Browse Prior Art Database

System and Method for Automatically Detecting and Recovering Stolen Automobiles Using User Activity Mining

IP.com Disclosure Number: IPCOM000248515D
Publication Date: 2016-Dec-12
Document File: 3 page(s) / 51K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is an automatic detection and recovery system for stolen vehicles. Using machine learning, the system identifies and stores user behaviors, and automatically can determine that the current driver is an unauthorized user by referencing the stored driving habits of the original/correct owner(s)/user(s).

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 50% of the total text.

1

System and Method for Automatically Detecting and Recovering Stolen Automobiles Using User Activity Mining

Currently, when a vehicle is stolen, the recovery efforts start after the owner reports that the vehicle is stolen.

The novel contribution is an automatic detection and recovery system for stolen vehicles. The system can automatically determine that the current driver is an unauthorized user by mining and storing the driving habits of the original/correct owner(s)/user(s).

The core novelty of this system and the associated methods are the ability to learn a user profile through observation the actions that a set of users takes. The system applies a machine learning model to discriminate those users from a large number of other users. When a user drives the car, the model computes the probability that the user is not in the set of pre-computed users. If the probability is beyond a threshold, then the system tries to authenticate the user through some other method. If the user fails such authentication, then the system executes a preconfigured action (e.g., stopping the car, notifying the user/owner, notifying law enforcement etc.)

The novel system assumes the following technical components/assets/capabilities to exist as necessary background:

• The ability to setup user profiles in a car

• The ability to collect information from various sensors including connectivity to other mobile devices

• Ability to detect location (e.g., with a global positioning system (GPS))

• System to upload information to central system from the car and download trained machine learning models into the car

• System to prompt the user for input

• System to stop the car and report its location to a central server

When a user procures a new vehicle, the system begins keeping a log of the user’s/ owner’s idiosyncrasies and driving habits. These idiosyncrasies could include (but are not limited to):

• Driving patterns (e.g., sudden breaking, average speed, etc.)

• Refueling patterns (e.g., favorite refueling stations, etc.)

• Sequence of actions (e.g., starts the car before putting on the seatbelt, immediately turns air conditioning on high setting, etc.)

• Connection of various devices (e.g., Bluetooth* device address)

• Driving locations (e.g., office, home, etc.)

When the system determines that the vehicle has a high probability of possession by an unauthorized user (i.e., the vehicle is stolen) it requires the driver to self-identify through some other ways (e.g., a passcode sent to a phone, password, fingerprint, etc.).

2

At a high level, the implementation methods follow: 1. Set up user profiles, collect information from various sensors in an automobile,

and store these sensor inputs for each user 2. Send the sensor information corresponding to each user to a central server 3. Build a machine learning model to discriminate each user from other users, using

the given sensor information and the ability to send the trained model back to the automobile

4. Apply t...