Hands-On Reinforcement Learning for Games

Understanding visual state

Up until now, we have observed state as an encoded value or values. These values may have been the cell number in a grid or the x,y location in an area. Either way, these values have been encoded with respect to some reference. In the case of the grid environment, we may use a number to denote the square or a pair of numbers. For x,y coordinates, we still need to denote an origin, and examples of these three types of encoding mechanism are as follows:

Three types of encoding state for an agent

In the preceding diagram, there are three examples of encoding state for an environment. For the first example, which is ...