Simulation and Machine Learning for Decision Support and Autonomy in Air Combat Operations

Abstract

This thesis presents a research framework for integrating simulation and machine learning to support decision-making in air combat operations. Focused on Beyond Visual Range (BVR) air combat, which represents the modern modality of aerial warfare, the study explores five key areas: simulation tools and services, weapon systems, in-flight tactical decision-making, aerospace data analytics, and aerial autonomous agents. It reviews existing applications, identifies research gaps, and proposes targeted solutions across these domains. The contributions include the development of simulation tools and services tailored to defense scenarios; the modeling of weapon systems with real-time engagement zone prediction and kill probability estimation; the use of machine learning techniques to enhance in-flight tactical decision-making through improved missile launch timing, maneuver planning, and situational assessment; the application of data-driven approaches to support large-scale aerospace analytics; and the training of autonomous agents using imitation, generative, and reinforcement learning to perform complex air combat tasks. The proposed research framework is intended to support ongoing efforts to advance decision support and autonomy in modern air forces.

@phdthesis{dantas2025simulation,
  author = {Dantas, Jo\~{a}o Paulo de Andrade},
  title = {{Simulation and Machine Learning for Decision Support and Autonomy in Air Combat Operations}},
  school = {Instituto Tecnol\'{o}gico de Aeron\'{a}utica},
  year = {2025},
  url = {http://www.bdita.bibl.ita.br/tesesdigitais/80511.pdf}
}