Autonomous Virtual Vehicle Using Reinforcement Learning

Abstract

Artificial intelligence plays a vital role in self driving vehicle. The direction in which the automotive industry is headed it is expected that autonomous vehicles capable of driving without human supervision will be released to market in the next decade. The development of such intelligence is fairly in its early stages. The problem of creating such intelligence is that the real-world environment is ever-changing. To solve this problem the autonomous vehicle needs to have such intelligence that it can cope with the changing real-world environment. We though can achieve such intelligence by simulating an autonomous self-driving agent in a virtual 2D platform where it will be able to follow the track on its own. Using supervised learning to solve this problem will not be an efficient approach to this problem because no matter how much training and testing is done, it will not be able to keep up with the dynamic real-world environments. Therefore, we are proposing a novel approach for creating autonomous AI powered virtual vehicle using Reinforcement Learning.