State-of-the-Art on Artificial Intelligence for Risk-based Decision-making in Autonomous Marine Vehicles

Abstract

Risk assessment (i.e., risk analysis and risk evaluation) is routinely performed in marine operations by vessel crews to inform and support the decision-making process for the ship’s operations. With the inevitable ubiquity of autonomous marine vehicles, the need for effective autonomous decision-making methods which considers risk factors increases, to enable these vessels to perform at least as safely and reliably as manned ones. This paper presents an initial non-exhaustive systematic review of the literature on risk-based decision-making methods for Autonomous Marine Vehicles. Specifically, we are interested in the intersections between risk science, artificial intelligence, and autonomous marine vehicles, and in defining the state-of-the-art in this specific scope. We look at the last five years of the literature using the snowballing systematic review technique, selecting 23 papers after one iteration. After data extraction on the 23 papers, we compile the most common methods used in implementing risk-based decision-making in autonomous marine vehicles, and discuss the state-of-the-art.