Tech to detect when a driver is dozing off

By Heather Kelly, CNN

While Google, universities and car companies work on perfecting self-driving vehicles, flawed and sometimes sleepy human drivers still fill our roads.

But new technology could help detect when those drivers start to feel tired and possibly prevent dangerous accidents. A research project at the University of Leicester has combined eye-tracking and brain monitoring to calculate when a driver's alertness starts to wane.

    Researchers have used the two tracking technologies on their own before, but Dr. Matias Ison, who led this project, said they've found a new way to combine them for more accurate information about a person's state of mind.

    "There are a variety of behaviors that are related to sleepiness and distractions," said Dr. Ison. "Some of them, such as blinking more frequently, changing our eye movements’ pattern, or not fixating on the road ahead are well suited to be detected with an eye tracker. However, brain activity changes during sleepiness and low cognitive alertness state can only be detected with an EEG."

    To detect eye movements, researchers aim an array of LED lights at the subject's eyes. An infra-red camera sensor detectes the reflection of the lights off of the eyes.

    They tracked brain activity using an electroencephalograph, or EEG. A cap covered in electrodes detects what part of the brain are active. A computer can tie that data to specific types of thoughts, and together with eye-movements, the information collected can tell a computer system if a person is getting sleepy.

    Currently still in the bulky test stages, engineers could streamline the technology for use outside of the lab in three to five years.

    "We´re only starting to understand what our brain does under real-life situations, where we are constantly moving our eyes at a rate of 3-4 times per second," said Dr. Ison.

    According to the National Highway Traffic Safety Administration, there are more than 100,000 crashes, 40,000 injuries and 1,550 deaths a year in the United States as a result of drivers nodding off at the wheel.

    Distracted driving is also a hurdle for researchers working on self-driving cars. The first wave of automated vehicles won't completely take over the job of driving. Early features will focus on specific tasks like getting through traffic or fitting into tight parking spots. The problem with semi-automatic driving, though, is that when it's time to hand the controls back to humans, the driver will need to be alert and ready to steer.

    Grabbing their attention could be much easier if the car has more information about their current state, possibly avoiding turning over the car to a driver before they are ready.

    The experiments were part of a 14-month research project at the university's Centre for Systems Neuroscience. It has potential interesting applications beyond just cars.

    "This could help people with severe motor disabilities, bringing tremendous benefits in terms of communication and motor control," said Dr. Ison.

    It could also be used for hands-free video games in which players control the action with their eye movements and thoughts, as well as diagnosing dyslexia.