ME5406P1-Frozen-lake-Reinforcement-Learning

This project explore three reinforcement learnings method to solve the 4x4 Frozen Lake problem on OpenAI gym and it's variation with a 10x10 grid.

Three reinforcement learning methods used in this projects are

• First Visit MonteCarlo Control without Exploring Start
• SARSA
• Q-learning

All results of the three methods are displayed with heatmap shown below

4x4 Frozen lake Problem

Total episodes run: 2000

1. First Visit MonteCarlo Control without Exploring Start

2. SARSA

3. Q-learning