Rigorous validation and verification for dependable and safe software systems
The unique challenge here is in the development and operation of cognitive CPS. These systems are capable of autonomous response and decision-making and can also learn from data and experiences by means of artificial intelligence (AI) methods. These cognitive systems are thus able to adapt their behavior to changing environments. The underlying prerequisite for the successful deployment of autonomous CPS-based products is thus the availability of effective and affordable validation methods.
In our research activities, we rely primarily on mathematical-logical methods – linear programming, logical fulfillment and model testing as an example - for solving derived problems. We also develop methods designed for scenario-based testing and test approaches based on coverage-driven fuzzy testing, which is used to analyze the extent to which software is susceptible to errors in order to potentially pinpoint safety vulnerabilities. Our research activities contribute to the development of novel methods for validating and certifying autonomous cyber-physical systems, thus enabling the deployment of AI technologies in mission-critical, autonomous software systems and services as well.