Research On Collision Avoidance Strategy For Multi-ship Encounter Situation Based On Target Ship Intention Prediction

According to a large number of shipping data surveys,the ship collision accident is one of the key factors affecting the safety of shipping industry.The main reason causing ship collision accidents is human factors,such as misdetection of target ships,misunderstanding of human-driven intention of target ships,failing to comply with the marine rules and so on.In addition,ship collision accidents usually happen in the congested water area with high density of ships,where multi-ship encounter situations are frequently experienced.With the development of shipboard electronic equipment and artificial intelligence technology,research in autonomous collision avoidance system for intelligent ships has attracted the extensive attention.In this paper,a complete autonomous collision avoidance system based on the target ship intention prediction is established for the multi-ship encounter situation.For reactive collision avoidance planning,the real-time precise prediction of target ships motion is very important for effective planning.A prediction method for human-driven intention based on Hidden Markov Model is proposed.The prediction results of target ships intention are combined with Velocity Obstacle to form the collision risk evaluation.Motion primitives are employed to generate the collision avoidance trajectory satisfying ship maneuverability constraints.Successful simulation tests and on-water tests of autonomous collision avoidance with the trimaran model demonstrate the validity and reliability of the proposed algorithm,which is significant for the further development of intelligent ships.