Evolution of Formal Model-Based Assurance Cases for Autonomous Robots

An assurance case should carry sufficient evidence for a compelling argument that a system fulfils its guarantees under specific environmental assumptions. Assurance cases are often subject of maintenance, evolution, and reuse. In this paper, we demonstrate how evidence of an assurance case can be formalised, and how an assurance case can be refined using this formalisation to increase argument confidence and to react to changing operational needs. Moreover, we propose two argument patterns for construction and extension and we implement these patterns using the generic proof assistant Isabelle. We illustrate our approach for an autonomous mobile ground robot. Finally, we relate our approach to international standards (e.g. DO-178C, ISO 26262) recommending the delivery and maintenance of assurance cases.