Compressed Parallel Architecture for AR/VR Training in Data Constrained Terrain (COMPACT)

We propose to design and demonstrate a Compressed Parallel Architecture (COMPACT) for scalable AR/VR training at the point of need. Using a system that Charles River Analytics has demonstrated can support over 15 million simulated entities, COMPACT will provide a low latency solution that can be used in data constrained environments. COMPACT builds upon innovative methods for optimization, tailored to the specific needs of Warfighters using the Synthetic Training Environment (STE). COMPACT achieves optimization through a focus on data compression, and prioritizing data transmission over bandwidth-restricted channels. These optimizations enable COMPACT to support a broad array of environmental conditions (e.g., massive numbers of entities in a Megacity environment) that would otherwise be infeasible to support. COMPACT includes a customizable benchmarking tool, enabling review of its optimization performance. COMPACT will enable the representation of more realistic, immersive, and interactive training environments, helping to ensure that training will transfer to real-world conditions.