Charles River Analytics was awarded a ~$500K contract from the US Navy under the Office of Naval Research’s Minerva Innovative Naval Prototype (INP) program to develop Probabilistic COA Reasoner Enhanced by Meta Reinforcement Learning (PERCIVAL-MERLIN).Potential adversary actions will be predicted using a unique reinforcement learning approach within AI agents, and presented to commanders in a way that’s easy to understandUnder PERCIVAL-MERLIN, we are conducting research to formalize course-of-action (COA) modeling in a novel way that allows commanders and analysts to encode realistic and high performing friendly and enemy COAs in both a human-understandable and computationally defined format.The PERCIVAL-MERLIN system will use this new method to rapidly generate and evaluate candidate friendly and adversary COAs, accurately predicting adversary COAs and delivering the results in a way that makes it easy for commanders to make decisions regarding mission planning, re-planning, and execution.PERCIVAL-MERLIN uses Charles River’s Neural Program Policies, a novel approach to reinforcement learning, to construct artificially intelligent agent models.Contact us to learn more about PERCIVAL-MERLIN and our other AI and probabilistic modeling capabilities.
This material is based upon work supported by the Office of Naval Research under Contract No. N00014-20-C-2063. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Office of Naval Research.Distribution Statement A: Approved for public release. Distribution is unlimited.
This material is based upon work supported by the Office of Naval Research under Contract No. N00014-20-C-2063. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Office of Naval Research.Distribution Statement A: Approved for public release. Distribution is unlimited.
