New Methods for Estimating Water Current Velocity Fields from Autonomous Underwater Vehicles

Water current velocities are a crucial component of understanding oceanographic processes and underwater robots, such as autonomous underwater vehicles (AUVs), provide a mobile platform for obtaining these observations. Estimating water current velocities requires both measurements of the water velocity, often obtained with an Acoustic Doppler Current Profiler (ADCP), as well as estimates of the vehicle velocity. Presently, vehicle velocities are supplied on the sea surface with velocity from GPS, or near the seafloor where Doppler Velocity Log (DVL) in bottom-lock is available; however, this capability is unavailable in the mid-water column where DVL bottom-lock and GPS are unavailable. Here we present a method which calculates vehicle velocities using consecutive ADCP measurements in the mid-water using an extended Kalman filter (EKF). The correlation of the spatially changing water current states, along with mass transport and shear constraints on the water current field, is formulated using least square constraints. Results from the Sentry AUV from a mid-water surveying mission at Deepwater Horizon and a small-scale hydrothermal vent flux estimation mission suggest the method is suitable for real-time use. DVL data is denied to simulate mid-water missions and the results compared to ground truth water velocity measurements estimated using DVL velocities. Results show quantifiable uncertainties in the water current velocities, along with similar performance, for the DVL and no-DVL case in the mid-water. This method has the potential to provide geo-referenced water velocity measurements from mobile ocean robots in the absence of GPS and DVL as well as estimate the uncertainty associated with the measurements.