The artificial intelligence system aiming to save lives on Italian roads
A system developed by Politecnico di Milano with Unipoltec promises to predict accidents before they happen. It's called RoadSafe AI 2.0 and it could reshape urban mobility as we know it.
Every year, more than 170,000 road accidents leave a trail of grief, costs, and unanswered questions across Italy. Over 3,000 people lose their lives on the country's roads, and the social toll of this silent emergency exceeds 20 billion euros, accounting for healthcare expenses, material damage, lost productivity, and, above all, shattered lives. These are the figures elaborated by the Urban Mobility Council, and they hit hard precisely because of their raw brutality: more than 70% of accidents occur in cities, the very places where life concentrates and where people would expect to feel safest.
Faced with these numbers, the Politecnico di Milano chose not to simply observe. In collaboration with Unipoltech, the technology arm of the Unipol insurance group, researchers have developed RoadSafe AI 2.0, an artificial intelligence system that goes beyond analyzing what has already happened and instead attempts to anticipate what might happen next. The difference, in terms of potential impact on road safety, is enormous.
At the heart of the system is a predictive logic that continuously processes millions of telematic data points collected from vehicles on the road, cross-referencing them with real-time traffic information and the physical characteristics of the road network. The output is a set of dynamic risk maps capable of showing, hour by hour and zone by zone, which stretch of road is most dangerous at a given moment of the day or under specific traffic conditions. These are not static snapshots of danger, but living, constantly updated representations of urban risk, almost breathing in real time alongside the city itself.
Yet the project does not stop at mapping. According to a study carried out with the contribution of the MIT Senseable City Lab, those maps could become the foundation for concrete urban redesign interventions. Traffic islands, smart speed bumps, new sidewalks, speed limit adjustments calibrated on the actual behavior of drivers: all physical tools that, guided by artificial intelligence data, could transform the most dangerous roads into significantly safer environments.
This represents a radical shift in thinking. For decades, road safety policies have operated in a reactive mode: an accident occurs, investigators try to understand what went wrong, and then something is done about it. With RoadSafe AI 2.0, the logic is reversed. The goal becomes prevention, not treatment. And in this context, prevention translates directly into human lives that are not cut short.
The impact on the broader ecosystem of artificial intelligence applied to mobility is equally significant. This kind of system demonstrates that AI is not merely a tool for optimizing business processes or generating content, but can become a social protection infrastructure capable of reading the complexity of the real world and translating it into decisions that save lives. It is arguably one of the most concrete and urgent applications that technology can deliver today.
Open questions remain, naturally: around the privacy of collected telematic data, the governance of predictive algorithms, and the ability of public administrations to translate the system's recommendations into real and timely interventions. But the direction is set, and the potential is hard to dismiss. More than 3,000 Italian families each year might not have to mourn a son, a parent, a friend. If even a fraction of those lives could be saved by an algorithm, it would be worth every effort to bring this system to the streets of every Italian city.
Sources: Urban Mobility Council – Urban Mobility Report; Politecnico di Milano / Unipoltech – RoadSafe AI 2.0; MIT Senseable City Lab.