Automatic vehicle crash detection using onboard devices

• US 9,392,431 B2
• Filed: 09/30/2014
• Issued: 07/12/2016
• Est. Priority Date: 09/30/2014
• Status: Active Grant

• Current Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Original Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Inventors: Barfield, James Ronald Jr., Welch, Stephen Christopher
• Primary Examiner(s): AJIBADE AKONAI, OLUMIDE
• Application Number:
  US14/502,423
• Publication Number:
  US 20160094964A1
• Time in Patent Office: 651 Days
• US Class Current: 1/1
• CPC Class Codes: G07C 5/008 communicating information to a remotely located station transmission systems for measured values G08C
• CPC Class Codes: G07C 5/085 using electronic data carriers
• CPC Class Codes: G08B 25/016 Personal emergency signalling and security systems emergency non-personal manually actuated alarm activators G08B25/12
• CPC Class Codes: G08B 29/185 Signal analysis techniques for reducing or preventing false alarms or for enhancing the reliability of the system
• CPC Class Codes: G08G 1/205 Indicating the location of the monitored vehicles as destination, e.g. accidents, stolen, rental
• CPC Class Codes: H04Q 9/00 Arrangements in telecontrol or telemetry systems for selectively calling a substation from a main station, in which substation desired apparatus is selected for applying a control signal thereto or for obtaining measured values therefrom
• CPC Class Codes: H04W 4/44 for communication between vehicles and infrastructures, e.g. vehicle-to-cloud [V2C] or vehicle-to-home [V2H]
• CPC Class Codes: H04W 4/70 Services for machine-to-machine communication [M2M] or machine type communication [MTC]
• CPC Class Codes: H04W 4/90 Services for handling of emergency or hazardous situations, e.g. earthquake and tsunami warning systems [ETWS]

Abstract

Vehicle collisions may be automatically detected and reported based to a call center. The collisions may be automatically detected based on a collision detection model that receives sensor data, or other data, as input, and outputs an indication of whether there is a collision. The collision detection model may be trained on historical sensor data associated with potential vehicle collisions, where the historical sensor data is labeled to indicate whether the data corresponds to an actual collision.

Claims

1. A device comprising:

an accelerometer to measure acceleration of the device; and

processing circuitry to;

calculate, based on the measured acceleration of the device, an acceleration magnitude value,

compare the acceleration magnitude value to a threshold,

determine, when the comparison indicates that the acceleration magnitude value satisfies the threshold, at least one feature based on a plurality of acceleration magnitude values, the at least one feature including an N-point pre max sum calculated as a sum of a plurality of acceleration magnitude values corresponding to a window of acceleration magnitude values centered around an acceleration magnitude value that occurs before the acceleration magnitude value that satisfied the threshold,

evaluate, when the comparison indicates that the acceleration magnitude value satisfies the threshold, a collision detection model based on the determined at least one feature, the evaluation of the collision detection model indicating whether the device is involved in a vehicle collision, and

alert an emergency call center when the evaluation of the collision detection model indicates that the device is involved in the vehicle collision.

2. The device of claim 1, wherein the determination of the at least one feature includes determining a plurality of features that are additionally determined based on one or more of:

data received from an On-Board Diagnostics (OBD) port of a vehicle, or

sensor data relating to a gyroscope, a barometer, a compass, or a speedometer.

3. The device of claim 1, wherein the device further includes:

an On-Board Diagnostics (OBD)-II connector to connect to an OBD-II port of the vehicle.

4. The device of claim 1, wherein the at least one feature additionally includes an N-point post max sum calculated as a sum of a plurality of acceleration magnitude values corresponding to a window of acceleration magnitude values centered around an acceleration magnitude value that occurs after a particular acceleration magnitude value that satisfied the threshold.

5. The device of claim 1, wherein the at least one feature additionally includes an N-point pre max sum calculated as a sum of a plurality of acceleration magnitude values corresponding to a window of acceleration magnitude values centered around a maximum acceleration magnitude value that occurs within a first predetermined number of samples prior to a minimum acceleration magnitude value that occurs within a second predetermined number of samples prior to an absolute maximum acceleration magnitude value that occurs within a third predetermined number of samples from a particular acceleration magnitude value that satisfied the threshold.

