Design of a Controller for Autonomous Vessel Recovery Utilizing the Prediction of Host Vessel Motions

In this paper predictive algorithms are used to forecast host vessel motions during USV recovery operations. This forecasting is utilized as part of a recovery controller to trigger transition from loitering to initiation of the entry maneuver. This paper provides discussion of the controller and predictor components of the recovery system with application to a hypothetical USV-host vessel combination. Two separate predictors were considered and evaluated against the requirements of the controller, one based upon artificial neural networks, and a statistical based approach utilizing the identification of like points in past trajectory data.