A systems-based application for autonomous vessels safety: Hazard identification as a function of increasing autonomy levels

During the last decade, a series of autonomous vessels projects have been conducted strengthening the vision of the positive impacts on human safety, cost savings and environmental protection using advanced Artificial Intelligence (AI) based on data fused algorithms from various sensors. Although these projects have demonstrated the technological feasibility of this idea, the hypothesis that autonomous ships will be safer needs to be tested and safety constraints must be established. The main objective of this paper is to determine the hazards for the system as a function of the vessel's autonomy level. This is accomplished by implementing the System Theoretic Process Analysis (STPA) on specific levels of increasing autonomy, as described by the International Maritime Organization (IMO), where the human element is limited to remote monitoring and the on-board control system makes most of the decisions and takes most of the actions. STPA assumes that accidents are caused by unsafe interactions among system components, none of which may have failed. STPA presents an advantage over other hazard analysis tools, by being more flexible in terms of required information and by having a top-down functional model approach. The STPA's results are used to determine how hazard occurrence changes, considering the interactions among system components and hard to identify failure points. These results will be used to evaluate the hypothesis that autonomous shipping will be safer and potentially contribute to the ongoing discussion for the regulatory framework that must be implemented for autonomous shipping to become a large-scale reality.