Communications of the ACM
Smartphones Hit the Road For Distracted-Driving Research
The windshield-mounted, custom-build monitor records images and data as soon a vehicle starts to move.
Credit: UAB
Few willingly admit their indiscretions behind the wheel, particularly when they involve sneaking a peek at an incoming text or checking Facebook when traffic slows down. Maybe once in awhile, they say, but nothing like those jerks all around me on the interstate!
For several years, Despina Stavrinos, director of the Translational Research for Injury Prevention Laboratory at the University of Alabama at Birmingham, has been using a detailed driving simulator to study how participants respond to distraction. This past summer, however, she began taking her research on the road using custom-built, smartphone-based driving monitors. The studies are designed to offer particular new insight on the behavior of teens and seniors, the two groups most likely to be involved in motor-vehicle accidents.
"We wanted to use existing technology," says Stavrinos, an assistant professor in the UAB College of Arts and Sciences Department of Psychology. "The problem is that it costs thousands of dollars to lease the equipment for several months for research purposes, which is not feasible in today's funding climate." Another shortcoming: The devices rely on accelerometers — highly sensitive movement sensors — to decide when to record data. Waiting for signals such as hard braking or acceleration is smart when you're interested in saving memory. That's why the devices can record for up to a year. "But that's not the best way to study something like distracted driving," Stavrinos says.
So she built her own. Along with colleague and co-principal investigator Lesley Ross, and supported by a faculty development grant from the UAB Faculty Senate, subsequent grants from the U.S. and Alabama Departments of Transportation, and the UAB Edward R. Roybal Center for Translational Research in Aging and Mobility, Stavrinos purchased 20 top-of-the-line smartphones — the HTC Evo model — and worked with a developer to design a custom app to collect the data she needed. She also worked with students in the UAB School of Engineering to design a case that shields the phones from the elements and conceals them from potential thieves.
The system uses both cameras on the phone — one pointed at the driver, the other at the road ahead. The app records images from both cameras simultaneously. It time-stamps the photos so researchers can match the action outside the vehicle with the driver's actions behind the wheel. Wide-angle lenses are attached to the phone's stock cameras to improve the field of view. The device plugs into the car's cigarette lighter for continuous power.
"The app uses the accelerometer to start as soon as you move," Stavrinos says. "It records an image every second, and automatically turns off when the car has been motionless for five minutes." The unit's GPS also records where drivers have been, how many trips they take in a day, and the routes they have chosen to get there. The team is particularly interested in these data for the older drivers due to reports that novice and older drivers will go to great lengths to avoid left-hand turns — because they don't feel comfortable turning in front of oncoming traffic, Stavrinos says. "Our preliminary data supports this claim; we've seen people make as many as six right-hand turns to avoid turning left."
What, Me Worry?
But surely, won't drivers be on their best behavior when they know a camera is monitoring them at all times? Actually, previous studies have shown that, within a day of the cameras being installed, drivers forget about them and revert to their normal driving habits. "We're confident that we can capture a driver's real behavior over two weeks," Stavrinos says.
Stavrinos and her colleagues are still evaluating the data from their initial pilot study, which involved 120 drivers. In addition to the driving images and other information, the researchers also collected self-reports from the drivers; they plan to compare the two data sets to verify the accuracy of the self-reports, which are currently the principal tool used to study driving behavior. The researchers also gave participants an extensive battery of cognitive and physical assessments before they began. They plan to analyze how participants' performance on these tests predict their performance in the driving study.
Meanwhile, the researchers are working to overcome challenges identified in their pilot study, including the units' propensity to shut down on very hot days. (The cellphones have an internal thermometer that automatically powers the machines off to preserve their circuitry when temperatures get too high.) Small fans built into the protective case around the phones could be one option to fix this, Stavrinos says.
The researchers have also looked into the commercial possibilities of their devices. As part of the study, they asked participants — seniors, teens, and the teens' parents — if they would be willing to have such a device installed permanently. Many of the teens weren't thrilled, predictably. But "the older adults were more open to allowing others to monitor their driving," Stavrinos says.
"There is enormous potential here."
No entries found