Vehicle acceleration estimation using smartphone-based sensors

Abstract

Recent advances in smartphone technology, including motion sensing and wireless communications, have resulted in these devices being used for vehicle-based driver behaviour sensing applications, replacing existing bespoke vehicle-based solutions. Acceleration is normally used as the primary indicator for recklessness. Despite the many benefits of using a smartphone to determine vehicle acceleration, the mobility of the phone relative to the vehicle, and the vehicle relative to the earth, causes the earth's gravitational force to obscure the true vehicle acceleration as perceived by the phone. The design and test results in this paper demonstrate how quaternions and an unscented Kalman filter can be used to remove the gravitational vector from the sensed acceleration, which enables reckless driving detection.