AI in Urban Mobility: Autonomous Vehicles and Smart Infrastructure (Towards the Synchronized City)
Imagine a city without traffic jams, where traffic lights "see" cars and direct them in real time. It's not a utopia: it's Smart Urban Mobility. The integration
Imagine a city where the traffic light turns green exactly at the moment you arrive at the intersection. Where your car communicates with the one ahead to avoid a sudden brake. Where you don't look for parking, because the parking spot finds you (or your car drops you off at the entrance and goes to park itself).
This is not a scene from Minority Report. It is the promise of Smart Urban Mobility, a sector where Artificial Intelligence doesn't just drive vehicles, but orchestrates the entire city flow. According to the most recent estimates, the integration of AI and V2X (Vehicle-to-Everything) systems can reduce urban congestion by 30-40% and cut emissions by 20%. Not by building new roads, but by using the ones we have better.
In this article for the AI for Sustainability column, we will analyze how AI is transforming passive infrastructures into active "brains," explore the digital twins of Milan and Bologna, and see how autonomous vehicles (AVs) are moving from science fiction to real-world logistics.
1. The Problem: Analog Cities in a Digital World
Our cities were designed in the 19th or 20th century. The streets are rigid, traffic lights follow pre-programmed timers, and planning decisions are based on years-old census data. The result? Traffic, pollution, and inefficiency.
The Invisible Orchestrator
AI intervenes as an orchestra conductor. It's not just about having smart cars; it's about having infrastructure that "sees" and "predicts." As highlighted by Cultura Digitale (culturedigitali.eu), AI for Traffic Management doesn't react to traffic; it anticipates it. By analyzing data from cameras, sensors in the asphalt, and smartphone GPS, Machine Learning algorithms can predict a traffic jam 60 minutes before it forms and modify traffic light timings in real time to smooth the flow.
Beyond Autonomous Driving: V2X Connectivity
The real qualitative leap, described in our in-depth look at AI and Future Mobility: Autonomous Driving, is V2X (Vehicle-to-Everything) communication. The car doesn't just use its own sensors (Lidar/Radar), but receives data from a blind intersection ("Warning, a cyclist is approaching") or from the car ahead ("I'm braking for an obstacle"). This "shared consciousness" is what will make roads mathematically safer.
2. Smart Infrastructure: When the Traffic Light Has a Brain
Even before cars drive themselves, the roads must become intelligent.
Adaptive Traffic Lights and "Green Waves"
Digital BlueFoam (digitalbluefoam.com) explains how AI uses predictive models for route optimization. Unlike old-generation smart traffic lights (which used simple induction sensors), new systems use Computer Vision. They recognize if a bus full of people (high priority) or a single car (low priority) is waiting and adapt the green light accordingly. This creates dynamic "Green Waves" for public transport and emergency vehicles, reducing public transport travel times and encouraging its use.
Digital Twins: The Italian Case (Milan and Bologna)
Italy is at the forefront of creating Digital Twins. According to AIDIA (aidia.it), cities like Bologna and Milan are creating virtual replicas of the entire city.
- What are they for? Before changing a traffic direction or building a bike lane, the administration simulates the impact on the Digital Twin. AI populates the virtual city with autonomous agents that simulate citizen behavior.
- Milano ForestaMi: AI also helps strategically place trees to maximize the reduction of heat islands and CO2 absorption, integrating mobility and sustainability.
This ability to simulate complex scenarios is based on advanced data analysis technologies. To understand the basics, read our guide on Predictive Analysis for Businesses.
3. Autonomous Vehicles (AVs): Safety and MaaS
Autonomous vehicles aren't just for reading the newspaper while driving. They serve to rethink the concept of car ownership.
From "Ownership" to "Service" (MaaS)
A study on ScienceDirect (sciencedirect.com) emphasizes how the greatest impact of AVs will be in MaaS (Mobility as a Service). Instead of having millions of cars parked 95% of the time (occupying precious public space), we will have fleets of constantly moving Robotaxis. A single shared autonomous vehicle can replace up to 10 private cars. This frees up space in cities for parks, wider sidewalks, and bike lanes.
XAI for Safety (Explainable AI)
But can we trust them? A paper in Nature Scientific Reports (nature.com) addresses the crucial topic of XAI (Explainable AI) in autonomous vehicles. When an AV decides to swerve sharply, it must be able to "explain" why it did so (e.g., "Detected child between two parked cars with 90% probability of crossing"). Without this explainability, it will be impossible to certify these vehicles for mixed urban use and, most importantly, assign liability in case of an accident.
The issue of liability is central. Who is at fault if the AI makes a mistake? We delve deeper in Who Judges the Algorithm? Ethics and Responsibility in AI Decisions.
4. Case Study: Who is Already Doing It?
It's not theory. Cities are already changing.
Hamburg and PTV Group: Predicting the Future
PTV Group (ptvgroup.com) describes the Hamburg project. Using the Machine Learning-based software PTV Optima, the city is able to predict traffic conditions with 60 minutes of advance notice. If the algorithm predicts a blockage downtown in an hour, variable message signs and connected navigators start diverting traffic now, preventing the traffic jam from forming in the first place. It's the difference between curing and preventing.
