Sampling-based Collision and Grounding Avoidance for Marine Crafts

Collisions and groundings account for a great deal of fiscal losses and human risks in the statistics of marine accidents related to ocean going vessels. With highly automated vessels offering a high degree of situational awareness, algorithms can anticipate developments and suggest timely actions to avoid or deconflict critical events, in accordance to safe navigational practices and in compliance with the Convention on the International Regulations for Preventing Collisions at Sea (COLREGs). To avoid such accidents related to navigation, this article proposes a Short Horizon Planner (SHP) for decision support or automated route deviations, as a means to mitigate prevailing risks. The planner adopts a
sampling-based planning framework that uses the concepts of cross-track error and speed loss during a steady turn, together with sampling spaces directly extracted from the electronic navigational chart to compute optimal and COLREGs compliant paths with the least deviation from the ship’s nominal route. COLREGs compliance (rules 8, 13-17) is achieved through an elliptical-like representations of the given COLREGs, which rejects samples based on modified ship domains. High fidelity simulations show properties of the method and the information made available to human- or automated execution of route alterations.