Alexander Kittelmann

Context: Skeditor is a graphical editor for modeling, specifying, and verifying skill graphs. Skill graphs are a means for modeling cyber-physical systems in abstract and modular fashion (https://github.com/AlexanderKnueppel/Skeditor).

Problem: Although important properties in the context of cyber-physical systems, we did not consider non-functional requirements and uncertainty. In this direction, enabling redundant modeling of functionality and sensors based on uncertainty measurements would increase robustness of real-world application of skill graphs. For instance, we assumed ideal values from our sensors in our verification process, whereas sensor quality depends on numerous factors, such as employed hardware and weather conditions. Adding means for specifying such assumptions allows developers to add redundant sensors and perception units in case of degraded functionality.

Envisioned Solution: Extend Skeditor with means for specifying and verifying non-functional requirements. A starting point is a formalization of the theory of belief functions (Dempster–Shafer theory), which allows for reasoning with uncertainty.

Prerequisites: Basic knowledge of Java and Eclipse EMF (Skeditor is based on Graphiti).