Pivot point estimation based advanced ship predictor evaluation with vessel maneuvers under sea trial conditions

To enhance Situation Awareness (SA) in the context of autonomous ship navigation within a complex navigation environment, the Advanced Ship Predictor (ASP) is proposed as a solution framework aimed at predicting ship maneuvers. This can be used to identify potential ship close encounters and collision scenarios in advance, where appropriate collision avoidance actions should be taken. The implementation of the localized ASP is divided into three stages. In the first stage, Kalman Filter (KF)-based techniques with kinematic motion models are employed to estimate vessel navigation states. The second stage involves calculating the pivot point (PP) from these estimates using a Gaussian Process Regression (GPR) model. Finally, in the last stage, a trajectory prediction algorithm that accounts for the characteristics of the vessel PP is employed to provide trajectory predictions. This study also aims to validate the local-scale prediction of the ASP by using sea trial experimental data in real ocean environments. Therefore, several data sets from two ship maneuvers executed by the UiT research vessel, Ymir RV, are used to validate the proposed ASP. The real-world validation results demonstrate that the applied KF-based algorithms and kinematic motion models are consistent with the simulation results. It is concluded that the vessel state estimation and calculated PP of the ASP are adequate for real ship navigation situations and have the potential to reduce the risk of collisions and near-miss incidents. • A local-scale advanced ship predictor (ASP) for autonomous ships is proposed. • The KF-based algorithm is implemented to estimate vessel states. • Gaussian process regression (GPR) method is used to calculate the vessel pivot point (PP). • The calculated PP is used to support the trajectory prediction in local-scale.