Adaptive Markov Inference Game Optimization (AMIGO) for Rapid Discovery of Evasive Satellite Behaviors

Space superiority requires space protection and space situational awareness (SSA), which rely on rapid and accurate space object behavioral and operational intent discovery. The presence of adversaries in addition to real-time and hidden information constraints greatly complicates the decision-making process in controlling both ground-based and space-based Air Force surveillance assets. The focus of this project is to develop robust and near real-time algorithms that rapidly discover the behavioral patterns and operational intent of potentially evasive and/or ambiguous active resident space objects (RSOs). The problem of behaviorally evasive intent identification is challenging for several reasons: (i) partial observable actions; (ii) evasive resident space objects; (iii) uncertainties modeling and propagation; and (iv) real-time requirement and computational intractable algorithms. In phase I, we have developed a game theory enabled machine learning solution. We have tested the solution on Lockheed Martin’s Space Fence dataset and obtained supportive and promising results. In the Phase II, we will refine and expand the Phase I technologies to unify game theory, uncertainties propagation, and machine learning for the rapid discovery of evasive satellite behaviors. We will evaluate the performance of the developed executable prototypical system by using realistic metrics, real-world data (Lockheed Martin Space Fence), and operational constraints.