Steering Machine Learning Mechanism Based on Big Data Integrated Cooperative Collision Avoidance for MASS

Abstract

The current collision avoidance implementation is based on unilateral static information, but not the bilateral movement information of the two vessels. In this paper, we utilized the Machine Learning (ML) Mechanism Integrated Vessel Networks (MLMIVN) to collision avoidance cooperatively for the two vessels especially for the (Maritime Automatic Surface Ships, MASS). The device onboard has the capability of big data analysis and edge computing and Vessel Networks is based on Device-to-Device(D2D) communication. The safety and economy of collision avoidance route can be improved by training historical navigation data. First, we put forward the concept of cooperative collision avoidance that considering the motion state of each vessel, and a two-step-turn cooperative collision avoidance method is utilized. Then a improved genetic algorithm combined with K-Means algorithm is used to train the big data.