Talk: AI for Command and Control of Multi-Domain Operations

The speed and complexity of Multi-Domain Operations (MDO) against a near-peer adversary in hyperactive environments call for high-speed decision-making and execution that may often exceed human cognitive ability. Recently emerging AI techniques such as Deep Reinforcement Learning (DRL) have outperformed human world champions in complex, relatively unstructured, partial-information, strategic games such as Dota 2 and Starcarft II. This may suggest the potential of such AI to contribute to C2 of MDO. However, many questions about the behaviors and limits of such new AI techniques remain unanswered. As part of the ARL Director’s Strategic Initiative (DSI) program, “AI for C2 of MDO”, we are investigating whether DRL might support the future agile and adaptive C2 of multi-domain forces that would enable the commander and staff to exploit rapidly and effectively the fleeting Windows of Superiority. This might be especially relevant to forces with a high proportion of robotic combat assets. To aid this investigation, we have developed two novel C2 testbeds and DRL-based learning within these testbeds. The presentation will include an overview of the project and demonstrate research results where an “artificial commander” executes an integrated planning-execution process in a simulated brigade-scale battle.

Dr. Priya Narayanan is an Engineering Researcher at DEVCOM Army Research Laboratory. She received her doctorate from the Mechanical Engineering Department at the University of Maryland Baltimore County (UMBC), masters and bachelors from the Aerospace Engineering Department at University of Florida (UF), and Indian Institute of Technology, Madras (IITM) respectively. Prior to ARL, She was a National Research Council (NRC) Fellow in the Navy Center for Applied Research in Artificial Intelligence (NCARAI) at the Naval Research Laboratory (NRL). She has led key research programs in aerial and ground robotics including perception integration activities for the final capstone experimentation of the Robotic Collaborative Technology Alliance (Robotic CTA) Program. She currently serves as the Program Manager for “AI for C2 of MDO” Director’s Strategic Initiative (DSI) program that focuses on the development of AI-enabled systems for Command and Control of Multi-Domain Forces.

Dr. Nicholas Roy Waytowich is a Machine Learning Research Scientist with the DEVCOM Army Research Laboratory currently stationed at Aberdeen Proving Grounds in Aberdeen MD. He received his BS from the University of North Florida in Mechanical Engineering (2010) and his MS in Electrical and Computer Engineering (2013) and his Ph.D. in Biomedical Engineering from Old Dominion University, VA. Upon completion of his Ph.D., he began a joint postdoctoral position in the Laboratory for Intelligent Imaging and Neural Computing at Columbia University, NY and ARL developing novel brain-computer interfaces. Two years later, he joined ARL as a civilian researcher and focuses primarily on human-in-the-loop machine learning and AI. Using a variety of machine learning techniques, his expertise and research interests include reinforcement learning, human-in-the-loop learning, multi-agent modeling, and developing better and more computationally efficient AI by leveraging human interaction.

Ms. Anne Logie is a Computer Scientist for the DEVCOM Army Research Laboratory. Her primary research interests include deep reinforcement learning and machine learning. Her previous work applying multi-agent reinforcement learning and generative adversarial networks to predicting missed communications in unreliable environments has led to current research on the Director’s Strategic Initiative project, Artificial Intelligence for Command and Control of Multi-Domain Operations.

ArtIAMAS (AI and Autonomy for Multi-Agent Systems) is a cooperative agreement between the Army Research Laboratory, the University of Maryland College Park, and the University of Maryland Baltimore County.