A team from Rutgers University and Stevens Institute of Technology have designed and tested a smart phone application that can pinpoint where in the car a cell phone user is sitting—i.e., on the driver’s side or the passenger’s side—and then take steps to reduce distractions if its user is a driver.
For example, it can silently forward incoming calls and texts to message boxes for later retrieval. It could also respond automatically to a caller or texter, saying that the owner is currently driving and will reply later. Or it could offer to put a voice call through if a caller or texter indicates the matter is urgent.
For outgoing communication, the app could disable texting and make placing certain calls less difficult, perhaps by offering a short list of frequent contacts shown as large on-screen buttons.
The National Highway Traffic Safety Administration estimates that 3,000 fatal traffic accidents nationwide last year were the result of distracted driving. Studies have found that one in 20 traffic accidents involve a driver talking on a cell phone and that talking even while using a hands-free device carries as great a delay in reaction time as having a blood alcohol concentration of .08, the legal limit.
Earlier suggestions on how technology could fight the problem, such as measuring how fast a cell phone is moving and cutting off conversations above a certain speed, were dismissed as overreaching.
The trouble was, that would cut you off if you were a passenger in a car or if you were riding on a train.—Marco Gruteser, Rutgers associate professor of electrical and computer engineering
Gruteser and his colleagues devised a way for a cell phone to work with a car’s sound system to distinguish between the driver and passenger. It requires a stereo sound system with Bluetooth connectivity—a capability working its way into the mid-priced car market.
|Flow of the detection algorithm. Source: Yang et al. Click to enlarge.|
The phone generates high-pitched beeps and transmits them to the car stereo over the Bluetooth connection. The beeps are spaced in time across the left, right, and if available, front and rear speakers. After sampling the beeps, the app uses a sequential change-point detection scheme to time their arrival, and then uses a differential approach to estimate the phone’s distance from the car’s center. From these differences, the app makes a “passenger” or “driver” classification.
A car with four-channel audio can perform the check more accurately, and may one day even be able to distinguish between front- and back-seat phone users.
The concept, while simple, had to prove itself in the cabin of a moving car, where acoustics are far from perfect; the app achieved a 90% success rate, said Rich Martin, associate professor of computer science in the Rutgers School of Arts and Sciences, who is also, like Gruteser, a member of the university’s Wireless Information Network Laboratory (WINLAB). Classification accuracy rose to 95% with some calibrations. They also found a low false positive rate, on the order of a few percent.
The team wrote its initial app to run on an Android device and plans to develop one for the iPhone. The concept merited a best paper award at last year’s MobiCom, a leading academic and professional conference for mobile computing and wireless networking technology.
The engineers hope their demonstration spurs cell phone makers to pursue commercial development of the concept.
Contributing to the research from Rutgers were Gayathri Chandrasekaran, Tam Vu and Nicolae Cecan; and from Stevens Institute of Technology, Jie Yang, Simon Sidhom, Hongbo Liu and Yingying Chen. The work was funded by the National Science Foundation.
Jie Yang et al. (2011) Detecting Driver Phone Use Leveraging Car Speakers. MobiCom ’11