Development of Contextual Collision Risk Framework for Operational Envelope of Autonomous Navigation System

This paper introduces the ‘Contextual Collision Risk (CCR)’ method, a novel collision risk assessment approach for autonomous navigation systems. As global shipping traffic increases, marine traffic becomes more complex and busier, requiring advanced risk assessment methods for autonomous vessels. The CCR establishes the Operational Envelope (OE) of autonomous navigation systems by integrating the complexities of real-time marine traffic and the specific maneuvering capabilities of own ship. The central aspect of CCR involves setting up Reachable Velocities (RV), which includes the dynamics of own ship. Additionally, the Velocity Obstacle (VO) algorithm is implemented to identify potential collision risks from other vessels. By integrating RV and VO analyses, CCR provides a framework that effectively quantifies collision risk in congested maritime environments. To demonstrate the effects of CCR, we employ data from transoceanic voyages across various scenarios. Furthermore, simulations of collision avoidance maneuvers are conducted, focusing on high-risk situations such as near-misses and potential collisions. In particular, the analysis of transoceanic voyage data shows that CCR holds promise for enhancing navigation safety in complex maritime environments.