and components. Some embodiments of system and / or methods in accordance with embodiments of the present subject matter are now described below, by way of example only, and with reference to the accompanying figures.

[0033] The drawings show in:

[0034] Fig. 1 a schematic illustration of a motor vehicle comprising a vehicle event prediction device; and

[0035] Fig. 2 a schematic illustration of a method to operate a vehicle event prediction device.

[0036] In the figures the same elements or elements having the same function are indicated by the same reference signs. DETAILED DESCRIPTION

[0037] In the present document, the word "exemplary" is used herein to mean "serving as an example, instance, or illustration". Any embodiment or implementation of the present subject matter described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

[0038] While the disclosure is susceptible to various modifications and alternative forms, specific embodiments thereof have been shown by way of example in the drawing and will be described in detail below. It should be understood, however, that it is not intended to limit the disclosure to the particular forms disclosed, but on the contrary, the disclosure is to cover all modifications, equivalents, and alternatives falling within the scope of the disclosure.

[0039] The terms “comprises”, “comprising”, or any other variations thereof, are intended to cover a non-exclusive inclusion so that a setup, device or method that comprises a list of components or steps does not include only those components or steps but may include other components or steps not expressly listed or inherent to such setup or device or method. In other words, one or more elements in a system or apparatus preceded by “comprises” or “comprise” does not or do not, without more constraints, preclude the existence of other elements or additional elements in the system or method.

[0040] In the following detailed description of the embodiment of the disclosure, reference is made to the accompanying drawing that forms part hereof, and in which is shown by way of illustration a specific embodiment in which the disclosure may be practiced. This embodiment is described in sufficient detail to enable those skilled in the art to practice the disclosure, and it is to be understood that other embodiments may be utilized and that changes may be made without departing from the scope of the present disclosure. The following description is, therefore, not to be taken in a limiting sense.

[0041] Fig. 1 shows a schematic illustration of a motor vehicle comprising a vehicle event prediction device.

[0042] The motor vehicle 10 may be designed as a car or a truck. The vehicle event prediction device 12 may be configured to provide a mechanism to predict vehicle system events using input data 16 comprising technical vehicle data as well as personal identifiable information from a user related to the motor vehicle 10 to increase a reliability of predictions of vehicle system events without sharing or compromising personal identifiable information with an external system outside the motor vehicle 10. The vehicle event prediction device 12 may be centered around a quantized prediction calendar 26, describing a likelihood of an occurrence of a vehicle system event at a given future point in time.

[0043] The vehicle event prediction device 12 may comprise a data collector module 14. The data collector module 14 may be responsible for collecting the input data 16 comprising data points for a training process as well as for collecting vehicle event data 18 describing the vehicle system events from various input sources.

[0044] The data collector module 14 may be responsible for collecting the input data 16 that may be used as learning data as well as the vehicle event data 18 related to the vehicle events. The data collector module 14 may connect to various data sources in the motor vehicle 10 via a vehicle network system 34 such as CAN-Bus, Flexray or automotive ethernet. Through the connectivity to the vehicle network system 34, the data collector module 14 may have an ability to read the input data 16 such as location data of the vehicle 10, GPS, a state of charge of a motor vehicle battery or any other arbitrary data points. In addition to enabling the data collector module 14 to collect the input data 16 that is stored on the motor vehicle 10 such as calendar or contact data, it may also use a vehicle interface system 36 to access the input data 16 on other connected devices such as a Smartphone or a tablet external to the vehicle 10. This may entail contacts, calendar, or location history of the user. Further any to the motor vehicle 10 connected online accounts may provide additional data. Any kind of available signal can be used as the vehicle event data 18 describing the vehicle events. A vehicle event may be a change of a state, such as a charger plugged-state, an engine start / stop or mobile network connectivity status. Due to technical limitations of temporary storage, compute and vehicle bus bandwidth, the data collector module 14 may be configured remotely by the off-board backend 32 to define which data shall be ingested and what vehicle events shall be collected. The configuration may comprise a transmission of a data collector configuration 42 from the off-board backend 32 to the data collector module 14.

[0045] The vehicle event prediction device 12 may comprise a machine learning training module 24 that may be configured to ingest the input data 16 and the vehicle event data 18 from the data collector module 14 and to learn autonomously based on the defined vehicle events. The training may be self-supervised as the prediction of a vehicle system event can be correlated with an actual occurrence or absence of the vehicle system event. A vehicle system event in this case could be a successful connection and communication of the vehicle 10 to the off-board backend 32. The vehicle event prediction device 12 can then evaluate whether the actually occurring vehicle system event was correctly predicted. The training is exclusively done on-board the motor vehicle 10. Besides reducing data transfer volume, this enables the vehicle event prediction device 12 to use sensitive personal identifiable information data without compromising the privacy of the user as no data leaves the motor vehicle 10.

