On the Quest for Criticality Searching for Interactions that Matter When Hunting for Bugs

Franz Wotawa

Abstract: Quality assurance of mobile and autonomous systems is essential for ensuring trust and avoiding harm, especially when such systems are safety-critical, heavily interacting with humans in our physical environment. Considering only requirements or models of systems for generating test cases or formal verification may not be sufficient due to the huge number of possible interactions that the requirements or system model might not cover. Therefore, this paper discusses a testing approach that utilizes environmental models for test case generation to mitigate this challenge. For representing the environmental models, we suggest using formal ontologies from which we can automatically generate test input models. The latter can be used directly for test case generation. Although the proposed approach was initially closely paired with combinatorial testing, it is agnostic regarding the underlying testing methodology. It can also be combined with search-based or random testing, as discussed in this paper. Besides introducing the foundations and presenting an algorithm that allows us to compile ontologies into test input models, we show an application scenario from mobile robots and summarize already reported results of an experimental evaluation in the context of automated driving assistance systems.