A simple, deterministic real-time strategy game environment designed for AI and reinforcement learning research.
microRTS is a small, efficient real-time strategy game environment built for AI and reinforcement learning research. It provides a simplified, deterministic platform to test algorithms like minimax and Monte Carlo Tree Search, allowing researchers to experiment with theoretical ideas before applying them to complex commercial games.
AI researchers and academics focusing on reinforcement learning, game theory, and real-time strategy AI, particularly those needing a lightweight, configurable testbed.
Developers choose microRTS for its simplicity, deterministic nature, and built-in AI tools, which enable faster iteration and experimentation compared to heavier alternatives like StarCraft with BWAPI.
A simple and highly efficient RTS-game-inspired environment for reinforcement learning
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Offers predictable gameplay for reproducible AI experiments, as the README states it supports deterministic settings via configuration flags.
Includes hard-coded bots and game-tree search methods like minimax and MCTS, enabling rapid prototyping without external libraries.
Supports both fully-observable and partially-observable modes through flags, allowing diverse research scenarios for AI testing.
Hosts a dedicated AI competition for benchmarking algorithms, providing a standard evaluation platform as mentioned in the README.
The project is no longer updated or supported as of August 2025, limiting its viability for future work and bug fixes.
Requires compiling and running Java code with specific commands, which can be cumbersome compared to modern, script-based environments.
Lacks the complexity of commercial RTS games, which might not fully prepare algorithms for real-world applications despite its research focus.