Deep Reinforcement Learning Enabled Warfighter Assistant (DICE)

Deep reinforcement learning (DRL) provides strong capabilities to make decisions under uncertain and complicated scenarios. However, there is a lack of explainability of DRL when applied to assist warfighters in training and operational scenarios. To address this critical need, we propose to develop Deep Reinforcement Learning Enabled Warfighter Assistant (DICE). The key innovation of DICE is utilizing DRL to analyze rich battlefield data and make explainable and interpretable recommendations, along with their quantitative assessment with respective to the warfighter’s mission. Approved for Public Release | 22-MDA-11339 (13 Dec 22)