Tees Valley let an AI twin of its road network retune traffic lights itself

Tees Valley Combined Authority built a 'digital twin' — a virtual replica of its road network — that collects traffic data from sources including GPS-tracked buses and roadside sensors, and uses AI to predict where problems will occur before they snarl the network.

In its first six-month pilot phase the system cut delays by 13.7%. It also provides an automated response at 11 hotspots in the region: if traffic is held up somewhere, it is aware, and it can look at the cycle of traffic lights and adjust them. TVCA's Sean Fryer says it mitigates the everyday occurrences that usually frustrate people getting home or to work.

Darlington Council leader Stephen Harker says humans retain oversight, but the software 'is coming up with better solutions than the humans can', far more quickly, and has been 'quite clever about where it thinks transport should be diverted'. The Green Party welcomed the technology while warning it would not outpace rising car miles on its own.

The pilot produced a hard number — 13.7% less delay — within six months, rather than a promise.

It moves signal control from scheduled human retiming to continuous automated response across a whole network.

Keeping a human oversight layer is what let the council accept software making faster, better decisions than its engineers.

Where decisions are frequent, local and measurable, automate the response and supervise rather than approve — speed is the value.

Next phases of the Tees Valley Traffic Digital Twin Project will add new routes and data, including freight, active travel and environmental information.