Supervisory risk control of autonomous surface ships

Abstract

The objective of this paper is to develop online risk models that can be updated as conditions change, using risk as one metric to control an autonomous ship in operation. This paper extends and integrates the System Theoretic Process Analysis (STPA) and Bayesian Belief Networks (BBN) with control systems for autonomous ships to enable supervisory risk control. The risk metric is used in a Supervisory Risk Controller (SRC) that considers both risk and operational costs when making decisions. This enables the control system to make better and more informed decisions than existing ship control systems. The novel control system is tested in a case study where the SRC can change: (i) which machinery system is active; (ii) which control mode to run the ship in; and (iii) which speed reference to follow. The SRC is able to choose the optimum machinery, control mode, and speed reference to maintain safe control of the ship over a route in changing conditions.