Next Generation Open Mission System (NOMS) Weapon Integration Support

The Weapons Automated Artificial Intelligence Planner (WAAP) utilizes SaigeTEC™ Foundry for data exploration and mining to produce optimized Machine Learning (ML) training data for ML Replicas. The ML Replicas replace legacy components and increase calculation speed by as much as 5000 times despite being 97% smaller on disk. The WAAP process produces operating system (OS)-agnostic replacement components with reduced compute requirements and fewer software dependencies. This provides a basis for modernization efforts, improving scalability, and integration with modern system architectures. The Replicas can be optimized for small computing devices to enable edge deployments in low Size, Weight, and Power (low-SWaP) environments with broad applicability for the DoD and other Government Agencies. The WAAP process was also able to identify invalid outputs (i.e., glitches) from legacy Fly-Out Models (FOMs) and remove them from the ML training data, producing a more trustworthy model than the legacy FOM. The first ML FOM replica was successfully integrated into the deployed Next-Generation Open Mission Systems (NOMS) weapons planner. This effort will be focused to create an ML replica of a classified weapon system FOM. Each weapon FOM is different, and the TACFI investment will expand upon the Phase II technology to prove it can be used to create replicas of modern weapons with more complex behavior. Additional ML technologies will be researched and applied as necessary to model behavior of more complex weapons such as Large Multimodal Models. Information technologies and compute will be purchased for mining data from classified FOMs and training models on classified data. Weapons will be selected by the TPOC.Programmatic and technical risk include timing GFI deliveries and complex weapon fly-out behavior. For timing of GFI deliveries for the selected weapon FOM, a mitigation will be to work with the TPOC to identify classified model with proper government usage rights. In capturing more complex weapon fly-out behavior, a mitigation may be research and testing of new AI/ML technologies for modeling said behavior.Scope of the TACFI includes purchasing server hardware to support ML replication of classified weapon FOMs, and develop ML replicas of weapon FOMs for us by mission planning. Deliverables include ML replicas of classified legacy weapon FOMs, and integration of ML replicas into NOMS weapon planner as a micro-service. Innovations to the warfighter include using AI/ML to quickly integrate weapon FOMs into planning of modern tactical aircraft saving orders of magnitude time and resources. ML Replicas can be run in the field on limited hardware unlike legacy models. Lastly, because of the high speed and low hardware requirements, ML Replicas can be used to support rapid autonomous decision making