Of the 7,000 fatalities each year in the USA caused by car accidents at intersections, 700 were caused by drivers running red lights. This staggering percentage has inspired a team at MIT to develop a system to try and reduce those accidents, by warning other drivers approaching the intersection of the potentially impending collision. The algorithm uses a combination of a car's speed, deceleration, and distance from the light to work out the likelihood that the the driver will be able to stop in time, all calculated within five milliseconds. This is then communicated to other cars via vehicle-to-vehicle (or V2V) communications — technology that the Department for Motor Vehicles is currently investigating in partnership with Ford, amongst other manufacturers, though this is still nowhere close to production.
When your car receives this warning, rather than slowing itself automatically, the researchers suggest that a warning light could be added to a heads-up display within the car, giving the driver foresight of the potential collision. The weak link in the system is obviously the need for more cars to be added into the network for it to be truly effective, and it remains to be seen whether peer-to-peer connections are the way forward or whether a hub-like system could at least work as a stop-gap solution.