By Prof. Dr. Evy Rombaut, MOBI Research Centre, Vrije Universiteit Brussel. Presentation at Flanders Technology & Innovation Festival.
In 2016, self-driving trucks made their Belgian debut on Flemish motorways as part of the European Truck Platooning Challenge. That same year, the first fatal accident involving an autonomous vehicle occurred in Florida, and Google's self-driving cars had already clocked up millions of kilometres. A decade on, De Lijn has launched its first autonomous bus in Leuven, and Waymo has taken its first steps onto European streets in Munich. So what has changed, and what have we actually learnt?
At the MOBI Research Centre's autonomous and electric vehicle lab at Vrije Universiteit Brussel, we have spent ten years investigating not just whether autonomous vehicles (AVs) can drive, but whether society is ready for them and how we can make that transition work for everyone. Our research has followed four core questions: who rides, and why? How do AVs fit into the Brussels context? What role does infrastructure play? And what happens to professional drivers?
Who Rides and Why?
Our first field experiment took place in June 2017, when participants took the wheel of a Level 2 autonomous Tesla, equipped with adaptive cruise control and lane keeping, along a 23-kilometre stretch of motorway. Before and after the ride, we measured their attitudes towards autonomous vehicles.
The headline finding was clear: experience matters enormously. Across all participants, attitudes towards AVs improved after the test ride. The effect was most striking among those with high levels of technology anxiety, precisely the group you might expect to be hardest to convince.
We took that lesson directly into our next project. In June 2019, MOBI ran its own pilot with an autonomous shuttle at UZ Brussel, while MIVB simultaneously deployed a shuttle in Woluwe Park. Surveying users across both sites, we found broadly positive experiences: general experience and driving behaviour scored well, though comfort received more mixed reviews, partly because the technology of the time occasionally caused the shuttle to brake for falsely perceived obstacles, such as falling leaves.
From 2020 to 2024, we scaled up significantly through the European SHOW project, tracking pilot deployments across the continent. We found that user ratings across nine European cities, from Frankfurt to Tampere to Klagenfurt, remained consistently high. Safety perception scored particularly well. The one recurring weak point was speed: at the time autonomous shuttles tended to travel slowly by design, which passengers noticed and sometimes found frustrating.
Lastly, we also explored the urban spatial dimension through a virtual reality study, inviting participants to experience a future Brussels in which shared AVs have reduced the need for parking, freeing up space for green infrastructure. The response was encouraging, suggesting that the broader urban vision, not just the vehicle itself, can strengthen public acceptance.
The key takeaways from a decade of user research are these: younger, tech-savvy users are quickest to adopt; real experience and pilot projects remain indispensable; safety perception is often the most important factor driving acceptance, and that perception is currently high. The open question is how to bring everyone on board, including those who are slowest to embrace new technology.
The Brussels Context: Can Shared AVs Replace the Private Car?
Brussels is a city moving steadily away from private car use. Low-emission zones, car-free areas, and active car-sharing schemes are reshaping urban mobility. Against that backdrop, we asked: could shared autonomous vehicles (SAVs), combined with the city's existing Park & Ride infrastructure, offer a genuine alternative to the private car?
To answer this, we developed a detailed Brussels simulation model using MATSim, fed by data from StatBel, BISA, Proximus, VIAS, UrbIS, Brussels Mobility, GTFS, and OpenStreetMap. The model captures real mobility flows at the micro level and projects them onto macro-level outcomes, allowing us to compare today's mobility patterns with a future SAV scenario in which private cars park at the city's edge and travellers continue their journey by public transport or shared autonomous shuttle.
The results are promising. Depending on pricing assumptions, CO₂ emissions dropped by between 32% and 63%, with a high replacement ratio of private vehicles by autonomous alternatives. However, pricing proved critical: if SAV services are offered free of charge, they risk generating additional traffic in the areas surrounding Brussels and cannibilizing public transport, an unintended consequence that planners would need to manage carefully.
Infrastructure: To Build or Not to Build?
One of the most debated questions in autonomous mobility is whether dedicated infrastructure is needed, or whether vehicles should simply adapt to existing road conditions. In the United States, the prevailing approach has been to deploy vehicles into existing infrastructure and expect them to cope. In Europe, the question has been approached differently, with the European Commission actively investigating whether targeted infrastructure investments could accelerate and improve the rollout of AVs.
Within the Augmented CCAM project, we examined specific edge cases where infrastructure-to-vehicle communication could make a meaningful difference: interactions with emergency vehicles, pedestrians crossing in low-visibility areas, roadworks zones, and motorway merging. These use cases were tested at sites in France, Spain, and Latvia.
The findings were compelling. Infrastructure support expanded area coverage from 32% to 91.3% and extended the average anticipation distance from 47.5 metres to 343.2 metres, giving drivers — including those still behind the wheel of conventional vehicles — an additional 10 to 15 seconds to respond in critical situations. Crucially, the benefits were not limited to fully autonomous vehicles: any connected vehicle gains from this kind of smart infrastructure.
Two broader lessons emerged. First, context determines impact: results varied considerably between pilot sites, reflecting the importance of local conditions and implementation details. Second, implementation costs for infrastructure solutions varied enormously depending on context, demonstrating that even modest investments can deliver significant safety gains.
What About the Drivers?
No serious discussion of autonomous mobility can avoid the question of employment. Within the EU RESKILLING project, we are currently modelling the labour market impact of autonomous shuttle deployment across the full value chain, from research and development through to operations, end-of-life, and recycling, and across all relevant stakeholders, including operators, governments, workers, passengers, and society at large.
Our benchmark model compares an autonomous shuttle service against a human-driven equivalent across four reference years: 2026, 2030, 2040, and 2050. Preliminary results show that in 2026, the net social benefit per vehicle kilometre is negative. At this early stage, a safety operator monitoring a single vehicle remotely replaces the driver's wage cost, while the automation kit still carries a significant price premium. Both costs decline over time as supervision ratios improve and technology matures.
From 2030 onwards, the picture shifts. Labour cost savings begin to outweigh technology costs, but the distribution of those gains matters greatly. Operators capture the largest share of the net benefit. Government revenues decline as employer contributions fall. Workers, on net, are worse off due to displacement costs, while passengers benefit from improved service availability. Compensating displaced workers for their losses would consume roughly 14% of operator gains in 2040, falling to 9% by 2050 — a manageable but politically significant figure. This is still work in process and more figures will follow soon.
Looking Ahead
We are now in 2026, and the landscape has shifted dramatically from where we started. The challenge is no longer whether autonomous vehicles can drive. It is whether we can integrate autonomous mobility successfully into society, making it safe, equitable, and economically sustainable for all.
Ten years of research point to four priorities. Experience builds trust: pilot projects are not optional extras but essential tools for public engagement. Shared autonomous mobility offers real opportunities in urban contexts, particularly when combined with existing public transport infrastructure. Targeted physical and digital infrastructure investments, even small ones, can make a significant difference. And scaling up requires an integrated approach that accounts for users, workers, operators, and the wider public interest alike.
The technology has arrived. Now comes the harder work.
Prof. Dr. Evy Rombaut is professor at our Autonomous and Electric Vehicles Research group at the MOBI Research Centre, Vrije Universiteit Brussel. For more information, contact evy.rombaut@vub.be