Logistics Management of Drifters and Predictive Methods for Ocean Sensing

We propose AI capabilities that improve modeling complex ocean dynamics and assist in the logistical management and planning of uncrewed deployment campaigns. Uncrewed sensor platforms sense and collect data for many interests, including tracking chemical and biological tracers, assimilating ocean dynamics observations into forecast models, and enhancing naval situational awareness. The recent emergence of the Ocean of Things has motivated initial development of technologies which serve these domains, including our Float Field Geometric Health (FFGH) software. FFGH provides drifter position and regional sensor coverage, command and control of on-board sensors, and flexible interfaces for streaming float data. We predict future positions of floating platforms with a blend of leading ocean flow models, first-principles physics, and learned float dynamics via FFGH. To improve the logistical management and planning of deployment campaigns, the model will be combined with planning tools, such as command and control capability, deep multi-agent reinforcement learning, and other analytical measures. Phase I will lay the groundwork for novel analysis of uncertainty and sensitivity, data assimilation methods for floats of varied design, and larger scale float planning. Our Phase I effort will seek to implement these software capabilities in an experimental, scalable, deployment to occur in Phase II.