Hands-On Q-Learning with Python

Toy text

Toy text environments such as Taxi, which we solved in Chapter 4, Teaching a Smartcab to Drive Using Q-Learning, are simple tasks designed to get you started with RL algorithms. Their state spaces are small, discrete, finite, and easily enumerated.

The following is a rendering of the Taxi environment we solved in Chapter 4, Teaching a Smartcab to Drive Using Q-Learning. Notice that toy text environments lend themselves to discrete state spaces because the states depend on the positioning of the characters onscreen.

On the other hand, a simulation of a real-world environment, such as in a video game, necessarily has to have a continuous state space because the rendering of the game changes multiple times per second:

Toy text problems, ...