Abstract
Increasing intensity in maritime traffic pushes the requirement in better preventionoriented incident management system. Observed regularities in data could help to predict vessel movement from previous vessels trajectory data and make further movement predictions under specific traffic and weather conditions. However, the task is burden by the fact that the vessels behave differently in different geographical sea regions, sea ports, and their trajectories depends on the vessel type as well. The model must learn spatio-temporal patterns representing vessel trajectories and should capture vessel’s position relation to both space and time. The authors of the paper proposes new unsupervised trajectory prediction with prediction regions at arbitrary probabilities using two methods: LSTM prediction region learning and wild bootstrapping. Results depict that both the autoencoder-based and wild bootstrapping region prediction algorithms can predict vessel trajectory and be applied for abnormal marine traffic detection by evaluating obtained prediction region in an unsupervised manner with desired prediction probability.
Highlights
The maritime logistic industry is a very crucial component of the global trade economy with an expanding volume, traffic intensity, and requirements
This paper investigates and proposes a method based on long short term memory (LSTM) autoencoder [18] to predict vessel trajectory and evaluate prediction region
Blue ellipse bounds prediction region calculated by wild bootstrap method
Summary
The maritime logistic industry is a very crucial component of the global trade economy with an expanding volume, traffic intensity, and requirements. The vessel traffic anomaly detection task can be defined as a task of outlier detection, where vessel traffic data are being analyzed as multiple standalone vessels positions/navigational vectors (point-based) or in trajectory-based manner, where vessel’s vectors are structured to time series sequences [19]. Automated marine traffic data gathering systems returns huge vessels trajectory/navigational data sets, which are challenging for human-based analysis and traffic anomaly detection [28]. Marine traffic is a dynamic system, where vessel’s traffic properties change in space and time. Such type of data can be defined as spatio-temporal time series. Despite advances in prediction of spatio-temporal data with deep neural network, the authors do not propose prediction/confidence interval evaluation that is crucial for marine traffic anomaly detection with this method. Published works take advantages of extended LSTM (Long Short Term Memory) neural networks to learn spatio-temporal dependencies (see [12, 15, 20])
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
