ASSUME: Assumption Management Framework for Automotive Embedded Software Development
|Coordinator||Volvo Personvagnar Aktiebolag - Volvo Personvagnar AB|
|Funding from Vinnova||SEK 5 200 000|
|Project duration||August 2013 - July 2017|
|Venture||Electronics, software and communication - FFI|
Purpose and goal
The ASSUME project had several goals, of which the main was to mitigate assumptions in distributed software-mechantronics development, initially by means of an assumption management framework. However, the results shows clearly that an assumption framework cannot solve this issue, and that continuous integration (fast test loops) has to be used instead to find and fix faulty assumptions. One framework has been tested, but with expected and negative results. Hence, the suggested project result in the application was falsified, but we have gain much understanding in the field.
Expected results and effects
The project has strengthened the understanding, and conviction, for the need for massive and automated testing in (distributed) software development. In a product complex as a modern premium car, it is not possible to find and document the assumptions made during the analysis phase. Instead, these must be detected by testing, continuously and automatically. The project has also studied methods of virtual testing (model based) of mechatronic control software and organizational gaps between system design and software development. Both have strengthened R & D´s strategy work.
Planned approach and implementation
The project has been conducted as a series of case studies and a survey, where the first study explored what types of assumptions are made and what they have for effect, then investigated how these can best be identified and relieved and finally what happens in organizations that develop automotive, or complex mechatronic systems when that type of change is introduced, considering both architectural changes and local functional changes. One of these case studies has also been to test a framework for managing assumptions, together with a research team from the Netherlands