Research Project “ATLAS-L4”
How Driverless Hub-to-Hub Transport Becomes a Reality
Autonomous lorries may soon be a common sight on Germany’s expressways.
MAN Truck & Bus
A joint project between MAN and eleven partners delivers a blueprint for how autonomous hub-to-hub transport will be implemented in Germany — and why SDV architects are now setting the pace.
With the ATLAS-L4 project, a twelve-member consortium has
demonstrated that highly automated lorries on German expressways are no longer
a vision of the future. The real enabler behind the first driverless hub-to-hub
truck is not a new generation of sensors, but a fully software-defined vehicle
architecture. This architecture allows functions to be delivered iteratively
via over-the-air updates, enables critical updates within days rather than over
product cycles, and integrates the vehicle into a DevOps-like pipeline — a
paradigm shift for commercial vehicle OEMs previously focused on hardware.
From Rolling Chassis to Software-Defined Trucks
At the heart of the demonstrator is a three-layer software
stack:
- Perception & Localisation aggregates data from lidar, radar, and cameras into a unified environmental model.
- Planning & Control generates trajectories based on a rule-based decision layer implemented as a microservice.
- Vehicle Interface Layer communicates via a service-oriented onboard network bus with redundant actuators for braking, steering, and power supply.
All layers operate on a containerised computing platform
with deterministic real-time hypervisor partitioning. The clear separation of
safety and non-safety domains simplifies ISO 26262 ASIL D certification, while
maintaining open OTA update corridors for non-safety-critical services.
What is ATLAS-L4?
ATLAS-L4 (short for “Automated Transport between Logistics
Hubs on Motorways at Level 4”) marks the first demonstration of a fully
autonomous lorry operating without a driver in hub-to-hub traffic on German
expressways. The project, funded by the German Federal Ministry for Economic
Affairs and Climate Action (BMWK), aimed to develop a production-ready Level 4
system architecture — including redundant actuators, a control centre for
technical supervision, and a scenario-based validation concept — thereby putting
Germany’s autonomous driving legislation into practice.
The twelve-member consortium includes: MAN Truck & Bus,
Knorr-Bremse, Leoni, Robert Bosch Automotive Steering, Fernride, BTC Embedded
Systems, Fraunhofer AISEC, Technical University of Munich, Technical University
of Braunschweig, TÜV Süd, the Federal Motorway Company, and the Würzburg
Institute for Traffic Sciences (WIVW).
Safety by Design
True Level 4 capability requires more than traditional
Failure Mode and Effects Analysis (FMEA). ATLAS-L4 combines fail-operational
hardware (steer-by-brake, dual 48-volt systems) with a software-based
self-awareness layer that continuously performs plausibility checks. If the
system’s confidence score falls below a defined threshold, it autonomously
triggers a minimal-risk manoeuvre — without any human intervention.
For approval, the team relied on a scenario-based validation
concept: 20,000 virtual edge cases from an open scenario database were
simulated in SIL/HIL environments before 150 real-world test drives verified
the digital twin. ATLAS-L4 thus bridges the gap between regulatory clearance
and production-ready approval.
Teleoperation:
Keeping the Human in the Loop
Although the lorry drives autonomously, German legislation
mandates a “technical supervisor.” The project’s control centre streams vehicle
and sensor data with sub-100 ms latency. A teleoperator can intervene if the
vehicle encounters an unclassifiable situation. The software architecture
includes a secure handover protocol with authentication, end-to-end encryption,
and a robust degradation strategy.
Cybersecurity as a Fundamental Requirement
As the functional
scope expands, so does the attack surface. The Fraunhofer Institute for Applied
and Integrated Security (AISEC) was the first to apply holistic security
management to an automated lorry: system-level threat modelling, intrusion
detection hooks in the Ethernet backbone, and a policy-based update manager
ensure cybersecurity throughout the entire lifecycle. This approach
treats the vehicle from the outset as a “cyber-physical system” — essential for
any SDV approach.
Roadmap for Logistics 4.0
The proof-of-concept is just the beginning. Building on the
prototype software base, the consortium plans to transition towards a scalable
platform. Commercialisable functions such as fleet balancing or predictive
maintenance are to be monetised via “feature toggles” — i.e., functions on
demand.
To support data-driven development, the petabytes of driving
data collected during testing feed into continuous reinforcement learning to
further improve planning algorithms. This is complemented by standardised open
APIs enabling seamless integration with transport management and ERP systems —
positioning the lorry as an intelligent node in the digital supply chain.
In the long term, the project lays the groundwork for a
logistics model in which driverless lorries help tackle driver shortages,
congestion, and CO₂ emissions. For the Software-Defined Vehicle community,
ATLAS-L4 not only provides a technological blueprint, but also regulatory proof
that software-defined commercial vehicles are technically feasible and legally
admissible in Germany.
This article was first published at automotiveit.eu
