WPI develops an app to make drunk drivers toe the line

When your smartphone tells you you’ve had enough to drink, it’s time to quit. Engineers at Worcester Polytechnic Institute say they’re developing an Android app called AlcoGait that’ll do exactly that, by keeping track of the way you walk.

“When you’re over the limit, the phone notifies you with a text message, and also it will buzz,” said Emmanuel Agu, the associate professor of computer science who’s leading the AlcoGait project. Agu has been working on the app since 2014, with assistance from a host of WPI undergraduate and graduate students.

Consumers can already buy breath-test devices that let them test their own blood alcohol level by blowing into the device. But AlcoGait is inspired by the “walk-the-line” tests that police routinely use to decide whether a driver is intoxicated.


“The most accurate biomeasure is the breathalyzer,” said Agu, “but the next most accurate is the walk test.”

With AlcoGait there’s no need to spend $50 or more for a personal breathalyzer; an Android phone alone will do the job. Also, users won’t need a human observer to grade their performance. Instead, the phone itself senses his body movements.

It’s made possible by centuries of research into human gait — the way people move. Aristotle is believed to have written the first work on how humans and animals move. But serious research on the topic began in 17th century Italy. The invention of photography in the 19th century enabled physiologists to review every detail of human gait, one frame at a time.

Today’s gait analysts use video images of humans walking to diagnose diseases, improve athletic performance, and even as a way to identify people. For example, automaker Jaguar Land Rover is working on a system of cameras that would recognize the gait of people walking past the vehicle. If the owner steps up to the car, its doors would unlock automatically.


But AlcoGait doesn’t rely on cameras. Instead, it uses the motion-detecting accelerometer and gyroscope chips found in nearly all smartphones. An AlcoGait user would activate the app, place the phone in a pocket, and take a brief walk. The app uses data from the phone’s chips to create a “fingerprint” of the user’s gait when he’s sober.

When the user wants to go party, he or she would activate AlcoGait once more. The app primarily tracks the movements of the upper body, which tend to acquire a reeling, swaying motion after too many drinks.

The app displays an estimate of the user’s blood alcohol level, based on how much his body is swaying. AlcoGait can detect the exaggerated swaying before a user’s blood alcohol level approaches the legal limit. Once the limit is reached, an alarm goes off and the phone displays a message warning the user not to get behind the wheel.

Agu said AlcoGait could be programmed to prevent the user’s car from starting when he’s intoxicated. Or it could automatically place a call to a cab company or ride-sharing service, which could give the user a safe ride home. A future version of the app will run constantly, so a user wouldn’t have to remember to activate it.

Agu tested AlcoGait with the help of students wearing “Drunk Buster” goggles — special glasses that present a distorted view of the world, causing the wearer to walk as if intoxicated. But now, Agu needs to test AlcoGait in the real world.


He’s working with researchers at Boston University and Brown University, who plan to recruit 250 volunteers who will be given alcohol in various quantities. They’ll then be tested with a standard breathalyzer and with AlcoGait, to confirm the accuracy of the app’s blood alcohol estimates.

If all goes well, Agu expects AlcoGait to be available by late 2017. “Hopefully, it’ll be ready for next New Year’s,” he said.

Hiawatha Bray can be reached at hiawatha.bray@globe.com. Follow him on Twitter @GlobeTechLab.