Motion prediction for ship-based autonomous air vehicle operations

A ship operating in an open sea environment undergoes stochastic motions which make the deployment and the landing of air vehicles on a ship difficult and potentially dangerous. There is always a delay between the decision to commit and the moment of actual launch or recovery. This paper presents an artificial neural network trained using singular value decomposition methods, genetic algorithms and conjugate gradient methods for the real time prediction of ship motions. These predictions assist in determining the best moment of commitment to launch, or recovery. Predictions generated using these algorithms allow improvements in safety, by facilitating deployment and recovery of air vehicles commonly used on ship platforms. It is shown that the artificial neural network produces excellent predictions and is able to predict the ship motion satisfactorily for up to 7 seconds