Computational Functional Hazard Assessment: A New Paradigm for Autonomous Aviation Systems

Functional hazard assessment (FHA) is a cornerstone of safety assurance for civil airborne systems. However, as the National Airspace System’s ecosystem evolves and novel operations with varying levels of autonomy, complexity, and acceptable risk are introduced, challenges in performing FHA surface. We argue that there are opportunities to support safety assurance by extending FHA. Specifically, opportunities include using computation for hazard assessment, proposing a standardized model capture and representation, and incorporating historical data to augment FHA. We further present the computational functional hazard assessment (CFHA) paradigm as one potential means of realizing the identified opportunities, using an example of a specific CFHA implementation applied to a drone performing a post-hurricane imagery mission. The core enabling features of CFHA are knowledge capture from historical data, analysis traceability via a formalized functional architecture specification, and iterative, simulation-based analysis of hazardous scenarios. The proposed CFHA paradigm can improve the ability to perform FHA for novel aviation systems with increasing levels of autonomy by a) identifying a broader range of hazards and failure conditions, b) better managing the large and complex hazard state space, c) better capturing the system’s internal and external interactions, and d) analyzing failure propagation and system resilience properties.