A multi model filter based on fuzzy logic for the navigation of an autonomous underwater vehicle, preliminary results

Accurate position estimation is a key factor in any Autonomous Underwater Vehicle (AUV) mission. Generally, Inertial Navigation System which estimates the position and orientation by means of dead reckoning are employed by AUVs to provide location information. Inertial navigation is most accurate when the vehicle is traveling close to the sea bottom and the Doppler Velocity Log can acquire a bottom lock. The estimation is less accurate while the vehicle performs maneuvers such as diving, or resurfacing. This work presents a multi model filter based on a fuzzy logic classifier that detects the vehicle maneuvering state. The information is then provided to a multi model filter and the most suitable combination is selected. Preliminary results shown that the proposed classifier succeeds in estimating the vehicle maneuver while the multi model filter provides higher accuracy than the traditional Extended Kalman Filter.