Deep Learning in Games to Improve Autonomous Driving

No Thumbnail Available
Borghi, Rafael
Issue Date
MSc in Data Analytics
Dublin Business School
Items in eSource are protected by copyright. Previously published items are made available in accordance with the copyright policy of the publisher/copyright holder.
Autonomous Driving is a lively research area in Artificial Intelligence that aims to automatize procedures for self-driving vehicles. Machine Learning has overcome human performance in some simulators with the development of techniques like Deep Reinforcement Learning and Convolutional Neural Networks, as they have been applied with great success to a variety of applications, such as video games. The objective of this work is to demonstrate established Deep Learning techniques used in videogame simulators to show proficiency in Autonomous Driving. In this research, concepts of Autonomous Driving, Video Games and Deep Learning are presented. For that, a simulator is used to enable the use of Machine Learning. Also, a technique related to Deep Learning is evaluated and chosen for the demonstration of conduct vehicles automation in the videogame simulator. Variables related to the agent and the environment (such as distance, steer variation, among others) are taken into consideration.