From Pilot Project to Real-World Operation
What’s Still Missing Today
For autonomous mobility, it’s not just a matter of whether a vehicle can drive. What matters is whether it can be integrated into a reliable, routine public transportation system.
It is precisely at this point that the debate often becomes vague. Technical feasibility is important, but it is not the actual stress test. The real measure is whether autonomous mobility can be operated in a way that is sustainable, safe, economical, and organizationally resilient. This leads to the next reality check: What is still missing today on the path from pilot project to public system?
The federal government identifies the gap with unusual clarity. Its strategy states that the market is in a “phase between completed testing and a lack of scaling.” At the same time, it notes a lack of market-ready solutions as well as scalable operator and business models for autonomous shuttle services.
International perspective: France describes the transition from testing to deployment in much more operational terms. The responsible ministry writes: “The deployment of automated vehicles and mobility services in France is made possible by the legislative and regulatory framework.” The difference is important: not just testing, but thinking in terms of application and service.
Source: https://www.ecologie.gouv.fr/en/public-policies/automated-connected-road-transport
That is precisely where the core problem lies. Many projects demonstrate that autonomous vehicles can function in defined environments. But routine operations raise different questions. Who takes responsibility in day-to-day operations? How are technical supervision, the control center, maintenance, intervention, and service organized? How reliable is the service under real-world conditions? And how does a funded project become a sustainable operation? The Handbook on Autonomous Driving in Public Transport makes it clear that autonomous mobility in public transport must be conceived as an operational task. Among other things, it addresses operational models, personnel, finances, technical supervision, control centers, safety, and organizational implementation as key planning areas.
What is often underestimated here is that the difference between pilot and regular operations is not merely organizational. It is also physical and system-level.
In pilot operations, scenarios can be limited. Routes are clearly defined. Interventions can be safeguarded. In regular operations, this very implicit safeguard is gradually eliminated. Then it is no longer sufficient for a vehicle to be able to operate under good, predefined conditions. It must also remain controllable under varying conditions.
This changes the central question: it is no longer just whether a system works, but whether it remains capable of functioning even when things are not operating as intended.
This is precisely where a concept comes into play that is absolutely central to autonomous systems in general and to public transportation in particular—and one that often comes up too late in many debates: fail-operational. For autonomous systems in public transit, it is not enough to simply switch to a safe shutdown in the event of a failure. A vehicle that simply stops in the event of a failure blocks operations, reduces availability, creates new risks, and makes scaling economically fragile. Regular operation therefore means more: a system must be able to safely reach defined states, even if parts of the system are operating at reduced capacity.
This is not a matter of convenience. It is an operational issue. Because in real-world public transportation, availability, reliability, predictability, and controlled behavior under everyday conditions are what count.
Many pilot projects currently circumvent this problem through limited scenarios, safety drivers, remote backup, or additional intervention levels. In regular operation, such workarounds are no longer viable in the same way. Then, vehicle control becomes a critical system element.
Bitkom is calling for larger pilot regions and more realistic operating conditions in order to answer precisely these questions. Reliable insights into scalability, operation, and cost-effectiveness do not emerge from small-scale, exceptional operations, but rather under real-world conditions.
In public transit, what matters is not whether the technology works in individual cases, but whether a service can be operated on a daily basis in a stable and predictable manner. Regular operation is therefore not a later stage of development. It is the actual goal that must be taken into account from the very beginning.
This is precisely why the debate must shift. The question is not: Can the vehicle drive autonomously? But rather: Will vehicle movement remain controllable, reliable, and safe during real-world operation?
Answering these questions lays the foundation for a public transportation system.
Anyone who seriously wants to advance autonomous mobility in public transportation must not treat regular operation as a long-term goal. It must be the benchmark for system architecture from the very beginning.
