Indian Scientist Develops Algorithm That Can Predict Driving Error

Here’s some respite for drivers who tend to run over unwary people sleeping on curbs.

Professor Ashutosh Saxena, Professor at Cornell University has developed a program that can predict if you're going to commit an error while driving.


May 09th 2015 A-A+

The program is based on the on-board computer guessing the driver’s intention based on head movements and watching the road ahead.

Integrating driver anticipation with radar or camera inputs to locate other vehicles, the car's safety system could warn the driver when the anticipated action could be dangerous. In addition to this, the system could also alert the driver with a warning message before making an illegal turn.

The system works on the principle that the computer is fed with face detection and tracking software, which identifies head movements and associates it with lane changes and turns.         

Video inputs from ten drivers, who drove over 1888 kilometers of freeways and city lanes was fed into the computer.

In a test against another data set of videos with different drivers, the system correctly predicted the driver's actions 77.4 percent of the time, anticipating an average 3.53 seconds in advance.

About the Inventor

Ashutosh, a PhD from Stanford University, was voted among eight inventors to watch out for in 2015 by the Smithsonian Institute. He also won the Google Faculty Research Award in 2012.