Towards the development of an operational fail-safe approach for autonomous tugboats

Abstract

Ensuring adequate operational safety for future marine systems with high levels of automation and low levels of human involvement will require the development of approaches that facilitate failing with safety, as highlighted by guidelines by Classification Societies. Designing fail-safe behaviour for autonomous marine systems is currently based on different types of hazard analyses that mostly depend on expert opinion about how to achieve this in different scenarios. The objective of this paper is to present a methodology for determining the decision paths that lead to suitable and feasible Minimum Risk Conditions (MRCs) outside the operational envelope given specific triggering events. The methodology integrates system modelling techniques with the Systems Theoretic Process Analysis (STPA) and is demonstrated in a case study involving two autonomous tugboats that manoeuvre a vessel to its berth while being monitored by a remote operator. The analysis resulted in three main and one last resort MRC that were associated to different scenarios based on the available system functionalities. Our aim is to contribute towards determining design requirements related to the behaviour of autonomous marine systems outside their operational envelope, as well as developing formal approaches for verifying a minimum level of safety in abnormal conditions.