Germany has begun a pioneering trial of next-generation traffic lights that are controlled by artificial intelligence.
Installed in the small town of Lemgo in the country's north-west, the new lights are claimed to be capable of improving traffic flow from 10-15 per cent, at least when tested in computer simulations.
Said to be the first real-world test of AI-managed traffic lights, the Lemgo lights are capable of "deep reinforcement learning" and are said to get better with age, as its on-board AI learns how to reduce waiting times.
The lights were created by Munich-based Fraunhofer Institute for Optronics and work by using both HD cameras and radar sensors to enable the computers to understand traffic flow at the intersection.
The AI then devises a plan to optimise a signal pattern to help traffic run smoothly.
In the lab, computer simulations found the lights facilitate shorter waiting times that led to reduced localised air pollution compared to conventional lights.
But their ability to learn real-world testing should continue to improve on the initial findings.
As well as helping cars, the same technology has been developed to help pedestrians cross a busy road.
The same sensors that gauge traffic flow are also capable of counting how many pedestrians are waiting and can arrange a slot to stop traffic.
Perhaps more impressive, the same cameras can assess individual pedestrian needs. If the traffic lights spot you're in a wheelchair, or using crutches, it lengthens the time needed for you to safely make it to the other side of the road.
Researchers say the needs-based system can reduce waiting times by up to 30 per cent when there's a small crowd waiting to cross.
It's also estimated to reduce jaywalking by up to 25 per cent.