Metroville and EcoVille: Real Numbers
DigitalDefynd (digitaldefynd.com) analyzes the case of "Metroville" (pseudonym for an Asian metropolis), where the implementation of adaptive traffic lights reduced congestion by 30% and improved emergency service response times by 40%. In the case of "EcoVille," the integration between electric mobility and the Smart Grid reduced energy consumption for public charging by 20%, balancing the demand peaks of electric cars with the availability of solar energy.
The integration between electric cars and the power grid is a key theme of the transition. Read more about AI, Energy, and Sustainable Smart Grids.
5. The Technology Under the Hood: Mapping and Sensors
How does AI know where it is? GPS is not enough (it has an error of meters, the AV needs centimeters).
HD Mapping and SLAM
Autonomous vehicles use SLAM (Simultaneous Localization and Mapping) technologies. As they move, their Lidar sensors and cameras build a real-time 3D map of the environment and compare it with pre-existing HD maps. This allows the vehicle to understand not only where it is geographically, but semantically: "I am in the right lane, that is a solid line, that is a pedestrian looking at their phone (therefore distracted)."
Predictive Road Maintenance
AI isn't just for driving, but for maintenance. Connected cars, passing over a pothole, detect the anomaly with accelerometers and send the location to the road manager. This creates a real-time map of road degradation, allowing for surgical maintenance interventions before dangerous sinkholes form.
6. Challenges and Risks: Cybersecurity and Privacy
A connected city is a vulnerable city. If a hacker takes control of New York's traffic lights, they can paralyze the city or cause chain-reaction accidents.
The Cyber Risk
The security of V2X protocols is the absolute priority. Messages between cars and infrastructure must be encrypted and authenticated in milliseconds. AI plays a defensive role here, analyzing data traffic for anomalies that could indicate a coordinated attack.
The security of critical infrastructure is a matter of national survival. Learn more in Cybersecurity and AI: Low-Cost Hacking and Automatic Defense.
Privacy and Surveillance
To function, the Smart City must know where we are. License plate recognition cameras, GPS tracking, facial sensors. There is a thin line between efficiency and mass surveillance. In Europe, the GDPR mandates that data be anonymized at the source (Edge Computing), but the risk of re-identification remains. Administrations must guarantee transparency in the use of this data.
FAQ: Frequently Asked Questions about AI Mobility
1. When will we see fully autonomous cars in Italy? Level 3 cars (autonomous driving on highways/in traffic jams) are already legal in some contexts. Robotaxis (Level 4/5) without a steering wheel are being tested in cities like San Francisco and Phoenix. In Italy and Europe, urban complexity and regulatory prudence suggest widespread arrival around 2030-2035.
2. Will AI solve traffic without building new roads? Largely yes. Current roads are used inefficiently (gaps between cars, unnecessary red lights). By optimizing flow ("platooning" of cars, smart traffic lights), the capacity of existing roads can be increased by 30-50% without pouring a single extra gram of asphalt.
3. What is V2X? It stands for "Vehicle-to-Everything." It is the technology that allows the car to talk to:
- V2V (other cars): "I'm braking."
- V2I (Infrastructure): "The traffic light 200m ahead will turn red."
- V2P (Pedestrians): Pedestrians' smartphones to avoid collisions.
4. Do smart traffic lights respect privacy? Yes, if designed correctly (Privacy by Design). Cameras should not record faces or license plates for fine purposes, but only count "objects" (cars, buses, bikes) and turn them into anonymous numbers for the traffic management algorithm.
5. What happens if an autonomous car has to choose who to hit (Trolley Problem)? It's the classic ethical dilemma. Car manufacturers and legislators are working on ethical standards. In general, AI is programmed to always brake and minimize the kinetic energy of the impact, avoiding making qualitative choices ("save the young or the elderly?"), which remain an ethical and legal taboo.
Conclusions: The City as a Living Organism
AI in urban mobility is not just a technological upgrade; it is a biological paradigm shift. We are transforming the city from a collection of stones and dumb machines into a living organism, equipped with a nervous system (fiber optics and 5G), senses (IoT sensors and cameras), and a brain (Cloud and Edge AI).
The ultimate goal is not technology for its own sake, but Livability. A city where AI manages traffic is a city where the air is cleaner, where less time is lost in queues, and where children can cross the street more safely. The challenge for Italian urban planners and politicians is to embrace this complexity, investing not only in concrete, but in code and data.
To understand how AI is becoming pervasive in every aspect of our social and working lives, explore our MindTech section.
Bibliographic References and Sources
To ensure technical and visionary accuracy, this article drew from the following primary sources:
- Infrastructure and Smart City:
- Autonomous Vehicles and XAI:
- ScienceDirect – Smart Infrastructure and Autonomous Vehicles. La Bussola dell'IA · Articoli · Rubriche