Maze Solver——Q-Learning and SARSA algorithm

In this project, we simulate two agent by Q-Learning and SARSA algorithm and put them in interactive maze environment to train best strategy

In this project, we simulated the interactive maze environment in the MATLAB real-time editor environment, and implemented two classical Rl (reinforcement learning) algorithms - Q-learning and sarsa algorithm. By creating an agent to move interactively in the maze, two algorithms are used to train the highest incentive value reward and the best maze walking method. Finally, we compare the performance of the two algorithms.