This paper discusses a novel algorithm to automatically identify the position of a smartphone inside a moving vehicle, so as to detect whether it is being used by the driver or just a passenger of the car. This detection has applications to the prevention of distracted driving and can be used to automatically disable phone features such as texting when the phone is located in the driver’s seat. The challenges in the smartphone localisation problem come from the need to entirely use only accelerometers and gyroscopes already available on a typical phone, and the need to allow for any unknown 3-dimensional orientation of the phone while being carried by the driver or passenger of the car. First, the phone’s real-time orientation is determined by identifying the vehicle’s longitudinal and vertical axes in the phone reference frame. This provides the rotational matrix for conversion of accelerations and angular velocities measured on the phone to accelerations and angular velocities about the car axes. Next, the front-to-back pitching dynamics of the car during deceleration and the side-to-side roll dynamics during turning are characterised to detect whether the phone is in the driver’s seat. The characterisation of the roll and pitch dynamics are formalised using cross-covariances of sensor signals and a machine learning algorithm. Both simulations and extensive experiments are used to show that the developed system can accurately determine if the phone is being carried by the driver. The developed technology can be extremely useful for iPhones and other smartphones which can currently only detect whether the phone is on a moving car, but cannot detect whether it is being used by a driver or a passenger.
Bibliographical noteFunding Information:
This research was supported in part by a research grant from the National Science Foundation, NSF [grant number PFI-1631133].
- distracted driving
- phone location
- smartphone sensors
- support vector machine