Enabling autonomous navigation: adaptive multi-source risk quantification in maritime transportation

Abstract

Current studies on maritime navigation risks often overlook interactions between ships, dynamic surroundings, and static environmental factors, limiting insights into navigation safety in complex scenarios. This research presents an innovative methodology to quantify and integrate multi-source heterogeneous navigation risks, enabling a comprehensive assessment of overall risk levels. The framework comprises four components. First, a spatiotemporal risk monitoring domain model, developed using historical AIS data, incorporates risk monitoring and forbidden domains, enabling precise localisation and timing of risk evaluation. Second, heterogeneous navigation risk evaluation functions, addressing dynamic target and static environment risks, capture ships’ varying sensitivities to diverse risk sources. Third, risk quantification methods evaluate dynamic risks from temporal and spatial perspectives while categorising static risks into three types. Finally, an adaptive fusion method hierarchically aggregates multi-source risk data into a unified profile, reflecting navigators’ risk perception. Real-world AIS data validate the framework, constructing spatiotemporal risk models for three ship types and analysing navigation scenarios such as crossing, overtaking, and multi-ship encounters. Results demonstrate the framework's capability to enhance precision in navigation risk assessment, providing actionable insights and robust support for autonomous navigation and intelligent maritime systems. This methodology offers a promising tool for advancing safety in complex maritime environments.