B-ACE: Beyond Visual Range Air Combat Environment

B-ACE is an open-source, lightweight simulation environment designed to facilitate the experimentation and evaluation of autonomous agents in Beyond Visual Range (BVR) air combat scenarios. Built on the Godot game engine, B-ACE leverages Multi-Agent Reinforcement Learning (MARL) to explore advanced techniques in autonomous air combat agent development. The environment provides a flexible and accessible platform for the research community, enabling the rapid prototyping and evaluation of AI-based tactics and strategies in complex air combat settings.

Features

• Open-source and easily extensible: Researchers can easily modify and extend the environment to suit their specific needs.
• Integration with MARL frameworks: Compatible with popular reinforcement learning libraries such as Gymnasium and PettingZoo.
• Simplified yet representative BVR combat dynamics: Focuses on key aspects of air combat, offering a balance between realism and accessibility.
• Supports single and multi-agent learning scenarios: Ideal for exploring both individual and cooperative agent behaviors.

Documentation

Detailed use cases and setup instructions will be available soon. Stay tuned for updates!

Citation

If you use B-ACE in your research, please cite the following paper:

@inproceedings{kuroswiski2024bace,
  author    = {Andre R. Kuroswiski and Annie S. Wu and Angelo Passaro},
  title     = {B-ACE: An Open Lightweight Beyond Visual Range Air Combat Simulation Environment for Multi-Agent Reinforcement Learning},
  booktitle = {Interservice/Industry Training, Simulation, and Education Conference (I/ITSEC) 2024},
  year      = {2024},
  note      = {Accepted for presentation, to be presented in December 2024},
  paper     = {24464}
}

Acknowledgements

This work was supported by the Brazilian Air Force Postgraduate Program in Operational Applications (PPGAO).

License

This project is licensed under the MIT License - see the LICENSE file for details.