If you’ve ever been frustrated by stop-and-go traffic, you might have thought that traffic lights just don’t “get” what’s going on around them... and you’d be right. Traffic lights are programmed based on typical traffic patterns for the time and location, but are unaware of what’s actually happening at any one place or time (this wouldn’t include pedestrians hitting walk light buttons, or stopped cars activating sensors embedded in the asphalt). Not only is stopping and waiting for red lights irritating, but it is also a huge source of wasted fuel and extra CO2 emissions. Now, however, researchers have come up with something that may greatly reduce drivers’ periods in the “red light zones” – a system that allows traffic lights to monitor traffic in real time, and coordinate their signals accordingly.

Stefan Lämmer at the Dresden University of Technology and Dirk Helbing of ETH Zurich made a computer model of Dresden’s roads, in which the traffic streams flowed and merged not unlike water going through pipes. They then equipped the virtual traffic lights on those roads with sensors that monitored the local traffic flow. Using this input, each light calculated the expected number of immediately oncoming vehicles, and figured out how long it would have to stay green in order to let that traffic through.

The lights also communicated with each other and adjusted their timing based on what the lights up- and downstream were doing. There would be no sense in one light letting most of the traffic through, if it were all just going to be held up at the next one. Despite the fact that the resulting signal pattern appeared to be random and chaotic as compared to a pre-programmed pattern, the end result was a reduced waiting time of 10 to 30 percent.

Such a system would not only reduce traffic jams, but would also eliminate situations such as drivers having to wait at empty intersections, or lights cycling unnecessarily when no traffic is present.

Lämmer and Helbing are now working with a German traffic management agency to implement their system on real roads.