Detecting Driver Phone Use Leveraging Car Speakers
A student team led by Profs. Marco Gruteser (ECE) and Richard Martin (CS), both members of WINLAB, and Prof. Yingying Chen of Stevens Institute of Technology received the best paper award at the 2011 ACM International Conference on Mobile Computing and Networking (MobiCom).
The paper "Detecting Driver Phone Use Leveraging Car Speakers", authored by Jie Yang, Simon Sidhom, Gayathri Chandrasekharan, Tam Vu, Nicolae Cecan, Hongbo Liu, Yingying Chen, Marco Gruteser, and Richard Martin addresses the problem of sensing when a smartphone is used by a driver, with particular emphasis on distinguishing between a driver and passenger.
This is a key milestone for enabling numerous driver safety and phone interface enhancements. The project developed a detection system that leverages the existing car stereo infrastructure, in particular, the car speakers and handsfree Bluetooth system.
It uses an acoustic ranging approach wherein the phone send a series of customized high frequency beeps via the car stereo. The beeps are spaced in time across the left, right, and if available, front and rear speakers. After sampling the beeps, it times their arrival via a sequential change-point detection scheme, and then uses a differential ranging approach to estimate the phone's distance from the car's center. From these differences a passenger or driver classification can be made. Experiments with two different phones and two different cars showed that our customized beeps were imperceptible to most users, robust to background noise, and achieved a classification accuracy of 90-95 percent depending on the degree of calibration.
The project also received considerable attention from the popular press. It featured into news stories on National Public Radio and was the basis of a joke in Jay Leno's Tonight Show monologue. It also was covered extensively in the MIT Technology Review, the online blog section of the Wall Street Journal, an Inside Science TV segment, CNET news, and numerous other online news services.
A video demo clip on this work can be found here