Researchers at Cornell and Stanford University believe they have found a way keep drivers from getting into car accidents before they happen.
The solution comes in the form of a new face-tracking system made up of cameras inside the car monitor that watch subtle movements of the driver's head and body to figure out if he/she is going to turn or change lanes, according to CNN Money. These cameras, along with sensors outside the car, use information they collect to warn the driver that they are about to get into a crash.
Ashutosh Saxena from Cornell University led a team of researchers to develop the system by recording video of 10 drivers traveling about 1,200 miles through cities and highways over a two-month period. The team also recorded video looking forward from the cars while they drove.
After learning to match certain head and body movements with imminent driving maneuvers, the technology was tested with different drivers and was able to predict what they would 77.4 percent of the time, CNN Money reported. The system was also able to predict turns and lane changes more than 3.5 seconds before they happened.
The researchers said the technology still has its flaws, as the system can sometimes be confused by passing shadows and interactions between the driver and passengers. Different physical cues can also throw the system off in different situations, such as when the driver turns from a turn-only lane.
The system is currently in its early stages of development, which presents the opportunity for other technology to be included, such as more advanced cameras and sensors that can detect the driver's movements from the steering wheel and pedals, CNN Money reported.
Saxena said his team is currently in talks with automakers, which, if they go through, could provide a new tool for making roads safer.
