Real-world proof becomes mandatory
UN sets first global rules for autonomous vehicle approval
Driverless vehicles such as this Waymo robotaxi will be assessed against globally harmonised technical requirements under the new UN framework.
Waymo
The UN has adopted the first globally aligned technical framework for highly automated and driverless vehicles. OEMs must now prove safety through audits, simulation, track and road testing, while controlling updates and monitoring performance throughout operation.
The UNECE World Forum for Harmonization of Vehicle
Regulations has adopted the first globally aligned technical framework for
approving highly automated and driverless vehicles. The rules define the
evidence manufacturers must provide before an Automated
Driving System can be authorised for use on public roads.
The scope covers systems that perform the complete dynamic
driving task within a defined operational design domain. This includes
perception, decision-making and vehicle control, and is relevant to many Level
3 and Level 4 applications, including robotaxis,
automated shuttles and a limited number of
passenger-car functions.
What must manufacturers define before approval?
Approval begins with the operational design domain, or ODD.
Manufacturers must specify where and under which conditions the system can
operate, including road type, speed range, weather, geographical area and
traffic environment. Outside those limits, the automated function must not be
available.
Within its ODD, the system must achieve at least the safety
level of a competent and careful human driver. The new rules therefore turn the
globally aligned regulatory framework for autonomous
driving into a structured safety case: manufacturers must identify
hazards, assess risks and demonstrate how residual risks are controlled.
How will automated driving safety be tested?
A single successful road test will not be enough.
Authorities are expected to assess a combination of documentation audits,
simulation, test-track trials and driving in real traffic. The development
process itself is also part of the evaluation, alongside cybersecurity and the
manufacturer’s software-update strategy.
The key requirement is that safety
evidence must connect simulation with real-world performance.
Virtual tests must be representative, traceable and transferable to actual
traffic conditions rather than serving as isolated proof points.
Human interaction remains part of the approval process.
Where a driver may need to resume control, the vehicle must issue a clear and
timely takeover request and assess whether the person is ready to respond.
Driverless vehicles without conventional controls must allow passengers to
request a safe stop.
What happens after type approval?
Manufacturer responsibility continues after a vehicle enters
service. Safety-relevant events, critical situations,
system failures and software problems must be recorded and reported to the
authorities. Regular reports on operating time, mileage and incidents
are also envisaged.
This creates a continuous evidence chain from development
into operation. It also reflects a broader regulatory trend in which crash data
can expose weaknesses that controlled tests miss. Software updates must remain
manageable and must not undermine the approved safety concept.
Why does global harmonisation matter?
The framework is intended to prevent autonomous vehicles
from facing completely different technical approval criteria in every market.
National and regional implementation will still be required, but OEMs now have
an internationally coordinated catalogue of expectations covering safety,
validation, cybersecurity, updates and in-service monitoring.
For manufacturers, this provides greater clarity about
what must be demonstrated before driverless technology
can move beyond pilot projects. It also shifts the competitive focus:
technical capability alone is no longer sufficient; companies must be able to
prove, document and monitor safe behaviour throughout the vehicle lifecycle.
