At a Transport Research Arena (TRA) panel on smart mobility, participants argued that Europe is unlikely to win the global race to build foundational AI models — especially in generative AI — yet this may not be the most important battleground. Instead, the panelists said Europe can compete by adapting AI into safe, robust and genuinely usable transport systems.
AI in practice: traffic management, predictive maintenance and sensing
The discussion highlighted that artificial intelligence is already embedded in mobility systems, even if ordinary passengers do not always notice it. In road traffic and urban traffic control, AI supports route planning and the optimisation of traffic lights. A concrete example mentioned at the panel was Google’s Green Light project, tested in Budapest, which aimed to optimise traffic light switching with AI; the reported results showed a 30 percent reduction in vehicle emissions and a 10 percent reduction in the number of stops vehicles needed to make.
AI is also deeply integrated into environment perception: sensors, cameras and automated driving functions rely on algorithms to interpret a vehicle’s surroundings and assist decision-making. The technology can help predict unexpected road-user behaviour and support emergency braking systems. Moreover, AI increasingly contributes to engineering and development processes, accelerating manufacturers’ R&D — a strategic advantage in the fast-paced competition with countries such as China.
In rail transport, Francesco Flammini of the University of Florence identified three main AI use cases: predictive maintenance and failure protection, traffic control, and autonomous mobility. Predictive maintenance is particularly promising because operational and technical data from rail systems can help detect faults before they cause service disruptions. However, Flammini also pointed to a key constraint: rail systems often lack sufficient data for training and testing models, and teams currently spend considerable time sourcing usable datasets instead of developing new solutions.
Passenger-facing benefits: real-time multilingual disruption information
The panel also covered applications that passengers can directly perceive. Researcher Elodie Petrozziello mentioned a Paris trial of real-time, multilingual translation of disruption information. While less flashy than demonstrations of self-driving vehicles, improved and accessible passenger information can make public transport more usable for tourists, migrants and anyone who does not speak the local language.
Why aren’t more solutions deployed across Europe?
A principal answer from the panel was the gap between academic research and industrial deployment. Academic projects often operate at low technology readiness levels: exploring new ideas, testing prototypes and publishing scientific results. Industry requires mature, widely deployable systems. That divide is compounded by differing incentives: private firms ask how to monetise a solution, universities ask how to secure funding and pay staff, and the public sector balances sustainability, efficiency, safety and growth.
Somogyi Rita, head of the TERN transport company, emphasised that these differing priorities can hinder collaboration and that bringing stakeholders together is vital. The panel pointed to successful sectoral cooperation in aviation — for example through SESAR — as a model.
Visibility and communication: pilots must be shared to scale up
Good technologies do not spread automatically. Elodie Petrozziello cited the Hi-Drive project as a positive example because it actively communicated results, including via YouTube videos. She argued that researchers and project partners must make their work visible so industry and decision-makers learn about successful pilots and their practical applicability.
Moderated by Axel Volkery, head for innovation and urban mobility at the European Commission’s Directorate-General for Mobility and Transport, the discussion explored how the EU could do more to support AI adaptation in transport.
Europe’s realistic opportunity: be first in usable, safe transport AI
The panel concluded that Europe’s strategic position is not necessarily to lead the development of large foundational AI models, but to become the leader in safe, reliable and useful AI systems for transport. Europe’s long tradition in designing, regulating and operating complex transport systems — and some of the world’s most advanced public transport networks — gives it an advantage in integrating AI into service-critical infrastructure.
What may look like regulatory constraint can be a market opportunity: if European companies, researchers and authorities can develop transport AI that is demonstrably safe and robust, those solutions could be valuable well beyond Europe. The path forward emphasises adaptation, visibility and collaboration so that research outcomes evolve into practical, widely deployed systems.
