Leveraging JS']JSBSim and Gymnasium: A Reinforcement Learning Approach for Air Combat Simulation

This paper explores the application of reinforcement learning in air combat simulations, emphasizing the use of JSBSim and Gymnasium to create realistic combat scenarios. The objective is to leverage artificial intelligence to validate military designs and enhance tactics in the field. The ongoing efforts involve testing and surveying several prominent frameworks for their suitability for cross-domain application towards designing and developing an open-source testbed for RL techniques to execute and visualize the simulations. We share insights gained from experimenting with these tools and highlight the creation of customized scenarios for specific needs. The focus is on collaborative multiaircraft air combat, aiming to improve mission success and reduce casualties. Referencing DARPA's ACE project, the paper showcases real-world AI applications in air combat intelligence. This contribution adds to the discourse on air combat, flight simulation, and reinforcement learning, emphasizing practical implications for military advancements.