FrozenLake-v1
A compact grid navigation benchmark for reaching a goal while avoiding holes and handling slippery movement.
Published by Farama Foundation
Setup and evaluation
Use FrozenLake to compare planning and tabular learning in a small state space. With slipping enabled, the move an agent requests can differ from the move it gets. Report the slip setting and map alongside success rates.
Install with pip install gymnasium and instantiate gymnasium.make('FrozenLake-v1'). The is_slippery option lets you compare uncertain movement with a deterministic baseline.
- Framework
- Gymnasium
- Access
- Local Gymnasium environment. No account required.