Explorative Project 4 (EP): Cloud-based software architecture for high-automated vehicles (DRAFT)

Future vehicle developments face major challenges – particularly due to increasing automation, connectivity, entertainment functions, shorter development cycles and over-the-air updates. A key approach to reducing complexity is the separation of software and hardware into independent layers, as implemented in the ‘Software-Defined Vehicle’ (SDV) concept.

The aim of the DRAFT project was to develop an innovative SDV vehicle architecture that combines approaches from the Internet of Things (IoT), modern software development and automotive research into an integrated solution for safe, highly automated driving. Encapsulation, abstraction, load balancing through orchestration, continuous integration and continuous deployment – as key technological drivers – were incorporated into a modular, cloud-based architectural definition.

The architecture developed comprises three areas: the ‘Event- and Time-Driven Stack’ manages the software in the vehicle, with applications running in containers according to their criticality and urgency. These containers are abstracted from the operating system and can be flexibly scaled to different computing resources, supported by an orchestrator and CI/CD for dynamic deployment. The ‘Cloud-based Stack’ performs similar tasks to the in-vehicle stack, but independently of the vehicle; for example, non-critical workloads can be offloaded. Finally, the ‘Supervisor System’ handles supplementary tasks for research operations.

Result: The project successfully demonstrated that standardised software components can be deployed independently of the hardware. The architecture was implemented on a testbed from the successfully completed SAFIR project EP 1 – AutoBit (FKZ: 13FH7I3IA) and enables fast, robust and secure data collection for research operations.