State-based Machine Aided Real Time Strategy (SMARTS)

Military operations require fast, decisive, and accurate decision making to accomplish missions with optimal performance and minimization of exposure to risk. Military leaders are forced to make these decisions in high-pressure situations with changing circumstances, incomplete information, very short time frames, and minimal margin for error. Advancements in Artificial Intelligence (AI), specifically Reinforcement Learning (RL) and Deep Learning (DL), are enabling computers to accomplish tasks under similar conditions by recognizing patterns across massive data streams. However, the lack of transparency and explainability of AI systems has made it unfeasible for these decision makers to put lives at risk based on black-box algorithms. In addition, the integration of AI systems into existing military operations presents challenges in how military personnel communicate with autonomous systems. There is a direct need for a system that can understand both the underlying machinations of autonomous systems and the doctrine-based communication patterns of military operations. To solve these shortcomings, the Machina Cognita Technologies (MCT) team is developing the State-based Machine Aided Real Time Strategy (SMARTS) system and the SMARTS Translation Engine. The SMARTS engine provides users with the ability to analyze an array of potential sequences of actions (or decision tracks), the risks associated with each of these actions, and the required capabilities and effectiveness for units to execute the actions. The SMARTS Translation Engine allows autonomous systems and humans to communicate without requiring military personnel to modify existing processes, procedures, and training. In particular, the SMARTS Translation Engine will enable the two-way conversion between military doctrine-based and formatted communication and machine-understandable messages and control. Machina Cognita Technologies (MCT), in partnership with Covan Group and Unitary Labs, proposes to enhance the SMARTS system’s of military doctrine and operations. To accomplish this goal, the MCT team will improve the SMARTS system’s Natural Language Processing (NLP) pipeline and Semantic Reasoning capability built upon Bidirectional Encoder Representations from Transformer (BERT) models, develop an Unstated Knowledge Model including support for speaker personality/military role models, and generate a Socio-Pragmatic Knowledge Repository based on military operations. Through these components, the SMARTS system will have the added functionality of extrapolating unstated tasks, conditions, and criteria based on human-generated operations and communications, ensuring that the translations are fully described and based upon military doctrine and governing documentation.