Systems, media, and methods applying machine learning to telematics data to generate vehicle fingerprint

Abstract

Described herein are systems and methods for applying machine learning to telematics data to generate a unique vehicle fingerprint by periodically receiving telematics data generated at a plurality of sensors of a vehicle; standardizing the telematics data; aggregating the standardized telematics data; applying a trained machine learning model to embed the aggregated telematics data into a low-dimensional state; and generating a unique vehicle fingerprint, the vehicle fingerprint comprising a static component, a dynamic component, or both a static component and a dynamic component; including iterative repetition to update the dynamic component of the vehicle fingerprint.

What is claimed is: 
     
         1 . A computer-implemented system comprising: a digital processing device comprising: at least one processor, an operating system configured to perform executable instructions, a memory, and a computer program including instructions executable by the digital processing device to create an application applying machine learning to telematics data to generate a unique vehicle fingerprint for a specific vehicle, the application comprising:
 a) a software module periodically receiving telematics data generated at a plurality of sensors of a vehicle;   b) a software module standardizing the telematics data;   c) a software module aggregating the standardized telematics data;   d) a software module applying a trained machine learning model to embed the aggregated telematics data into a low-dimensional state; and   e) a software module generating a unique vehicle fingerprint for the specific vehicle, the vehicle fingerprint comprising a static component, a dynamic component, or both a static component and a dynamic component;   wherein b) - e) are iteratively repeated to update the dynamic component of the vehicle fingerprint.   
     
     
         2 . The system of  claim 1 , wherein the telematics data originates at a plurality of vehicle sensors connected to the vehicle’s controller area network (CAN) bus. 
     
     
         3 . The system of  claim 1 , wherein the telematics data is transmitted wirelessly via the vehicle’s connectivity module. 
     
     
         4 . The system of  claim 1 , wherein the telematics data comprises vehicle data. 
     
     
         5 . The system of  claim 4 , wherein the vehicle data comprises one or more of: travel speed, wheel speed, acceleration, orientation, engine revolutions per minute (RPM), engine temperature, coolant temperature, oil temperature, current gear, battery voltage, suspension activity, climate control system settings, window positions, door statuses, mirror positions, internal air temperature, tire pressures, seat belt tension, tire pressure, passenger occupancy, radar status, diagnostic trouble codes, historical maintenance triggers, and personalization settings. 
     
     
         6 . The system of  claim 1 , wherein the telematics data comprises environmental data. 
     
     
         7 . The system of  claim 6 , wherein the environmental data comprises one or more of: location, altitude, external air temperature, external humidity, precipitation, road type, light, and road condition. 
     
     
         8 . The system of  claim 1 , wherein the telematics data comprises driver data. 
     
     
         9 . The system of  claim 8 , wherein the driver data comprises one or more of: steering wheel position, steering wheel velocity, brake pedal position, braking force, gas pedal position, shifting, internal lighting use, headlight use, turn signal use, mirror adjustments, window adjustments, climate control system use, entertainment system use, and seat belt use. 
     
     
         10 . The system of  claim 1 , wherein at least some of the telematics data is sequential time series data. 
     
     
         11 . The system of  claim 1 , wherein the telematics data is received at least every 15 minutes, 10 minutes, 5 minutes, 1 minute, 45 seconds, 30 seconds, 15 seconds, 10 seconds, 5 seconds, or 1 second. 
     
     
         12 . The system of  claim 11 , wherein the telematics data is received substantially continuously. 
     
     
         13 . The system of  claim 1 , wherein the machine learning model comprises a neural network. 
     
     
         14 . The system of  claim 13 , wherein the neural network is a plurality of stacked recurrent neural networks. 
     
     
         15 . The system of  claim 13 , wherein the neural network comprises a plurality of recurrent neural networks and a fully connected layer. 
     
     
         16 . The system of  claim 1 , wherein b) - e) are iteratively repeated to update the dynamic component of the vehicle fingerprint at least every 15 minutes, 10 minutes, 5 minutes, 1 minute, 45 seconds, 30 seconds, 15 seconds, 10 seconds, 5 seconds, or 1 second. 
     
     
         17 . The system of  claim 16 , wherein b) - e) are iteratively repeated to update the dynamic component of the vehicle fingerprint substantially continuously. 
     
     
         18 . The system of  claim 1 , wherein the vehicle fingerprint comprises a component malfunction or vehicle system failure risk. 
     
     
         19 . The system of  claim 1 , wherein the vehicle fingerprint comprises a system-aggregated malfunction or failure risk. 
     
     
         20 . The system of  claim 1 , wherein the application further comprises a software module identifying vehicle fingerprints, from among a plurality of vehicle fingerprints, which are similar to each other. 
     
     
         21 . The system of  claim 20 , wherein the similarity is measured by a calculated similarity score. 
     
     
         22 . The system of  claim 1 , wherein the application further comprises a software module utilizing the vehicle fingerprint to predict future component-specific or vehicle-system-specific malfunctions or failures. 
     
     
         23 . The system of  claim 1 , wherein the application further comprises a software module utilizing the vehicle fingerprint to predict future system-aggregated malfunctions or failures. 
     
     
         24 . The system of  claim 1 , wherein the application further comprises a software module utilizing the vehicle fingerprint to identify and group vehicles based on common component-specific or system-aggregated malfunction or failure histories. 
     
     
         25 . A computer-implemented method of generating a unique vehicle fingerprint for a specific vehicle comprising:
 a) periodically collecting, by a computer, telematics data generated at a plurality of sensors of a vehicle;   b) standardizing, by the computer, the telematics data;   c) training, at a computer cluster, a machine learning model to embed the aggregated telematics data into a low-dimensional state;   d) applying, by the computer or a vehicle, the trained machine learning model to embed the aggregated telematics data into a low-dimensional state;   e) generating, by the computer or the vehicle, a unique vehicle fingerprint, the vehicle fingerprint comprising a static component, a dynamic component, or both a static component and a dynamic component; and   f) iteratively repeating steps b) - e) to update the dynamic component of the vehicle fingerprint.   
     
     
         26 . The method of  claim 25 , wherein the method further comprises:
 a) saving, by the computer or the vehicle, weights generated by the trained machine leaning model; and   b) inferring, by the computer or the vehicle, a unique vehicle fingerprint for the specific vehicle based on the weights for novel telematics data generated at a plurality of sensors of the vehicle.   
     
     
         27 . A system for applying machine learning to telematics data to generate a unique vehicle fingerprint for a specific vehicle comprising:
 a) at least one server processor configured to perform at least the following:
 i) periodically receive telematics data generated at a plurality of sensors of a vehicle; 
 ii) standardize the telematics data; 
 iii) aggregate the standardized telematics data; 
 iv) apply a trained machine learning model to embed the aggregated telematics data into a low-dimensional state; 
 v) save weights generated by the trained machine leaning model; 
 vi) transmit the saved weights to the vehicle; and 
 vii) iteratively repeating a) ii) - a) vi) to update the transmitted weights; and 
   b) at least one vehicle processor configured to perform at least the following:
 i) receive the transmitted weights; and 
 ii) infer a unique vehicle fingerprint for the specific vehicle based on the transmitted weights for novel telematics data generated at a plurality of sensors of the vehicle, the vehicle fingerprint comprising a static component, a dynamic component, or both a static component and a dynamic component.