Detecting maneuvers in maritime observation data with CUSUM

Understanding and building suitable marine traffic and prediction models is a difficult undertaking. To build such a model, it is necessary to understand vessels behavior and have knowledge about the driving patterns. In general, a complex dynamic model for several ship types is used. Maritime sensors like AIS or radar do not always provide the necessary information, like vessel size or type. That is why a simpler solution for detecting maneuvers in maritime observation data is needed. We present a light approach to detect maneuver in maritime observation data without information about the vessel dynamic. For the maneuver detection, we use the CUSUM method which is normally be used in change point detection and use it for geospatial data. Further we present our data fusion process for AIS and radar fusion, both provided from the LABSKAUS infrastructure, to get a better situation picture.