AI-friendly Digital Environments
Grid2Op – Power grid environment
Open-source Python platform designed to simulate and study power grid operations. It enables testing of AI algorithms for grid control under dynamic conditions, including outages and demand variations. Widely used in research and competitions, it supports realistic grid environments and reinforcement learning integration for decision-making under uncertainty.
AI4REALNET Repositories:
https://github.com/AI4REALNET/grid2op-scenario
https://github.com/AI4REALNET/grid2evaluate
https://github.com/AI4REALNET/grid2op
https://github.com/AI4REALNET/pypowsybl2grid
Hosted in Linux Foundation for Energy:
https://lfenergy.org/projects/grid2op/
https://github.com/Grid2op
Flatland – Railway network environment
Open-source multi-agent simulation framework for railway environments. Designed for AI research, it challenges agents to navigate complex rail networks efficiently while avoiding collisions. Featuring stochastic elements and scalable environments, Flatland supports reinforcement learning and planning algorithms, making it ideal for studying coordination, pathfinding, and resource management in constrained spaces.
AI4REALNET Repositories:
https://github.com/AI4REALNET/flatland-blackbox
https://github.com/AI4REALNET/maze-flatland
https://github.com/AI4REALNET/flatland-book
https://github.com/AI4REALNET/flatland-rl
https://github.com/AI4REALNET/flatland-scenarios
https://github.com/AI4REALNET/flatland-baselines
Flatland Association:
https://www.flatland-association.org/home
Bluesky – Air traffic management environment
Open-source air traffic simulator developed for research and education. It offers real-time, fast-time, and human-in-the-loop simulation capabilities for airspace operations. BlueSky supports customizable scenarios, aircraft behavior modeling, and integration with AI and optimization algorithms, making it ideal for testing air traffic management concepts and tools.
AI4REALNET Repositories: