AdvoCATE: The Assurance Case Automation Toolset at Age 14

Assurance cases (ACs) are finding increasing adoption in many safety- and security-critical domains as a paradigm focused on communicating to stakeholders why a system or service can be relied upon, through an explicit statement of assurance objectives and rationale that shows how those objectives are met. ACs give the flexibility to both keep pace with technological change, e.g., with the rapid introduction of new technologies such as machine learning (ML), and to tailor verification efforts to be proportional to the levels of risk posed. Since the Assurance Case Automation Toolset (AdvoCATE) was first introduced to the community in 2012, it has inspired the development of various similar toolsets and capabilities for creating ACs. AdvoCATE has continued to evolve from its initial focus on creating structured arguments to capture assurance rationale, using the Goal Structuring Notation (GSN), to encompass a broader suite of features for safety risk management, including hazard analysis, risk modeling using barrier models captured using Bow Tie Diagrams (BTDs), capturing requirements, and organization of substantiating evidence as evidence logs. We use the assurance of an autonomous visual landing capability with a machine learnt component as a running example to showcase the utility and applicability of AdvoCATE.