Deep Learning with R Cookbook

Getting ready

In this recipe, we will solve a grid navigation problem. The following image is a pictorial representation of the navigation grid. It represents a navigation matrix, where each state has been assigned a label. Each cell in the matrix represents a state, leading to a total of four states. The agent should navigate from any random starting state to the final goal state, 4. The agent can only move between states through the openings in the grid and cannot move off the grid walls. At each state, the agent can perform any action from the available set of actions; that is, they can move up, down, left, or right. When entering the goal state, the agent earns a reward of 100, and every other additional step costs a penalty of -1.