New AI Training Tool Targets Air Combat

RTDynamics has released a Python Gym-compatible environment for training AI agents in within-visual-range (WVR) air combat scenarios using reinforcement learning.

The environment uses RTDynamics’ simulation model library to offer a high-fidelity WVR scenario simulating one-on-one aerial engagements. Fully compatible with the OpenAI Gym interface, it can be integrated into modern AI training workflows.

Flight dynamics are powered by FixedWingLib CGF, a six-degree-of-freedom (6 DOF) simulation framework that includes a library of high-level maneuvers performed by a virtual pilot. Radar, electronic warfare, and weapons systems are modeled using EWAWS, RTDynamics’ suite of electronic warfare and weapon simulation modules.

Optimized for real-time and faster-than-real-time execution, RTDynamics’ models enable efficient AI training with reduced computational demands—helping to significantly lower development costs.