Intent Inference of Ship Collision Avoidance Behavior Under Maritime Traffic Rules

This paper proposes an algorithm to infer the maneuver intention of an obstacle ship and to check its compliance with the maritime traffic rules to avoid ship collision and ensure maritime traffic safety. A probabilistic graphical model is constructed to represent the relationship between motion observations of the obstacle ship and its hidden maneuver intention to comply with the traffic rules. The probabilistic belief of the ship's intention is modeled and quantified using probabilistic tools such as dynamic Bayesian networks. Three different intent inference models are formulated considering the different levels of observation configurations, and their calculation procedures are described. To demonstrate the feasibility of the proposed intent inference algorithm, Monte-Carlo simulations were conducted and the results are presented.