MazeSolver

Reinforcement learning is a subset of machine learning wherein agents have no prior knowledge of the environment but learn progressively as a result of rewards given to them when they perform an action that would lead them to complete a given task. Q-learning is an algorithm in reinforcement learning which employs a Q-Table to determine the total reward an agent may obtain by performing an action in any given state. This project shows how the maze problem can be solved in a more efficient manner by employing multiple agents using the concept of threading. The agents cooperate with each other by updating values to a shared Q-Table.