[0046] The vehicle event prediction device 12 may comprise a prediction model 20 module. The prediction model 20 module may store an actual trained prediction model 20, able to generate a prediction calendar 26.

[0047] The prediction model 20 may be a layered output of the machine learning training module 24. It may be used to regularly generate an up-to-date prediction calendar 26 based on the system configuration. Even though it consumes fewer computing resources and may run in more vehicle 10 states than the machine learning training module 24, the host system or user may have an ability to suspend the execution of the prediction model 20.

[0048] The vehicle event prediction device 12 may comprise a prediction calendar module 28, configured to host the prediction calendar 26. The prediction calendar 26 may be a quantized, discrete calendar containing the likelihood of a vehicle event at a future point in time. It may be based on a large amount of the input data 16 that may contain personal identifiable information data. The prediction calendar 26 may be generated by the prediction model module 22 using the prediction model 20. The key is that it only contains the predicted vehicle event and a respective occurrence likelihood. The prediction calendar 26 may be generated and updated by the prediction model module 22 when a defined threshold of the prediction likelihood has changed. A prediction length can be configured. The reliability may decrease the longer the prediction calendar 26 is planned for into the future. The prediction calendar 26 may only contain data about the probability of a defined vehicle event at a given future time. The training data used to train the model to generate the prediction calendar 26 may include personal identifiable information data, but it is impossible to deduce any kind of input data 16. To further increase privacy aspects, a generated or tokenized vehicle 10 identifier may be used instead of a Vehicle Identification Number VIN.

[0049] A schematic prediction calendar 26 is shown in the table Tab 1. Tab 1 shows an example quantization of 15 minutes for vehicle 10 network prediction. Day +1 Day +2 Day +3 Day +4 00:00 00:15 Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 50% Network prediction: - Status: 50% connected - Speed: 99% faster than 10 Mbit / s 00:15 00:30 Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 99% Network prediction: - Status: 99% connected - Speed: 99% faster than 10 Mbit / s Compute availability prediction: 50% Network prediction: - Status: 50% connected - Speed: 99% faster than 10 Mbit / s Compute Compute Compute Compute availability availability availability availability prediction: 50% prediction: 50% prediction: 50% prediction: 1% 00:30 Network prediction: Network prediction: Network prediction: Network prediction: — - Status: 1% - Status: 1% - Status: 1% - Status: 1% 00:45 connected connected connected connected - Speed: 10% - Speed: 10% - Speed: 10% - Speed 10% faster faster than faster than faster than than 5 Mbit / s 5 Mbit / s 5 Mbit / s 5 Mbit / s Tab 1

[0050] The vehicle event prediction device 12 may comprise a backend communication module 30. The backend communication module 30 may be configured to share the prediction calendar 26 of the prediction calendar module 28 with the off-board backend 32 outside the vehicle 10. The backend communication module 30 may have three main tasks: It may receive the data collector configuration 42 from the off-board backend 32 to configure the components of the vehicle event prediction device 12. It may detect changes in the prediction calendar 26 and compare them against a defined threshold to trigger the upload of the current prediction calendar 26. In addition, it may upload the prediction calendar 26 to the off-board backend 32 based either on detected changes or of a trigger event from the off-board backend 32. The backend communication module 30 may use an existing vehicle network and telecommunication unit TCll of the motor vehicle 10 to communicate with the off-board backend 32.

[0051] Fig. 2 a schematic illustration of a method to operate a vehicle event prediction device 12.

[0052] The method may comprise the following steps:

[0053] Step S1 may comprise a transmission of a data collector configuration 42 from an off-board backend 32 to a data collector module 14 of the vehicle event prediction device 12.

[0054] Step S2 may comprise a configuring of the data collector module 14 to collect input data 16 and vehicle event data 18 according to the data collector configuration 42.

[0055] Step S3 may comprise a collection of the input data 16 and the vehicle event data 18 according to the data collector configuration 42 by the data collector module 14.

[0056] Step S4 may comprise a transmission of the input data 16 and the vehicle event data 18 from the data collector module 14 to a machine learning training module 24. The transmission may be initiated by the data collector module 14, when a buffer of new input data 16 and / or new vehicle event data 18 exceeds a predefined threshold.

