Mobility is the category with the most obvious human stakes: it's the difference between a 20-minute commute and a two-hour one, between a hillside neighbourhood that's connected to the rest of the city and one that isn't. It's also the category with the widest range of "smart" — from a single AI-tuned traffic light to a cable car built to reach communities buses never could.
Traffic engineering has been "smart" in some sense since the first coordinated signal timing plans of the 1960s and 70s — fixed schedules based on rush-hour averages. What changed from the 2010s on was feedback: signals, buses and parking spaces that report their own state in real time, and software that reacts to it instead of following a pre-set clock. Pittsburgh's Surtrac, born out of Carnegie Mellon research, is a clean example — intersections that negotiate with their neighbours second by second rather than running a fixed cycle, cutting travel times by a quarter.
The other founding idea, older than any sensor, is that mobility is also a social-equity problem, not just an engineering one. Medellín's Metrocable, opened in 2004, made that case at city scale: cable cars folded into the metro network turned informal hillside barrios — until then reachable only on foot or by informal transport — into neighbourhoods with a real commute. It remains one of the most cited urban transformations anywhere, precisely because the "smart" part was the decision to build it, not a particular piece of software.
Three patterns show up across this dataset's 80+ mobility projects. Pricing as a policy lever: Singapore's Electronic Road Pricing, running since 1998, and its many descendants (Stockholm, London, Milan) use tolling not to raise revenue first but to keep expressways in a target speed band by making congestion cost something. City-scale AI coordination: Hangzhou's City Brain ingests camera and sensor feeds across the whole city to time lights and reroute emergency vehicles, treating traffic as one optimisation problem rather than thousands of independent intersections. Mapping what already exists: in cities where formal transit doesn't cover everything, the highest-leverage "smart" move has often been documentation, not new infrastructure — researchers GPS-riding roughly 130 Nairobi matatu routes to publish the city's informal minibus network as an open map, so riders (and eventually planners) could actually see it.
These sit on a spectrum from top-down (a control room watching the whole city) to bottom-up (citizens and researchers mapping what's already there), and both produce real, measurable results — a reminder that "smart" doesn't always mean "sensor-heavy."
Expect continued growth in three directions: congestion pricing spreading to more mid-size cities as the political cost of doing so keeps falling; AI traffic coordination moving from single-city pilots to standard procurement; and informal-transit mapping and formalisation efforts expanding well beyond Nairobi, since a huge share of the world's urban trips still happen on systems no official map shows. The project list below tracks all three as they happen.