Water Current Field, Flux, and Plume Dynamics Estimation for AUVs in the Absence of DVL Bottom-lock, Using ADCP - aided State Estimation
James Kinsey, Applied Ocean Physics & Engineering
Lashika Medagoda , Applied Ocean Physics & Engineering
2014 DOEI Funded Project
AbstractQuantifying ocean fluxes, along with localization and mapping of plumes, over large regions requires measuring water velocities from mobile platforms (e.g., AUVs, CTDs, gliders). A fundamental challenge in obtaining Earth-referenced water velocity measurements from these platforms is separating the motion of the vehicle from the measured water velocities. Recent advances in state estimation (i.e., estimating the position and velocity of vehicles) improves our ability to correct for vehicle motion and, in consequence, improve our ability to estimate water velocities from mobile platforms. This proposal seeks to apply these advances to obtaining improved water velocity measurements with AUVs. In addition to their autonomy, AUVs possess significant advantages over other platforms including possessing the navigation instrumentation necessary to apply the proposed techniques. AUVs are routinely equipped with acoustic Doppler current profilers (ADCPs) which provide concurrent water velocity and vehicle velocity measurements. Jointly estimating vehicle and environment state results in improved estimates for both since the estimation problem is coupled due to the states being highly correlated in the estimation process. Furthermore, probabilistic approaches quantify and maintain the uncertainty of the quantities in question, along with the correlations in the joint estimate, allowing data quality to be assessed objectively.
This research proposes to explore more sophisticated modeling of the water current environment in conjunction with the ADCP-aided navigation to provide further constraints. We also propose the correlation of the spatiotemporally changing water current states, along with mass transport and shear constraints on the water current field, will be formulated using probabilistic graphical models. The methods investigated in this proposal will be implemented and assessed on data obtained with the Sentry autonomous underwater vehicle (AUV) during an already funded cruise in summer 2014. During this cruise, Sentry will conduct control volume and near-bottom surveys to obtain temperature and water velocity data necessary for quantifying heat fluxes at Axial Seamount and Main Endeavor Field. All of the dives during this cruise will occur within the 200m of the seafloor (i.e., the Doppler velocity log (DVL) will have bottom-lock); however, through data denial we will evaluate the performance of the ADCP and ultra-short-baseline (USBL) solution in order to simulate high altitude trajectories where DVL bottom lock may not be available. If awarded, this proposal would enable a postdoctoral investigator to apply his background in stochastic state estimation and robotics to an important problem in deep ocean science. Expected outcomes of this work would include the methods described above and a peer-reviewed journal article.