Remote Operational Monitoring Unit Leveraging UUV Simulation (ROMULUS)

We propose to prototype an open, modular, UUV Operational Digital Twin called the Remote Operational Monitoring Unit Leveraging UUV Simulation, or “ROMULUS.” An Operational Digital Twin is a virtual representation of a UUV and its environment that is based on physics-based models and machine learning that continuously integrates data from a priori sources, the mission plan (MP), onboard sensors and autonomy services to predict multiple future scenarios of the UUV and its mission. ROMULUS enables decision-making agents (human and autonomous) to monitor the performance of the UUV, resolve emerging problems, and adapt MPs to improve mission outcomes. The ROMULUS prototype will overcome the communications challenges of operating undersea by synchronizing two DTs in parallel: the first onboard the UUV, the other offboard. Key innovations from Phase I were the (1) use of explicit expectations derived from DT predictions to compactly encode high-density UUV state information for efficient transmission over available communications networks, including acoustic channels, (2) tracking of state uncertainty between updates from the UUV, and (3) a modular design which is consistent with the Unmanned Maritime Autonomy Architecture (UMAA) so DTs can interface with a diverse array of autonomy services and add functionality over time. Prior to starting a mission, ROMULUS takes the programmed MP and a priori estimates of the environment in the OPAREA as inputs and simulates over the MP to quantify the distribution of expected future states. During the mission, ROMULUS’ Onboard DT continuously ingests sensor and autonomy data to update estimates of the UUV’s state. When the state deviates significantly from the prior expectation, that triggers encoding a message to synchronize the Offboard DT. This approach is aperiodic and minimizes the comms required to be feasible over low-bandwidth, intermittent channels. By synchronizing states between onboard & offboard DTs, we can forecast the consequences of emerging problems on the vehicle or in the environment, and aids decision makers to adapt the MP if needed. These forecasts are published as services to other elements of the UUV’s Autonomy, which will enable long-duration autonomy by providing the situational awareness needed to adapt the MP without human intervention. Importantly, the synchronization with the Offboard DT also provides critical decision aids for the command team while supervising the UUVs autonomous mission performance to intervene when necessary to improve mission outcomes. We propose to prototype a ROMULUS system and demonstrate multiple use cases that were developed by interviewing Navy stakeholders during Phase I. Demonstrations will leverage a modeling & simulation environment. However, we also propose to conclude the Phase II with a final in-water demonstration by leveraging an existing REMUS 100 and facilities at Woods Hole Oceanographic Institute (subcontractor).