APPLICATION OF REINFORCEMENT LEARNING TO AIR-TO-AIR FIGHTER TACTICS

Loading...
Thumbnail Image
Authors
Bromley, Christopher M.
Subjects
air-to-air combat
JAAM
BVR
tactics
reinforcement learning
Advisors
Huang, Jefferson
Bassett, Robert L.
Date of Issue
2024-06
Date
Publisher
Monterey, CA; Naval Postgraduate School
Language
Abstract
As threat nations continue to modernize the capabilities of their fighter aircraft and armament, U.S. fighters must update their tactics to adapt. Fourth generation aircraft, which prioritize kinematic defeat of missiles for survival, rely on the Joint Air-to-Air Model (JAAM) software to develop tactics. JAAM runs deterministic simulations of combat scenarios based on predefined weapons employments and maneuvers given by the user. Currently, subject matter experts across the services brute force runs of JAAM to test different scenarios and find tactics that provide desirable outcomes. We solve this problem by applying a Reinforcement Learning algorithm to alleviate the tedium and increase the creativity in developing beyond visual range (BVR) tactics. By providing a value function to a decision-making agent, we allow the agent to independently learn how to best achieve the desired outcome through practice. We discretize the actions available to the agent to only those actions we could expect a human to execute. Our goal is to develop a framework to rapidly modify tactics against current and future threats by allowing the agent to exercise creativity and to try novel methods to achieve the desired outcomes. Ultimately, this will relieve subject matter experts of the need to spend countless hours innovating new tactics and subsequently running iterations of JAAM to validate them.
Type
Thesis
Description
Organization
Identifiers
NPS Report Number
Sponsors
Funder
Format
Citation
Distribution Statement
Distribution Statement A. Approved for public release: Distribution is unlimited.
Rights
This publication is a work of the U.S. Government as defined in Title 17, United States Code, Section 101. Copyright protection is not available for this work in the United States.
Collections