Electronic system for dynamic, quasi-realtime measuring and identifying driver maneuvers solely based on mobile phone telemetry, and a corresponding method thereof

Patent metadata

Field	Value
Publication number	US-11216888-B2
Application number	US-201815944525-A
Country	US
Kind code	B2
Filing date	Apr 3, 2018
Priority date	Sep 6, 2017
Publication date	Jan 4, 2022
Grant date	Jan 4, 2022

Abstract

An electronic, real-time system performs maneuver recognition of vehicles based on dynamically measured telematics data, particularly the sensory data of smartphone sensors, and more particularly data from the accelerometer sensor and the global positioning system (GPS) sensor and/or the gyroscope sensor of a smartphone. The axes of the smartphone may be moving independently relative to the axes of the vehicle and thus do not need to be aligned with the axes of the vehicle. Driver behaviors and operational parameters are automatically measured and discriminated, based on automatically individuated and measured driver maneuvers within various measured vehicle trajectories, and an output signal is generated based upon derived risk measure parameters and/or crash attitude measure parameters. The system can use score-driven, especially risk-score driven, operations associated with motor vehicles or transportation modes for passengers or goods, and reliant on a dynamic, telematics-based data aggregation and dynamically measured driving maneuvers, respectively.

First claim

The invention claimed is:

1. An electronic maneuver detection system in real time based on dynamically measured telematics data, wherein driver behaviors and operational parameters are measured and discriminated based on automatically individuated driver maneuvers within various measured vehicle trajectories, and wherein measured sensory data are real-time filtered, and start and endpoints of driving events are isolated and identified in real-time, the electronic maneuver detection system comprising:
a plurality of sensors included in each of mobile telematics devices, the plurality of sensors including at least an accelerometer sensor, a global positioning system (GPS) sensor, and a gyroscope sensor, the plurality of sensors configured to capture the telematics data associated with a motor vehicle and users; and
circuitry configured to:
capture telematics data, by the plurality of sensors of each of the mobile telematics devices associated with a plurality of motor vehicles, the mobile telematics devices comprising one or more wireless connections, wherein, for providing the wireless connection, each of the mobile telematics devices acts as a wireless node within a corresponding data transmission network by antenna connections of the mobile telematics device, wherein each of the mobile telematics devices is an integral part of a mobile phone device and connected to a monitoring cellular mobile node application of the mobile phone device;
implement a dynamic, functional aggregation-based telematics circuit to which the mobile telematics devices are connected, wherein a data link is set between the dynamic, functional aggregation-based telematics circuit transmitting at least the captured telematics data from the mobile telematics devices to the dynamic, functional aggregation-based telematics circuit;
for driver maneuver identification, match sequential patterns of the captured telematics data with searchable predefined measuring parameter sets of driving maneuvers by the dynamic, functional aggregation-based telematics circuit, wherein the dynamic, functional aggregation-based telematics circuit stores the searchable predefined measuring parameter sets of driving maneuvers, wherein each driving maneuver is composed of a plurality of hierarchically structured operation-measuring parameter sets providing a hierarchical structure of vehicle operations or driving features, and wherein a specific operation or driving feature of a motor vehicle is separately triggerable based on a corresponding operation parameter set, wherein each driving maneuver is composed of sequential patterns of operations or driving features by the hierarchically structured operation-measuring parameter sets which are each extractable and measurable by measuring telematics parameters exhibiting during a lifespan of a driving maneuver;
for pattern template identification, store searchable predefined measuring parameter pattern templates of driving maneuvers, wherein each single driving maneuver is composed of a plurality of prototypical parameter patterns, wherein an individuated time interval is measured spanning a duration of a captured maneuver pattern by the dynamic, functional aggregation-based telematics circuit, wherein the measured individuated time interval is used as a time index for extracting portions of features effectively corresponding to a selected maneuver, wherein, for each captured maneuver pattern and each operation or feature, a similarity measurement of detected operation or feature profiles is performed ranging over possible trajectories used as a benchmark and highly correlated operations or features are selected;
automatically individuate captured driver maneuvers within various car trajectories provided by the matched searchable predefined measuring parameter sets of driving maneuvers and the searchable predefined measuring parameter pattern templates of driving maneuvers, wherein, for two temporal sequences, within the telematics data, a temporal-variation independent measuring parameter is measured, each time, by the dynamic, functional aggregation-based telematics circuit, wherein the dynamic, functional aggregation-based telematics circuit provides an optimal match between the two temporal sequences by non-linearly warping the temporal sequences in a time dimension, and wherein one of the two temporal sequences, which comprise two time series, is locally stretched or compressed while optimizing measured similarity between the two temporal sequences of the telematics data; and
perform machine learning based on a neural network, wherein the machine learning includes status observation and learning, wherein the status observation observes status variables of the sequential patterns of the captured telematics data, and wherein the learning performs a learning operation by linking at least one of the observed status variables of the sequential patterns to at least one of the searchable predefined measuring parameter sets of the driving maneuvers.

2. The electronic maneuver detection system in real time according to claim 1 , wherein the dynamic, functional aggregation-based telematics circuit is implemented as a dynamic, time warping-based telematics circuit, wherein, for driver maneuver identification, a matching of the sequential patterns is implemented based on dynamic time warping over the searchable predefined measuring parameter sets of driving maneuvers.

3. The electronic maneuver detection system in real time according to claim 1 , wherein the dynamic, functional aggregation-based telematics circuit is implemented as a dynamic, symbolic-aggregate-approximation-based telematics circuit or a dynamic, piecewise-aggregate-approximation-based telematics circuit, wherein, for driver maneuver identification, a matching of the sequential patterns is implemented based on dynamic, symbolic-aggregate-approximation or dynamic, piecewise-aggregate-approximation over the searchable predefined measuring parameter sets of the driving maneuvers.

4. The electronic maneuver detection system in real time according to claim 1 , wherein the machine learning includes reward computing and function updating, wherein the reward computing generates a reward based on at least one of the temporal sequences, wherein, for a temporal sequence within the captured telematics data, as observed by the status observation, a temporal variation-independent reward parameter is measured, each time, by the dynamic, functional aggregation-based telematics circuit, and wherein the function updating updates a function for deciding, from the observed status variables of the sequential patterns at present, based on the reward generated by the reward computing, at least one of the searchable predefined measuring parameter sets of driving maneuvers, and wherein the searchable predefined measuring parameter pattern templates of driving maneuvers provide an optimal match between the two temporal sequences.

5. The electronic maneuver detection system in real time according to claim 1 , wherein the machine learning is based on a convolutional neural network or a recurrent neural network or a standard backpropagation neural network.

6. The electronic maneuver detection system in real time according to claim 1 , wherein the mobile telematics devices comprise one or more wireless or wired connections and a plurality of interfaces for connecting with at least one of a motor vehicle's data transmission busses and/or a plurality of interfaces for connecting with sensors and/or measuring devices, wherein the mobile telematics devices are connected to onboard diagnostics systems and/or in-car interactive devices and/or a monitoring cellular mobile node application.

7. The electronic maneuver detection system in real time according to claim 1 , wherein t

Assignees

• Swiss reinsurance co ltd

Inventors

• Perl Andri
• BONGERS Sebastiaan
• Bassis Simone
• Apolloni Bruno

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/092

  Reinforcement learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G08G1/0112

  from the vehicle, e.g. floating car data [FCD] · CPC title