Investigation of Intra-vehicle Location Finding using Inexpensive Sensors, 10-R8222
John G. Whipple
Ted R. Wilmes
Brian K. Anderson
Inclusive Dates: 04/04/11 – 08/04/11
Background — Distracted driving is a growing problem that expands with the increasing role that smartphones play in day-to-day activities. Many solutions exist that attempt to curb this problem by first detecting when the user is in a moving vehicle and then disabling distracting functionality. These solutions lack the ability to determine if the smartphone belongs to the driver or a passenger and thus, every smartphone in the vehicle has reduced functionality. This research focused on finding a solution to this problem by data mining the accelerometers that already exist in smartphones.
Approach — Vehicles generate forces that act on objects differently based on their position in the vehicle. Accelerometer sensor technology can be used to detect these variations. This research began by using SwRI's test track to record driving actions such as turning, accelerating and braking from several different vehicles using a custom accelerometer recording device. Multiple instances of the device were placed in each vehicle to facilitate recording acceleration data from multiple locations within the vehicle at the same time. This data was then analyzed using statistical software to extract features that could be used to determine driver or passenger. Digital signal processing was used to filter and identify relevant features of the data. Machine learning techniques such as neural networks and genetic algorithms were then used to model the data. A classification of "driver" or "passenger" could then be determined by feeding the models previously unseen data.
Accomplishments — This project confirmed that the accelerometer sensor can be used to deduce its own location within a mobile frame of reference. This solution does not rely on any external infrastructure and is not limited to automobiles. The same process that was used to develop the automobile models could be applied to container ships, passenger ships, train cars and tractor trailers. The wide applicability of the solution has prompted the investigators to file a patent application.