Reinforcement Learning – The Roll-a-Ball Game

In this blog, I introduce the „Roll-a-Ball“ game: a small game I specifically developed for this topic.

The idea is simple: the player has to navigate though a small open pathway and reach a goal. An overview of such a pathway (from two different perspectives) is shown in Figure 1.

Figure 1 – Game Overview

 

This is just an example of one pathway; there are eight different options – one of which is randomly selected when the game starts. When the player reaches the goal, points are rewarded and the game is reset, as shown in Media 1. If the player falls, the game resets as well, but all points are lost, as shown in Media 2.

Media 1 – Rolling

 

Media 2 – Falling

 

Next week, I will look into methods of hacking games like this roll-a-ball.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.

neunzehn − 14 =