Enabling Autonomy in Command and Control via Game-Theoretic Models and Machine Learning with a Systems Perspective

摘要

A Command and Control (C2) system brings technology and humans together for achieving specific missions such as autonomous cars, disaster surveillance, and traffic management. In today's C2 systems, whether it is an air traffic control system or an integrated defense operation, the technological system only aids the human decision-making process and executes commands with limited opportunities to make decisions or perform actions on its own. However, future operational concepts of C2 systems demand agility and rapid decision-making in complex and uncertain environments. In these situations, intelligent systems could substantially complement human capabilities to meet the operational timeliness, precision, and complex needs. The objective of this paper is to formulate a conceptual design of future C2 system, composed of collaborative intelligent multiagent systems that autonomously conduct multi-step actions to efficiently and robustly achieve objectives. Using an example concept formulation of a cyber-physical command guided swarm (CGS), we develop a systems perspective of autonomous C2 that employs state-of-the-art machine learning techniques for developing the intelligent systems that utilize a game theoretic formulation for engineering collaboration strategies between these systems. The crux of the GAme theoretic Machine Learning C2 (GaMLC2) concept lies in introducing autonomy in the independent systems, which remain subordinate to their human commanders/operators while collaborating autonomously to accomplish the commander's intent.