The SOTIF Meta-Algorithm: Quantitative Analyses of the Safety of Autonomous Behaviors

This paper introduces a meta-algorithm for achieving Safety Of The Intended Functionality (SOTIF) in complex systems by integrating quantitative Fault Tree Analysis (qFTA), structured generation of UL 4600 Safety Performance Indicators (SPIs), and statistical hypothesis testing of SPIs in simulation and reality. The meta-algorithm can be used for specific autonomy stacks and Operational Design Domains (ODDs) to make practical the risk analyses in the automotive standard ISO 21448. By using the SOTIF Meta Algorithm, operators will make more informed deployment decisions for autonomous systems.