Potential Fields in Maritime Traffic
Modeling and Anomaly Detection
The issues of global marine traffic may be not as commonly apparent as these of road and air transport. Nevertheless, the water transport has a great, albeit indirect, impact on our lives. The coast guard, navigation officers, marine rescue services and many other institutions concentrate on assuring safe itineraries for all travelling vessels. The observation of marine traffic is enabled by various surveillance technologies such as radar or GPS. Nowadays ships are usually equipped with full system of marine instruments, one of which is an Automatic Identification System (AIS) transponder. The AIS is an automatic vessel tracking system used on ships and by Vessel Traffic Services (VTS) for identifying and locating vessels by electronically exchanging data with other nearby ships and AIS base stations.
The availability of the global AIS ship tracking data opened the possibilities to develop maritime security far beyond the simple collision prevention. My research at BTH focuses on detecting anomalies in marine traffic in a novel way. The novelty of the method lays in employing the technique of artificial potential fields used, e.g., for designing game AI.
The potential fields are based on actual physical phenomenon of a potential field, and are described in a similar way. The general idea about applying potential fields for marine traffic is for the potentials either to attract or to repel a vessel on the way. A conflict of potentials that have been observed in the past and the potential of a vessel currently in motion indicates an anomaly in the marine traffic.
One of the advantages of the method is the possibility to detect and signalize different gravities of safety threats (i.e., anomalies), depending on the intensity of potentials conflicts. Another desired property is the possibility of visualizing the potential fields. The map-based, visual representation of reported anomalies in marine traffic is more intuitive than a textual list of reported anomalous events.
The problem present in many existing maritime anomaly detection systems is that they inform their users about detected anomalies but do not provide a meaningful explanation why the detections were made in the first place. Additionally, there is a high possibility of false detections. The frequent misdetections and deficient explanation have a negative effect on users trust in the detection software. For this reason the intuitiveness and improved user awareness, enabled by the visualization of potential fields, is one of the aims driving the research work presented in this thesis.
The developed method has been examined in practice using a web-based anomaly detection system STRAND (for Seafaring TRansport ANomaly Detection). The applicability of the method has been demonstrated in publications included in the later chapters of this thesis, and the massive traffic surveillance dataset set up for this study provides an abundance of challenges for future studies.
There are various potential benefits and practical applications of the method, depending on the target user. From a ship navigator's point of view, the display of patterns of correct or normal behavior, aids the choice of the safest and most optimal path. From traffic safeguarding perspective, the anomaly detection based on potential fields may help quickly and comprehensively inspecting possible traffic incidents. Finally, from authorities' point of view, the clear overview of traffic may help recognize traffic regulation and legislation issues.