All Tools
M
AgentsFreeOpen Source
MINIGRID
Configurable grid-world environments for reinforcement learning
Apache-2.0
ABOUT
Full physics sims are too heavy when you need to debug exploration, memory, or sparse rewards. MiniGrid ships lightweight grid worlds with a Gymnasium API so researchers can iterate on algorithms, curriculum, and partial observability without waiting on expensive simulators.
INSTALL
pip install minigridINTEGRATION GUIDE
1. Prototype a sparse-reward exploration algorithm on MiniGrid navigation tasks
2. Benchmark memory and partial-observability agents on BabyAI-style missions
3. Teach RL concepts with configurable grid worlds that run on a laptop
4. Test curriculum learning by composing MiniGrid rooms, doors, and keys
TAGS
pythonreinforcement-learninggymnasiumgrid-worldenvironmentsfarama