US 12,282,963 B1
Automatedly generating and issuing accident data gathering recommendations following a vehicle accident
Angelica Nichole White, Spring Branch, TX (US); Sean Carl Mitchem, San Antonio, TX (US); Sean Michael Wayne Craig, San Antonio, TX (US); Subhalaskshmi Selvam, Allen, TX (US); Christopher Russell, The Colony, TX (US); Brian Howard Katz, San Antonio, TX (US); and Roberto Virgillio Jolliffe, San Antonio, TX (US)
Assigned to United Services Automobile Association (USAA), San Antonio, TX (US)
Filed by UIPCO, LLC, San Antonio, TX (US)
Filed on Feb. 8, 2022, as Appl. No. 17/667,300.
Int. Cl. G06Q 40/08 (2012.01)
CPC G06Q 40/08 (2013.01) 18 Claims
OG exemplary drawing
 
1. A method of automatedly issuing accident data recommendations for a vehicle accident occurrence, the method comprising:
detecting, for a vehicle, an occurrence of a vehicle accident;
in response to the detecting, providing an instruction to an individual to (a) obtain one or more first images of damage at a scene of the detected accident and (b) provide one or more language descriptions of the scene of the detected accident;
in response to the detecting, automatically retrieving initial accident data for the detected vehicle accident by retrieving a reading from at least one sensor, of the vehicle, comprising a proximity sensor, distance sensor, camera, and/or pressure sensor;
generating one or more accident data recommendations by:
converting, into input for a machine learning model, the one or more first images, the one or more language descriptions, and the initial accident data, wherein,
the one or more first images are converted to image values, the one or more language descriptions are converted into one or more semantic embeddings, and the initial accident data is converted into values representing sensor readings; and
the machine learning model is trained to predict one or more accident characteristics for the detected vehicle accident using training data comprising initial accident data for prior vehicle accident occurrences matched to accident characteristics, the initial accident data of the training data comprising images, language descriptions, and vehicle sensor readings;
applying the input to the machine learning model to obtain the accident characteristics for the detected vehicle accident, one or more accident characteristics comprising a vehicle damage type;
selecting the one or more accident data recommendations respectively mapped to the accident characteristics, the recommendations comprising c) generating one or more second images corresponding to one or more of the respective accident characteristics, and d) one or more of generating an oral and/or video accounting of the detected vehicle accident, obtaining a witness statement, determining whether any utility property was impacted as a result of the detected vehicle accident, or any combination thereof;
providing, to the individual involved in the detected vehicle accident via at least an augmented reality device, the selected accident data recommendations and receiving additional accident data, wherein the providing includes:
causing the augmented reality device to display an augmented reality overlay that provides guidance for obtaining the one or more second images;
compiling the initial accident data and the additional accident data into a vehicle accident package that represents a state of the scene of the detected accident; and
maintaining the machine learning model by updating the training of the machine learning model using at least the obtained one or more accident characteristics.