Modeling and cross-domain dependability analysis of cyber-physical systems

This paper discusses a novel method of modeling and formal verification to support dependability analyses. The method is demonstrated in an example of a fault management capability of robots that interacts with equipment and humans. Hazard analyses produce derived requirements for fault management capabilities. These include safety critical functions for collision avoidance and temporary autonomy. Derived requirements are represented formally in models that are used to produce dependability evidence using theorem proving, model-based test vector generation, test execution with code coverage analysis, and requirement-to-test traceability. To address the challenges of heterogeneity of modeling tools and languages, Semantic Web Technologies are used for model composition and model transformation from modeling tools to formal analysis tools.