Reduced-Order Underwater Vehicle Dynamics Model Determination Considering Navigation Performance

This paper proposes a novel approach to VDM(vehicle dynamics model) determination problem for model-aided navigation of an AUV(autonomous underwater vehicle) when the DVL(doppler velocity log) is unavailable. The conventional algorithms have been developed based on the well-known GHM(Gertler-Hagen hydrodynamics model). Unfortunately, the GHM is generally high-ordered so that it is hard to design GHM based applications working stable in actual situations with high uncertainties. To improve this problem, idea to use the transfer function for ground-heading control loop instead of GHM was proposed but it is still high-ordered to retain the stability problems. In this paper, we suggest a revolutionary reduced order VDM for AUV model-aided navigation. The order of our model is drastically reduced enough to show stable and consistent performances in real applications, while the errors due to the model reduction are mitigated under the newly introduced concept of zero model-output error integral to produce small dead-reckoning navigation errors. Simulation results show that our method demonstrates critically improved performances, compared to conventional EKF(extended Kalman filter) methods using the closed-loop heading control transfer function, in stability and consistency.