Third party funded individual grant
Start date : 15.06.2007
End date : 31.12.2011
This project is part of the INI.FAU collaboration between AUDI AG and the University of Erlangen-Nuremberg. It examines model-driven ways to integrate vehicle functions on electronic control units (ECUs). Moreover, the project develops supporting methods and tools for this task. The insights gained in the course of this project will be practically validated by integrating a damper control system into an AUTOSAR ECU. In the automotive industry it is common practice to develop in-car-software on a high level of abstraction and in a model-based way. To eliminate uncertainties concerning resource consumption and runtime it is necessary to test the developed software on the target hardware as early as possible. But due to cost and safety requirements the integration of the software into an ECU is very time-consuming and demands special expertise going beyond that of the function developer. AUTOSAR (AUTomotive Open Systen ARchitecture) is on the way to become a standard for the basic software on ECUs. But due to the novelty of this standard there are neither processes nor tools to support the integration of the developed in-car-software into an ECU. In 2008, we have examined the modeling expressiveness of AUTOSAR with respect to both its applicability and possible conflicts with existing standards and technologies that are currently in use at Audi. Furthermore, the automatic generation of an AUTOSAR software architecture from a single damper control component has been implemented. Since 2009, a model-driven approach that supports the integration of software into an ECU is being implemented and integrated into the tool set used at Audi. In particular we are looking at the automatic configuration of the bus communication by means of a bus database and the automatic task scheduling among the application processes. The model-driven development, which in this case is based on the Eclipse Modeling Framework, will enable easier tayloring of the emerging prototype to changing requirements. In 2010, we exploited the information that is available in an AUTOSAR project to automatically configure local and remote communication between software components. We have also developed a genetic algorithm that uses dependency information to automatically generate task schedulings that minimize communcation latencies between cooperating software components. The existing prototype has been extended with the above-mentioned methods.