Systems and methods for determining individualized driving behaviors of vehicles

Patent metadata

Field	Value
Publication number	US-11077850-B2
Application number	US-201916563789-A
Country	US
Kind code	B2
Filing date	Sep 6, 2019
Priority date	Sep 6, 2019
Publication date	Aug 3, 2021
Grant date	Aug 3, 2021

Abstract

In one embodiment, a computing system of a vehicle may capture, using one or more sensors of the vehicle, sensor data associated with a first vehicle of interest. The computing system may identify one or more features associated with the first vehicle of interest based on the sensor data. The computing system may determine a driving behavior model associated with the first vehicle of interest based on the one or more features of the first vehicle of interest. The computing system may predict a driving behavior of the first vehicle of interest based on at least the determined driving behavior model. The computing system may determine a vehicle operation for the vehicle based on at least the predicted driving behavior of the first vehicle of interest.

First claim

What is claimed is:

1. A method comprising, by a computing system of a vehicle:
capturing, using one or more sensors of the vehicle, one or more identifying features associated with a vehicle of interest located in a geographic region;
based on the one or more identifying features associated with the vehicle of interest, determining whether an individualized driving behavior model of the vehicle of interest is available for predicting behavior of the vehicle of interest, wherein the individualized driving behavior model is based on observed vehicle behavior of the vehicle of interest; and
subsequent to determining that the individualized driving behavior model of the vehicle of interest is unavailable for predicting the behavior of the vehicle of interest:
selecting, based on the geographic region where the vehicle of interest is located, a region-based driving behavior model from a plurality of driving behavior models associated with different geographic regions, wherein the plurality of driving behavior models are based on observations of vehicle behavior of vehicles in the different geographic regions,
predicting the behavior of the vehicle of interest in the geographic region based on at least the region-based driving behavior model, and
determining a vehicle operation for the vehicle based on at least the predicted behavior of the vehicle of interest.

2. The method of claim 1 , wherein the one or more identifying features associated with the vehicle of interest comprise an anonymous identifier of the vehicle of interest, and wherein the individualized driving behavior model is associated with the anonymous identifier of the vehicle of interest.

3. The method of claim 2 , wherein the anonymous identifier is generated based on a license plate of the vehicle of interest.

4. The method of claim 2 , wherein the anonymous identifier obfuscates an identity of the vehicle of interest and causes the computer system to be agnostic to the identity of the vehicle interest, and wherein the anonymous identifier anonymously associates the individualized driving behavior model to the vehicle of interest.

5. The method of claim 1 , further comprising:
determining specification information of the vehicle of interest based on an image comprising at least a part of the vehicle of interest, wherein the specification information comprises one or more dimensions of the vehicle of interest; and
determining a bounding box for the vehicle of interest based on the specification information.

6. The method of claim 5 , further comprising:
determining a future location of the vehicle of interest based at least on the bounding box of the vehicle of interest, wherein the vehicle operation is determined based at least on the future location of the vehicle of interest.

7. The method of claim 5 , wherein the specification information comprises a vehicle weight and one or more performance parameters of the vehicle of interest, and the method further comprises:
determining an acceleration envelope for the vehicle of interest based on the vehicle weight and the one or more performance parameters of the vehicle of interest; and
determining a predicted moving path for the vehicle of interest based on the acceleration envelope of the vehicle of interest.

8. The method of claim 1 , further comprising:
determining, using the one or more sensors of the vehicle, contextual data of an environment surrounding the vehicle of interest, wherein predicting the behavior of the vehicle of interest is based on at least the contextual data of the surrounding environment, and wherein the region-based driving behavior model is based on driving situations that are associated with similar contextual data to the contextual data of the surrounding environment of the vehicle of interest.

9. The method of claim 1 , wherein the vehicle operation comprises one or more of:
delaying a vehicle acceleration operation for a period of time;
executing a vehicle acceleration operation ahead of a previously planned time;
executing a speed reduction operation;
increasing a distance to the vehicle of interest;
adopting a new driving trajectory;
keeping a distance to the vehicle of interest above a threshold distance;
yielding a right of way;
sending warning signals to the vehicle of interest;
sending warning signals to a driver of the vehicle;
allocating more sensing resources for monitoring the vehicle of interest;
switching driving lanes; or
avoiding intersecting a predicted driving trajectory of the vehicle of interest.

10. The method of claim 1 , further comprising:
detecting, using the one or more sensors, an opt-out indicator associated with the vehicle of interest; and
excluding, in response to the detected opt-out indicator, the vehicle of interest from a data collection process for collecting driving behavior data related to the vehicle of interest.

11. The method of claim 1 , wherein the region-based driving behavior model is selected in response to a determination that a possibility score for the individualized driving behavior model satisfies a pre-determined threshold score.

12. The method of claim 1 , wherein the behavior of the vehicle of interest is predicted by feeding vehicle driving data associated with the region-based driving behavior model and contextual data of a surrounding environment of the vehicle of interest to a prediction model, and wherein the prediction model is trained based on comparisons between the observations of the vehicles in the geographic region and corresponding predictions for the vehicles using current weighting factors of the prediction.

13. The method of claim 1 , wherein the region-based driving behavior model is generated based on an aggregation of driving behavior data of one or more other observed vehicles that have driven in one or more other geographic regions that are within a similarity threshold with respect to the geographic region, and wherein the one or more other observed vehicles have the one or more identifying features.

14. The method of claim 1 , further comprising:
determining a similarity level between the geographic region where the vehicle of interest is located and an additional geographic region associated with an additional region-based driving behavior model;
in response to the similarity level satisfying a threshold level, selecting the additional region-based driving behavior model; and
predicting subsequent behavior of the vehicle of interest based on the selected additional region-based driving behavior model associated with the additional geographic region.

15. The method of claim 1 , further comprising:
determining a similarity level of a vehicle type of the vehicle of interest with respect to an additional vehicle type;
in response to the similarity level satisfying a threshold level, selecting a vehicle-type-based driving behavior model associated with the additional vehicle type and the geographic region where the vehicle is located; and
predicting subsequent behavior of the vehicle of interest based on the selected vehicle-type-based driving behavior model associated the additional vehicle type and the geographic region.

16. One or more non-transitory computer-readable storage media including instructions that are operable, when executed by at least one processor of a computing system of a vehicle, to cause the computing system to:
capture, using one or more sensors of the vehicle, one or more identifying features associated with a vehicle of interest located in a geographic region;
based on the one or more identifying features associated wi

Assignees

• Lyft Inc

Inventors

• Green Logan
• Vincent Luc
• Matthiesen Taggart
• Qureshi Sameer

Classifications

• G06V20/625

  License plates · CPC title

• G06V10/25

  Determination of region of interest [ROI] or a volume of interest [VOI] · CPC title

• B60W30/0956 Primary

  the prediction being responsive to traffic or environmental parameters · CPC title

• G06V20/584

  of vehicle lights or traffic lights · CPC title

• G06V2201/08

  Detecting or categorising vehicles · CPC title