Risk-Based Control of Autonomous Surface Ship

Abstract

The last few decades have seen tremendous technological development in the maritime industry. Improvements in navigation systems such as Global Navigation Satellite Systems have reduced measurement error and uncertainty for both position and velocity data. In addition, improvements in computational power have contributed to the development of more advanced ship control systems. Increased focus on greenhouse gas emissions have further driven the maritime industry into developing new and more efficient ship transport systems. More and higher levels of autonomy at sea is one way to address this challenge and utilize the technological improvements.

A crucial part of developing autonomous ships is ensuring their ability to operate safely. To address this, the overall objective of this thesis has been to develop methods and tools for assessing and controlling risk when operating autonomous ships. To achieve this, risk related to autonomous ships needs to be analyzed and modeled. Since autonomous ships are complex systems that include hardware components, software, interact with humans, and operate in highly unstructured environments, traditional risk analysis methods focusing on single component failures are not considered sufficient. Instead, newer methods such as Systems Theoretic Process Analysis focusing on system interactions and Bayesian Belief Networks for modeling the system, combined with additional methods for analyzing specific parts of the ship are used to get sufficient information.

The main result of the thesis is the Supervisory Risk Controller that can combine information from a risk model with operational measurements from a ship control system to handle a wider range of challenges compared to existing ship control systems. Existing ship control systems are great at optimizing efficiency and costs. However, they often lack the ability to assess safety. By introducing a risk model in the control system, both safety and efficiency can be evaluated as part of the decision-making process similar to the way human operators and crew do on traditional ships. This is an important contribution towards operating autonomous ships by improving the decision-making capabilities accounting for both safety and efficiency.