This model is valid for a region surrounding the Horten-Moss crossing in the Oslo fjord in Norway. The region can easily be changed by altering the available data. 

The model is capable of predicting ship behavior up to 50 minutes into the future. It is based on a kNN-PCA-based trajectory prediction approach that yields high accuracy required by autonomous ship operations.

The predicted routes can be used to identify possible future collisions and optimize departure times of autonomous vessels to minimize collision risk. This also reduces energy consumption from unnecessary collision avoidance manoeuvres.

The model has support for dockerization to function as a REST API.