Frozen Lake

Frozen Lake is a benchmark environment used in reinforcement learning that simulates an agent navigating a grid-like surface with frozen and non-frozen tiles, evaluating the effectiveness of learning algorithms in handling environments with inherent symmetries.

Papers using Frozen Lake (11)

Reinforcement Learning Using Known Invariances2026When to ASK: Uncertainty-Gated Language Assistance for Reinforcement Learning2026SafeAdapt: Provably Safe Policy Updates in Deep Reinforcement Learning2026Fast Non-Episodic Finite-Horizon RL with K-Step Lookahead Thresholding2026TSR: Trajectory-Search Rollouts for Multi-Turn RL of LLM Agents2026Internalizing World Models via Self-Play Finetuning for Agentic RL2025Cogito, Ergo Ludo: An Agent that Learns to Play by Reasoning and Planning2025Active Query Selection for Crowd-Based Reinforcement Learning2025Explaining Reinforcement Learning: A Counterfactual Shapley Values Approach2024Optimized Monte Carlo Tree Search for Enhanced Decision Making in the FrozenLake Environment2024An Open-source Sim2Real Approach for Sensor-independent Robot Navigation in a Grid2024