Surtrac — Pittsburgh's AI-Adaptive Traffic Signals Cut Travel Time 26% and Wait Times 41%

In 2012 the City of Pittsburgh, working with Carnegie Mellon University's Robotics Institute, installed Surtrac, an AI-based adaptive traffic-signal system, at nine intersections in the East Liberty neighbourhood, replacing fixed-timing signal plans.

Let each intersection build its own short-term signal plan from local sensor data in real time, then share plans with neighbouring intersections so clusters of vehicles move through corridors with fewer stops.

Carnegie Mellon spun out Rapid Flow Technologies to commercialise and maintain Surtrac; Pittsburgh scaled the system from 9 to roughly 50 signalised intersections by 2016, and the city has since committed over $20 million with state and federal partners to expand to around 200 intersections citywide.

The independent CMU evaluation of the original East Liberty pilot measured a 26% reduction in average travel time, a 41% cut in intersection wait times, a 31% drop in the number of stops, and a 21% reduction in vehicle emissions versus the previous fixed-time signals.

The quantified results are specific to the original 9-intersection pilot; later expansion phases have not published an equivalent independent before/after evaluation, though city officials and vendor materials still cite the pilot figures when describing today's programme.