6. The device of claim 1, wherein the at least one feature additionally includes a maximum acceleration magnitude value calculated in a window of acceleration magnitude values centered around a particular acceleration magnitude value that satisfied the threshold.

7. The device of claim 1, wherein the collision detection model includes a collision detection model based on a binary classification tree.

8. The method of claim 1, further comprising:

wirelessly transmitting, when the comparison indicates that the acceleration magnitude value satisfies the threshold, the at least one feature to a server device.

9. A method comprising:

determining, by a telematics device, three-axis acceleration measurements associated with the telematics device;

calculating, by the telematics device and based on the measured three-axis acceleration measurements, acceleration magnitude values;

determining, by the telematics device, that the acceleration magnitude value satisfies a minimum threshold value;

determining, by the telematics device and based on the determination that the acceleration magnitude value satisfies the minimum threshold value, at least one feature based on a time series of the acceleration magnitude values, the at least one feature including an N-point pre max sum calculated as a sum of a plurality of acceleration magnitude values corresponding to a window of acceleration magnitude values centered around an acceleration magnitude value that occurs before the acceleration magnitude value that satisfied the threshold;

evaluating, by the telematics device and based on the determination that the acceleration magnitude value satisfies the minimum threshold value, a collision detection model based on the determined at least one feature, the evaluation of the collision detection model indicating whether the telematics device is associated with a vehicle collision; and

alerting, by the telematics device, an emergency call center when the evaluation of the collision detection model indicates that the telematics device is associated with a vehicle collision.

10. The method of claim 9, wherein the determination of the at least one feature includes determining a plurality of features based on one or more of:

data received from an On-Board Diagnostics (OBD) port of a vehicle, or

sensor data relating to a gyroscope, a barometer, a compass, or a speedometer.

11. The method of claim 9, wherein the at least one feature additionally includes an N-point post max sum calculated as a sum of a plurality of acceleration magnitude values corresponding to a window of acceleration magnitude values centered around an acceleration magnitude value that occurs after a particular acceleration magnitude value that satisfied the threshold.

12. The method of claim 9, wherein the at least one feature additionally includes a maximum acceleration magnitude value calculated in a window of acceleration magnitude values centered around a particular acceleration magnitude value that satisfied the threshold.

13. The method of claim 9, wherein the evaluation of the collision detection model indicates a severity level of the vehicle collision.

14. The method of claim 9, wherein the collision detection model includes a collision detection model based on a binary classification tree.

15. A method comprising:

receiving, by one or more computing devices, sensor data generated by sensors associated with a plurality of vehicles;

extracting features from the sensor data, the features including features based on a time series of acceleration magnitude values associated with the plurality of vehicles, at least one of the features including an N-point pre max sum calculated as a sum of a plurality of the acceleration magnitude values corresponding to a window of acceleration magnitude values centered around an acceleration magnitude value that occurs before a particular acceleration magnitude value;

associating, by the one or more computing devices, the sensor data with sets of data corresponding to potential vehicle collisions, each of the sets of data being associated with a label indicating whether the set of data corresponds to an actual vehicle collision or a non-collision; and

training, by the one or more computing devices and based on the labeled sets of data, a classification model to distinguish between sensor data corresponding to vehicle collisions and sensor data corresponding to non-collisions.

16. The method of claim 15, wherein associating each of the sets of data with a label further comprises:

performing automated keyword analysis on call center logs to associate each of the sets of data with the label.

17. The method of claim 15, wherein the sensor data corresponds to sensor data that is received in response to automatic detection of potential collisions at the plurality of vehicles.

18. The method of claim 15, wherein the sensor data includes data sensed by an accelerometer, a barometer, a magnetometer, or a gyroscope.

19. The method of claim 15, wherein the label associated with each of the sets of data indicates a severity of the vehicle collision.

20. The method of claim 15, wherein the classification model includes a classification model based on a binary classification tree.