Real-time underwater positioning and navigation of an AUV in deep waters

Due to the absence of GPS, navigation of autonomous vehicles underwater requires the integration of various measurements to provide the best location estimate. Usually in littoral waters, adequate navigational accuracy may be obtained by integrating odometry measurements provided by a Doppler Velocity Log (DVL) into an Inertial Navigation System (Aided INS). However, due to the bulk attenuation of seawater at the acoustic centre frequency at which DVLs typically operate, odometry estimates become increasingly unreliable when the vehicle flies more than 200 m above the bottom (depending on the DVL central frequency). Such a case occurs during experiments in deep waters. This work addresses a theoretical and experimental study on the feasibility of navigating the AUVs using a multi-input Extended Kalman Filter (EKF) integrating proprioceptive measurements (i.e., INS data and speed-over-water observations from DVL) with a set of exteroceptive sensor data, when available. The filter was integrated on-board the CMRE Ocean Explorer Class Version C (OEX-C) AUVs, and tested at sea for the first time in deep water during the NATO exercise Dynamic Mongoose’17 off the South coast of Iceland (June-July 2017).