장기 AIS 데이터 기반 연안 선박의 원격 상황인식을 위한 효율적인 위치 확률분포 추정 기법

Abstract

Automatic identification system (AIS) data gathered at vessel traffic service (VTS) centers include various and useful information with respect to each vessel navigated in the control area of VTS. In particular, navigation patterns reflected in long-term AIS data of coastal vessels can be utilized to enhance the performance of situational awareness for autonomous ships. This study addresses an efficient method for estimating positional probability distributions of coastal vessels using a large amount of AIS data collected over a long period. For this, the use of a weighted kernel density estimation algorithm is proposed considering respective densities of data distribution and their different confidences. The validity of the proposed method is demonstrated using a real AIS dataset obtained at a VTS center