Played By A.I.
Against A.I.
As a means to teach about Artificial Intelligence in video games, as students, we were asked to design an A.I. that could play a game for as long as possible before either the game ended or the A.I. character, inevitably, died.
Seeing as I chose a rogue-like, I faced the latter situation. This is a simple game:
two types of enemies (Orc and Ogre)
six items (healing potion, lightning/fire/confusion magic scrolls, sword, and shield)
stairs to the next level
level system (health and damage)
built with Python/PyGame
As a gameplay programmer, my contributions to this game are focused on the AI decision-making, Actions, and Pathmaking, as is shown in slightly more detail below.
Finding The Way
To progress in the game, one must find the stairs that lead to the next levels.
To do so as an A.I., a pathfinder function leveraging the A* search algorithm was developed.
A* was used as its main drawback of high space complexity has little effect on such a simple game that uses such little memory.
In broad terms, the algorithm uses cost values attributed to each tile in the game to determine the least amount of “cost” to go from the A.I. character’s location to the stairs, ultimately, defining a path.
Decide…..
As Best You Can
After concluding the path to take, every action is a decision waiting to be pondered upon. Conditions, context, and information available, all come into play when choosing what to do next.
To design A.I. to be capable of executing such checks and deciding what to do next based on said information, Behavior Trees are the go-to.
