Development of Navigation Support System to Predict New Course of Ship

Abstract

To keep safety navigation in congested sea area, it is required early action to avoid collisions based on precision situation awareness include prediction of other ship behavior. However, prediction method of other ship's next behavior automatically has not presented yet. On the other hand, there is Advanced Navigation System in Tokyo University of Marine Science and Technology and navigational big data for the past about 5 years in Tokyo bay were collected in this system. Therefore, this paper proposed prediction method of ship's next course at the exit of traffic route in Tokyo bay. The method was made use of traditional machine learning technique called Support Vector Machine and AIS data in Tokyo bay. As the result, in the exit of traffic route, prediction accuracy of ship's next course was 87.9%. However, to use the method practically, it was necessary to increase the predictive accuracy.