Futures Emergency Management through Artificial Intelligence

As the Army Emergency Management (EM) Leader, researching and developing performance enhancements to Installation technologies and tactics, techniques, and procedures across the DoD, ARDEC strives to continually advance knowledge and expertise to optimize processes to ensure mission readiness and installation preparedness across the Joint community. Installation Emergency Management aligns to the Army Modernization Network/C3I priority. The Acting Assistant Secretary of the Army for Installations, Energy and Environment, Mr. J. Randall Robinson's vision and future focus on "Installations of the Future" includes integrating Resources, Communities, Infrastructure, Services, Soldiers and Ranges and Land to improve force protection, individual and unit readiness across the ARMY. This Small Business Innovation Research Phase I project will develop knowledge in AI computing algorithms and methods for the purpose of identification, prevention, response, and recovery of human-initiated emergency incidents as an enhancement to the ARDEC developed Physical Security Integration Framework (PSIF). Currently the PSIF Enterprise Architecture allows for the integrated use and management of Installation processes, technologies, personnel, and business practices in the areas of Daily Operations, Pre-planned Events and No-notice Incidents. An understanding of emerging AI techniques, as well as the applicability of various data sources to key personnel during an emergency event will be refined and better understood to optimize Installation Emergency Management functions. The research conducted in Phase I will inform the development of an integrated application to the existing PSIF framework that would be able to support machine learning of key words and relationships and threat data analytics that when correlated would present a trigger or alert for a security officials to review or act upon. There is no consolidated criminal investigation database to allow for the search or analysis of criminal behavior that could affect the ability to detect a potential threat. Therefore, a feasibility study of related, existing databases would be conducted to determine relevant data and potential access and privacy concerns for each considered data source. An investigation of data mining, machine learning, and synthetic perception will be conducted to further understand the architecture of an EM-driven AI software system. Evaluation of content to detect emerging events and threats as they're developing, and methods of pushing alerts to users based on user-defined areas and topics of interest would also be included in Phase I. This research and application can ultimately provide Installation EM personnel and law enforcement with predictive trends, which would feed decision-making during all phases of an emergency.