Autonomous driving is one of the leading trends in the automotive industry. Therefore, Electronic Control Units (ECUs) must execute ever more and more complex driver assistance systems. This increasing degree of automatization leads at the same time to a growing demand for high availability of critical functionalities since the human driver is no more available as a backup layer in case of a failure. In addition, new tasks for example from entertainment find their way into cars and must be executed. As those tasks have a different criticality in terms of safety, the whole system becomes a mixed criticality system.
This project investigates flexible and resource efficient fail operational concepts for component-based software architectures in heterogeneous hardware environments. In case of a fault in an executing entity, software components including critical functionalities should be executed on another entity, even though this entity might be based on a completely different hardware or software architecture. This topic will be investigated for the AUTOSAR Classic and Adaptive software architectures and ECUs based on either microcontrollers or microprocessors.
The main topics under investigation for this project are therefore:
- failure handling strategies, including fault detection, maintenance of critical functionalities, real-time readiness of the fault reaction and data consistency of the primary and backed up functionality,
- virtualization concepts for temporal and spatial separation of a mixed criticality system and
- the interchangeability of Software between different hardware architectures like microprocessors and microcontrollers and different software architectures like AUTOSAR Adaptive and AUTOSAR Classic.
This work is funded by: Vitesco Technologies Group AG