Artificiell intelligens till spel

Sammanfattning: Focus in this examination project is put on researching basic techniques for structuring decision-making and pathfinding for the artificial intelligence in computer games. Already implemented examples of these structures are summarized and compared to each other to choose fitting solutions to use together with the game engine Unity for the development of a specific game called Unkown. For decision-making a plugin called Behave was chosen. Behave implements behavior trees in a way that makes them easy to create, modify and review. Behavior trees handle decisions in several levels in a hierarchical structure which handles actions separate from decisions and is acyclic in the transitions between decisions. A' pathfinding was the plugin chosen for pathfinding purposes. It is a flexible plugin with support for several different methods for representing the game environment in a searchable graph. The graph is searched with the A' algorithm with different alternativs available for heuristics and smoothening of the path. The method chosen for representing the environment was a grid due to the simplicity and ease of generating it with A' pathfinding. With Behave and A' pathfinding an early version of an agent was created with the ability to navigate past obstacles to different places in the environment, to follow the player and execute specific actions upon contact with the player.