Formal approaches to design and verification of safe control systems for autonomous vessels

Abstract

Maritime autonomy can positively impact society by cutting costs and emissions while enabling new solutions for transportation and mobility. Autonomous vessels must be capable of performing complex tasks under significant uncertainty in an unstructured environment, which requires fundamental innovations in the control systems. Introducing fundamentally new technology, combined with increased system complexity and criticality, introduces new risks that must be identified and mitigated to ensure safety. The success of maritime autonomy is ultimately hinged upon whether autonomy developers, regulators, and classification societies can find tractable solutions for safety assurance of autonomous vessel control systems. Formal Methods (FMs) are a family of mathematically based methods for design and verification that have been used actively for assurance of safety-critical systems in several other industries. Recently, there has also been active research on FMs applied to autonomous systems. The maritime industry has, however, no tradition of using FMs. This thesis has investigated how FMs can contribute to solving the safety assurance challenges for autonomous vessel control systems. The research has involved simulation studies as well as full-scale and model-scale experiments. The research has been conducted in close collaboration with industry partners.

The thesis has proposed several novel methodologies using FMs for safety assurance of autonomous vessels. Some key contributions include a formal methodology to automate large-scale simulation-based testing, a methodology for using formal contracts to enable modular design and verification, and a novel algorithm to estimate vessel motions with a formal proof of correctness. In addition to developing new methodology, the doctoral work has reviewed the literature on FMs and identified several existing tools and methodologies which have the potential to address the safety assurance challenges for autonomous vessels.