New AUV Adaptive Behaviors for Subsea Data Exfiltration

This paper describes new adaptive behaviors native to the REMUS autonomous underwater vehicle (AUV). The new behaviors represent the expansion of the vehicle's capabilities to monitor vehicle and mission health and progress. We implemented the new framework to facilitate the application of REMUS vehicles to the data mule problem in general and specifically to the scenario of retrieving data from an Ocean Bottom Seismometer (OBS). Access to high-quality data from these off-shore stations would improve our understanding of the internal structure and rupture properties of faults, and could allow advanced warning of short-term seismic risk. However, retrieval of off-shore data remains complicated. Fiber optic cable stations are expensive to deploy; as is sending research vessels to retrieve data from autonomous stations. Oceanographic Systems Laboratory (OSL) collaborated with the Ocean Bottom Seismograph and Optical Modem Labs at Woods Hole Oceanographic Institution (WHOI) to develop a solution. [1] Using an AUV to transport data from the sea floor requires specific autonomy to enhance vehicle endurance, reliability and communications quality awareness. The new capabilities described here address limitations of reactive vehicle monitoring strategies, which focus on problems rather than mission progress. Our new implementation features vehicle introspection infrastructure that allows the AUV to assess progress and make robust choices for performance improvement. OSL developed a suite of REMUS capabilities to allow the AUV to make a long distance transit to the seafloor station, autonomously launch and monitor retrieval of the data, and make the transit back to deliver the data. Depending on the size of the data, data samples or meta data may be communicated by the vehicle over the horizon via Iridium before the vehicle performs the final transit. To broaden the impact of the new vehicle introspection controls, we are currently organizing a system of mission progress metrics and plan to leverage that system to expose adaptive capabilities systematically across objectives.