New AI software can detect when people text and drive

Scientists have developed a new artificial intelligence (AI) system that can accurately detect and alert drivers when they are texting or engaged in other distracting activities, and possibly help avoid accidents.

The system uses cameras to detect hand movements that deviate from normal driving behaviour and grades or classifies them in terms of possible safety threats.

Fakhri Karray, a professor at University of Waterloo in Canada, said that information could be used to improve road safety by warning or alerting drivers when they are dangerously distracted.

As advanced self-driving features are increasingly added to conventional cars, signs of serious driver distraction could be employed to trigger protective measures, he said.

"The car could actually take over driving if there was imminent danger, even for a short while, in order to avoid crashes," said Karray.

Algorithms at the heart of the technology were trained using machine-learning techniques to recognise actions such as texting, talking on a cellphone or reaching into the backseat to retrieve something. The seriousness of the action is assessed based on duration and other factors.

That work builds on extensive previous research on the recognition of signs, including frequent blinking, that drivers are in danger of falling asleep at the wheel. Head and face positioning are also important cues of distraction.

"It has a huge impact on society," said Karray, citing estimates that distracted drivers are to blame for up to 75 per cent of all traffic accidents worldwide.