[0057] Step S5 may comprise a training of a prediction model 20 in a prediction model module 22 based on the input data 16 and the vehicle event data 18 by a machine learning training module 24. The prediction model 20 may be a self-supervised model training based on a comparison of predicted vehicle events and actual vehicle events.

[0058] Step S6 may comprise a generation and / or an update of the prediction model 20 stored in the prediction model module 22.

[0059] Step S7 may comprise a generation and / or an update of a prediction calendar 26 in a prediction calendar module 28 using the prediction model 20 by the prediction model module. 28 The step may be initiated, when a prediction calendar length is shorter than a predefined threshold.

[0060] Step S8 may comprise a retrieving of the prediction calendar 26 by a backend communication module 30.

[0061] Step S9 may comprise a transmission of the prediction calendar 26 from the backend communication module 30 to the off-board backend 32. Reference signs 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 Tab vehicle vehicle event prediction device data collector module input data vehicle event data prediction model prediction model module machine learning training module prediction calendar prediction calendar module backend communication module off-board backend vehicle network system vehicle interface system vehicle sensor system vehicle gateway data collector configuration table

Claims

1. A vehicle event prediction device (12) for a vehicle (10) characterized in thatthe vehicle event prediction device (12) comprises- a data collector module (14), configured to receive and to store input data (16) and vehicle event data (18) for training a prediction model (20), the vehicle event data (18) related to an occurrence of predefined vehicle events;- a machine learning training module (24), configured to retrieve the input data (16) and the vehicle event data (18) from the data collector module (14), and to train a prediction model (20) based on the input data (16) and the vehicle event data (18), the prediction model (20) giving a prediction of an occurrence of a respective one of the predefined vehicle events,- a prediction model module (22), configured to host the prediction model (20), and to generate a prediction calendar (26) based on the prediction model (20), the prediction calendar (26) describing a probability of an occurrence of the vehicle events in predefined future timeslots;- a prediction calendar module (28), configured to host the prediction calendar (26), - a backend communication module (30), configured to provide the prediction calendar (26) to an off-board backend (32).

2. The vehicle event prediction device (12) according to claim 1, characterized in thatthe predefined vehicle events comprise a change of a charger plugger state, a change of an engine operation state, and / or a mobile connectivity status of the vehicle (10).

3. The vehicle event prediction device (12) according to claim 1 or 2, characterized in thatthe input data (16) comprise vehicle state data, the vehicle state data comprising a current location of the vehicle (10), a current state of charge of the vehicle (10).

4. The vehicle event prediction device (12) according to any one of claims 1 to 3, characterized in thatthe input data (16) comprise personal data, the personal data comprising a calendar of a vehicle related person, a contact list of the vehicle related person and / or a location list of the vehicle related person.

5. The vehicle event prediction device (12) according to any one of claims 1 to 4, characterized in thatthe data collector module (14) is configured to receive input data (16) from the vehicle (10).

6. The vehicle event prediction device (12) according to any one of claims 1 to 5, characterized in thatthe data collector module (14) is configured to receive input data (16) from a device outside the vehicle (10) via a wireless communication network.

7. The vehicle event prediction device (12) according to any one of claims 1 to 6, characterized in thatthe data collector module (14) is configured to receive a data collector configuration (42) from the off-board backend (32) , and to collect the input data (16) and / or the vehicle event data (18) according to the data collector configuration (42).

8. An off-board backend (32), characterized in thatthe off-board backend (32) is configured to- provide a data collector configuration (42) to a vehicle event prediction device (12), the data collector configuration (42) describing input data (16) and / or the vehicle event data (18) to be collected by a data collector module (14) of the vehicle event prediction device (12), and- receive a prediction calendar (26) from the vehicle event prediction device (12).

9. A method to operate a vehicle event prediction device (12) for a vehicle (10) comprising the following steps of:- receiving and storing input data (16) and vehicle event data (18) for training a prediction model (20), by a data collector module (14) of the vehicle event prediction device (12), the vehicle event data (18) relating to an occurrence of predefined vehicle events;- retrieving the input data (16) and the vehicle event data (18) from the data collector module (14) by a machine learning training module (24);- training a prediction model (20) in a prediction model module (22), based on the input data (16) and the vehicle events, the prediction model (20) giving a prediction of an occurrence of a respective one of the predefined vehicle events;- generating a prediction calendar (26) in a prediction calendar module (28) by the prediction model module (22), based on the prediction model (20), the prediction calendar (26) describing a probability of an occurrence of the vehicle events in predefined future timeslots; and- providing the prediction calendar (26) to an off-board backend (32) by a backend communication module (30).

Images