An autonomous path finding strategy for an Artificial Intelligence enabled Lego Mindstorms robot

The objective of this Master thesis is to research how Artificial Intelligence can contribute to make robots more intelligent in order to cope with real world applications. The vision is a self-driving robot which uses different Artificial Intelligence methods and algorithms such as reinforcement learning and deep reinforcement learning in order to make the robot more intelligent by enabling it to learn. The basis is that robots can recognize ways and obstacles by using different sensors. We extended our research to provide the comparison between Q-learning and Deep Q-Network as pathfinding agent in grid map environment. The simulation results show that both of algorithms work well with small dimensional state spaces. Nevertheless, Deep Q-Network can perform a better performance and more stability than Q-learning when states spaces increasing. Afterward, A Lego Mindstorms EV3 applying reinforcement learning algorithms in order to find the path to the target and avoid obstacles will be proposed, developed, and discussed.