The Frozen Lake environment, available in the OpenAI Gym library, presents a classic challenge in Reinforcement Learning. In this Grid World scenario, the agent navigates a frozen terrain to reach a designated goal state. This project is dedicated to exploring fundamental RL algorithms and concepts within the context of the Frozen Lake environment. By experimenting with various RL techniques, we aim to develop strategies that enable the agent to effectively learn and navigate the icy terrain toward the ultimate goal. Additionally, the environment offers the flexibility to introduce stochasticity by enabling the slippery mode, adding an extra layer of complexity to the learning process. Through this project, we delve into the intricacies of RL, investigating how agents adapt to both deterministic and stochastic environments to accomplish their objectives.

• Iterative Policy Evaluation

• Policy Iteration

• Q-Learning