Systems Theoretic Process Analysis for Layers of System Safety

Abstract

The move toward autonomous vehicles (AV) and mobility services extends the scope of traditional vehicle systems to include a much broader system definition. Questions often arise on how to identify a practical level of system hierarchy for the analysis. System Theoretic Process Analysis (STPA) is a hazard analysis technique that can be used to handle emergent properties of large, organized systems. However, STPA lacks the degree of formalization required to implement it successfully in a fast‐paced industrial environment. This paper introduces a framework to apply top‐down STPA on complex systems at multiple levels of decomposition and abstraction. In particular, we leverage the hierarchy within the system of interest (SOI) in tandem with STPA and proactively analyze complex dynamic processes that could potentially lead to accidents. This leads to multi‐level hierarchical progression of STPA on SOIs, in addition to analyzing each system element at different abstraction levels. The outcome of the analysis is a set of requirements for safety, cybersecurity, performance and business needs. The paper also provides guidelines on using these derived requirements to steer the system design. We demonstrate the efficacy of the proposed methodology by applying it to an AV ride‐sharing ecosystem.

The Challenges and Opportunities of Human-Centered AI for Trustworthy Robots and Autonomous Systems

“…Traditional hazard analysis methods include Fault Tree Analysis (FTA) [31], Hazard and Operability Analysis (HAZOP) [32], Failure Modes, Effects and Criticality Analysis (FMECA) [33] and Systems Theoretic Process Analysis (STPA) [34]. These methods can precisely identify the potential risks and model fault scenarios during the design phase.…”