US 12,254,724 B2
System and method for providing user feedback regarding driving behavior and the effects on gas consumption and vehicle wear
Callum Brook, Piedmont, CA (US); and Theobolt N. Leung, San Francisco, CA (US)
Assigned to QUANATA, LLC, San Francisco, CA (US)
Filed by Quanata, LLC, San Francisco, CA (US)
Filed on Jul. 27, 2020, as Appl. No. 16/940,215.
Prior Publication US 2023/0130875 A1, Apr. 27, 2023
Int. Cl. G07C 5/08 (2006.01); G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06V 20/59 (2022.01); G07C 5/00 (2006.01); H04W 4/40 (2018.01)
CPC G07C 5/0816 (2013.01) [G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06V 20/597 (2022.01); G07C 5/008 (2013.01); H04W 4/40 (2018.02)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for providing feedback, the method comprising:
receiving, by one or more processors, one or more vehicle tire images for a vehicle;
analyzing, by the one or more processors, the one or more vehicle tire images to determine driving behavior data associated with a user, the driving behavior data indicative of an impact associated with the user on longevity of the vehicle;
determining, by the one or more processors, a driving performance metric associated with the user based at least in part upon the driving behavior data by at least:
applying, by the one or more processors, the driving behavior data associated with the user to a trained machine learning model for determining the driving performance metric,
wherein the trained machine learning model comprises a neural network and was trained:
to compare features identified in the one or more vehicle tire images with template features comprising tire dimensions, tire shapes, and tire thickness metrics at one or more tire positions,
using a plurality of sets of driving behavior data associated with a plurality of users with a plurality of vehicles each having known longevity, and
using a plurality of indications of respective known longevities of the plurality of vehicles associated with the plurality of users;
wherein the trained machine learning model creates numerical representations of the features identified in the one or more vehicle tire images to generate feature vectors that are compared to feature vectors of template objects using a vector distance metric to generate a tire wear score that is used to determine the driving performance metric associated with the user; and
wherein the features are extracted from the one or more vehicle tire images by (i) detecting stable regions within the one or more vehicle tire images using a scale-invariant feature transform, (ii) locating keypoints in the stable regions, and (iii) forming a bounding box around the keypoints representing the feature; and
generating, by the one or more processors, the feedback associated with the user based at least in part upon the driving performance metric and related to the impact on longevity of the vehicle; and
providing, by the one or more processors, the feedback to the user.