Multimodal Sensemaking in SeaVision
By utilizing the DoD HPCMP systems, our team has developed and trained AI/ML algorithms to analyze large-scale Automatic Identification System (AIS) data and predict ship tracks. The project's primary objective is to enhance maritime operations, increase maritime security, and facilitate international knowledge-sharing. To achieve this goal, we have secured a research partnership with a web-based maritime situational awareness tool, which enables maritime operation center personnel to view and share an array of maritime information. Our approach has shown promising results in identifying anomalous tracks and predicting ship movements within a dynamic maritime environment. We believe that our research has the potential to significantly contribute to the development of more effective and efficient maritime operations and sanction enforcement. 

We utilized DoD HPCMP systems as a collaborative platform to host 1.1 Terabytes (TB) of collaborator data provided by SeaVision to train AI/ML workloads with large-scale AIS data. We also trained a track prediction algorithm using this data. Our team broadly leverages the HPC Portal, interactive jobs (qsub –I), and GPU resources available on Navy DSRC systems to test and evaluate various AI/ML algorithms with applicability to SeaVision use cases foremost of which is the TrAISformer and GeoTrackNet algorithms. 

The project's use cases of anomalous tracks, and track predictions leverage recent deep learning techniques, to discover complex maritime patterns and communicate these insights across the SeaVision community of users.  

In conclusion, the MUSES project is a cutting-edge research effort that utilizes AI/ML and HPC to enhance maritime situational awareness. Our approach has shown promising results, and we believe that our research has the potential to significantly impact maritime domain awareness. We look forward to continuing our research and exploring new ways to apply AI/ML solutions utilizing the HPC as a collaborative platform HPC.
IMPACT
We believe that our research has the potential to significantly contribute to the development of more effective and efficient maritime operations and sanction enforcement.
PRESENTER
Brand, Lodewijk
lodewijk.w.brand.civ@us.navy.mil
NIWC-LANT
CO-AUTHOR(S)
Meredith, Mason
mason.l.meredith.civ@us.navy.mil
Ek, Brian
bryan.t.ek.civ@us.navy.mil
CATEGORY
AI/ML for HPC
SECONDARY CATEGORY
Mod, Sim & Analysis for Decision Making
SYSTEM(S) USED
NARWHAL