Extraction of Ship Collision Avoidance Patterns From AIS Data

The maritime industry faces constant challenges in preventing ship collisions and ensuring adherence to the Convention on the International Regulations for Preventing Collisions at Sea (COLREGs). This paper presents a comprehensive analysis of ship collision avoidance patterns utilizing Automatic Identification System (AIS) data. By examining AIS data parameters, including vessel speed, course, heading and the dynamic change in position, crucial insights into the avoidance of collision can be extracted. In this paper the encounters are extracted from the AIS data collected near Los Angeles Port. The raw data is preprocessed and visualised in a relative distance domain from the own ship. The encounters are filtered using a collision risk model and the encounter situations are labeled and classified into multiple classes using GRU-RNN deep learning technique.