Risk Analysis of Autonomous Ships

Abstract

The main purpose of this paper is to address the feasibility of using the System Theoretic Process Analysis (STPA) for risk analysis and quantitative risk modeling of autonomous ships. The paper defines a set of requirements and questions, which are used to assess 35 potential risk analysis methods related to software, security, humans in the loop, and emerging risks. The assessment identifies seven main methods that can be used to expand the STPA analysis to enhance it for analyzing risk of autonomous ships, i.e., one method for software, security, and emerging risks, two for human factors, and two more general methods based on Bayesian Belief Networks (BBNs). The results show that almost all the applicable methods provide additional information compared to STPA only, but the results vary much from method to method. It is also a challenge to find a way of combining the results in a quantitative risk model that can be used to describe and evaluate the risk level for autonomous ships. The seven most relevant methods can be used as a starting point for further development of a general framework for risk analysis for autonomous ships.