(Solved): Consider the problem of creating a control system for a self-driving car. For our purposes, we sha ...

Consider the problem of creating a control system for a self-driving car. For our purposes, we shall consider a simple self-driving car which is just driving around the street without needing to go anywhere in particular (i.e. no navigation). The objective is for the car to avoid crashing and to keep moving. The car is completely autonomous and it has various sensors and motors to enable it to perceive the world, to drive and to steer. We are going to attempt to model this problem using a Markov Decision Process. (a) Name THREE elements of a Markov Decision Process and explain what they woul represent in the self-driving car control system problem. (b) You decide to use Q-learning to help with the solving of the Markov Decision Process for the self driving car. You decide to use a neural network to represent the Q-function. Explain what the Q-function does in the context of the self-driving car control system and how it relates to the Markov Decision Process. (c) What would be the training data for the neural network? Explain what the elements of the training data represent in relation to the self-driving car problem.