Implementing an autonomous self-driving car

We will now look at implementing an autonomous self-driving racing car that learns to drive by itself on a racing track using deep Q networks. The driver and the car will act as the agent, while the racing track and its surroundings act as the environment. We will be using an OpenAI Gym CarRacing-v0 framework as the environment. The states and the rewards are going to be presented to the agent by the environment, while the agent will act upon those by taking appropriate actions. The states are in the form of images taken from a camera in front of the car. The actions that the environment accepts are in the form of the three-dimensional vector a ∈ R³ where the first component is used for turning left, the second component is used for moving forward and the third component is used for moving right. The agent will interact with the...