Using Operational Design Domain in Hazard Identification for Automated Systems

Artificial Intelligence (AI) based components require new approaches to identify the risks they introduce in Automated Systems (ASs). This paper presents a methodology designed to support the identification of hazards in ASs. This methodology is built on the definition of scenarios derived from Operational Design Domain specification, a taxonomy of vehicle capabilities, and augmented keyword-based hazard analysis. An ontological model relates these elements to help discover specific hazards pertaining to AI integration in ASs. The method is implemented in a web-based framework and exemplified in a vehicle driving scenario. The demonstration video, keywords, and code are available at https://github.com/ollierGuillaume/AI-HARA.