Recursive identification of autonomous underwater vehicle for emergency navigation

This paper mainly presents an emergency navigation method based on the model of autonomous underwater vehicle (AUV). Dead-Reckoning navigation system depends highly on the reliability of Doppler Velocity Log (DVL) and the failure of DVL will result in divergence of the navigation which would lead the AUV to an emergent situation. So an alternative method is needed to provide velocity messages of the AUV in order to ensure the safety of vehicle itself and the accomplishment of mission in that emergent situation. A dynamic model with measurable inputs is appropriate to offer velocities without requirement for additional instruments. In order to obtain a model that can directly serve navigation, this paper presents a recursive least squares model identification method implemented in double integral way which introduces navigation results straight into prediction errors. Simulation results represent the feasibility and effectiveness of the proposed identification algorithm and emergency navigation method.