Autofahrt in einer Stadt bei Nacht

Bild: lassedesignen / Adobe Stock

Auf dem Weg zu vertrauenswürdigen hoch-automatisierten Systemen

Ob Verkehr, Medizin oder Energie – in sensiblen Bereichen hängt der Erfolg von hochautomatisierten Systemen entscheidend davon ab, ob wir Menschen diesen Systemen vertrauen können. Dr. Thorsten Weyer beleuchtet in einem Vortrag, was dies für das Requirements Engineering bedeutet.

Dr. Thorsten Weyer hält seinen eingeladenen Vortrag am 01.09.2020 auf der IEEE International Requirements Engineering Conference im Rahmen des Industry Days.

On the way to trustworthy highly-automated CPS — Some thoughts on safety and security requirements in the age of autonomous systems and machine learning

Abstract — In the traditional view coined by Edward A. Lee, a cyber-physical system (CPS) consists of embedded systems which are interconnected by a technical network. Nowadays, CPS are no longer restricted to fixed (invariant) connections with other systems throughout their life cycle. Today, and even more in the future, such systems are able to dynamically form cooperating or collaborating groups in order to fulfil their mission in the best possible way. CPS with these capabilities are the technological foundation for many of our future scenarios, like in autonomous transportation, Industry 4.0, smart medical devices, and smart energy networks. In these scenarios, CPS typically need to operate autonomously to a great extent but not necessarily in all operational situations — viz., they are “highly automated”. Whether highly-automated CPS actually meet with broad acceptance will largely depend on whether we as humans can build trust in such systems. A core aspect for building trustworthy highly-automated systems is to guarantee safety and security in operation, even in the case of unforeseeable situations in the operational environment.

Using examples from the autonomous driving field, this talk examines overall challenges in the requirements engineering field to lay the foundation for safe and secure highly-automated CPS in dynamic and collaborative environments. From the perspective of requirements engineering, the talk also demonstrates how techniques from the areas of self-adaptive systems, collaborative systems and machine learning can be used to engineer trustworthy highly-automated CPS.