Probabilistic Intent Inference for Predicting the Compliance with COLREGs between Two Surface Ships

Abstract

The international regulations for preventing collisions at sea (COLREGs) are the rules of the road for marine surface vessels. However, certain ships fail to comply with COLREGs and their non-compliance poses a greater danger to the maritime safety. This study proposes a probabilistic model for intent inference of ship maneuvers which consist of an intent model, a dynamics model, and a measurement model. An algorithm based on the proposed graphical model is implemented to infer and predict the ship's intent of non-compliance/compliance with COLREGs, which enables making proper decisions for collision avoidance maneuvers even when the opponent ship violates the marine traffic rules. In order to demonstrate the feasibility of the proposed algorithm, the results of extensive traffic simulations are presented and discussed. Copyright (C) 2020 The Authors.