Collision warning system works in tune with driver attention

Researchers are developing a collision warning system that takes into account driver attention to give appropriate warnings.

A team at Leeds University headed by Dr Natasha Merat and Prof Oliver Carsten has been awarded nearly £500,000 by the EPSRC for a project called FORWARN: Towards an intelligent Forward Collision Warning System.

‘What we’re trying to create is a forward collision warning [FCW] system that would know when the driver has detected the lead vehicle or obstacle and would also know when the driver hasn’t — so we would reduce unnecessary and annoying warnings for the driver,’ Carsten told The Engineer.

FCW systems use sensors and radar to scan the area ahead of the vehicle and aim to avoid rear-end collisions, or reduce their impact, by advising drivers to brake. Some newer systems even intervene in some cases to avoid a collision.

There are considerable potential benefits of such systems, which have recently been proven in a large-scale field operational test in North America. However, there is a danger that systems that have permanently fixed criteria will be viewed by a significant number of drivers as presenting too many ‘false’ or unwanted warnings. Indeed, in those US tests the drivers were keen to be able to tune the system to their personal preferences.

‘It’s a question of driving style,’ said Carsten. ‘We did an experiment previously where we showed that if we adapted the [FCW] system to somebody’s habitual reaction times, then the people who had very short reaction times and liked breaking late got fewer warnings, and that improved their trust in the system — we were able to increase acceptance without any real decrement in observed safety.’

The challenge now is to develop a system that can do this automatically through FORWARN. One of the methods will be to use infrared eye tracking in conjunction with Leeds University’s state-of-the-art vehicle simulator facility.

‘They [the eye tracking systems] are well used and robust in experimental situations, but they’re not quite out there yet in a package that will work reliably in the real world. That’s going to be coming soon, so we’re trying to look ahead to when that potential exists,’ said Carsten.

He envisages a system where, upon approach to a hazardous condition, an intelligent FCW system will only be triggered after if it has ascertained that the driver is truly distracted and unable to respond to the hazard in good time. There may, however, need to be a period of calibration to account for exact driver- and vehicle-related metrics, Carsten added.