TECHNOLOGY AREA(S): Battlespace, Information Systems

ACQUISITION PROGRAM: PEO-C4I/PWM-120 Littoral Battlespace Sensing (LBS); PMS-406;

OBJECTIVE: Develop reduced order modeling (ROM) techniques for geophysical fluid dynamics models that will enable their use aboard unmanned platforms, including the ability to assimilate local environmental data into the model state. Computational efficiency (low power/memory) and the ability to provide estimates of the flow field to the local platform control system are required.

DESCRIPTION: The Navy operates high-resolution (O(km)) environmental models to provide forecasts of the operational environment to ensure safe and efficient maritime operations. These prediction systems assimilate observations from both in situ and space-based sensors and are run on High Performance Computing platforms. The resulting data sets are large, and the relevant weather and ocean forecast products must be transmitted to naval platforms for use. ROM methods are sought that can predict the evolution of the maritime environment around a platform at sea, incorporating in situ environmental observations to forecast the four-dimensional ocean fields (i.e., temperature, salinity, and velocity vectors) with sufficient fidelity to allow the platform to exploit the information (e.g., optimal path planning or positioning within the water column).

The capability would enable unmanned platforms to sense the local environment and use that information to estimate the ocean conditions for use in autonomous command and control algorithms. Given limited power and computational capability on these systems, ROM techniques are likely to be useful for this application if they are able to provide a useful characterization with limited observational data. The ROM capability must be capable of operating in real time over iridium data rates, incorporating in situ observations, and returning a solution for the predicted ocean environmental information in time to enable exploitation by the platform.

The Phase II effort will require implementation of the developed techniques aboard an unmanned platform. For this purpose, commercial systems of similar capabilities to existing platforms are sufficient � prototyping aboard actual naval unmanned systems is not required. Coordination and testing of software and techniques using existing GFE platforms will be possible but not required. The desired size and power constraints would be aimed for a light weight Unmanned Underwater Vehicle (UUV) class vehicle.

PHASE I: Design and develop ROM method(s) for prediction or reconstruction of oceanographic fields appropriate for use by an in situ unmanned platform. Estimate the capabilities of the proposed method(s). Determine the feasibility of the proposed method(s). Develop a Phase II plan.

PHASE II: Develop tools that incorporate the output of the ROM solution into the autonomous control system of the vehicle. Demonstrate the usage of the predicted environmental information to inform the control algorithm. Integrate and test prototype ROM tools onboard surrogate unmanned systems, which may include platforms operating on or below the ocean surface.

PHASE III DUAL USE APPLICATIONS: Finalize and transition the ROM tool to platform developers to test on U.S. Navy unmanned systems. This technology has the potential to provide better situational awareness for unmanned platforms deployed at sea. Other federal agencies and the ocean technology industry operate unmanned systems that could make use of this capability.

REFERENCES:

1. Majda, A.J. and Qi, D. �New Strategies for Reduced-Order Models for Predicting the Statistical Responses and Uncertainty Quantification in Complex Turbulent Dynamical Systems.� SIAM Rev., 2017a. http://www.ipam.ucla.edu/abstract/?tid=14237

2. Subramani, D.N., Wei, Q.J., and Lermusiaux, P.F.J. �Stochastic Time-Optimal Path-Planning in Uncertain, Strong, and Dynamic Flows.� Computer Methods in Applied Mechanics and Engineering, 333, 218�237, 2018. http://mseas.mit.edu/publications/PDF/Subramani_et_al_stochastic_timeoptimal_planning_CMAME2017.pdf

3. Feppon, F. and Lermusiaux, P.F.J. �A Geometric Approach to Dynamical Model-Order Reduction.� SIAM Journal on Matrix Analysis and Applications, 2017. https://arxiv.org/pdf/1705.08521.pdf

KEYWORDS: Reduced Order Modeling; ROM; Ocean Modeling; Data Assimilation; Dynamical Systems; Unmanned Underwater Vehicle; UUV; Unmanned Surface Vehicle; USV; Autonomy