Neural Network Line Follower
A line follower that teaches itself
About the project
The topic of my bachelor thesis was: Autonomous Mobile Robot Controlled by an Artificial Neural Network. The robot should be able to drive through a whole course from the start to the end without losing track of the designated path and to stop in a case of possible collision.
It has been achieved by building a physical robot, creating a computer simulation and implementing an artificial intelligence algorithm using reinforcement learning methods i.e. Deep Q-learning. The ability of the trained network model to make decisions on performing an optimal action for the current state provided by line sensors was tested and described for both the simulation and a real life environment.
Why I made this?
Autonomous robots are something that has fascinated me for my whole life even in their simplest forms e.g. robo-vacuums. Later on, during the studies, I learned about neural networks and instantly developed a great interest in them. Hence I wanted to create a project for my thesis which would combine those two fields and would grant me a better insight of them.