Motor de juego como plataforma para la aplicación de algoritmos de Inteligencia Artificial

摘要

In recent times an interest in the investigation of Arti cial Inteligence through videogames hasemerged. Areas such as procedural content generation, the creation o arti cial agents capable ofplaying games competently or the creation of agents whose conduct is undistinguishable from thatof a human player attract a growing amount of researchers nowadays. Research in the área of Arti cialIntelligence in games not only can improve the quality of videogames in the future, but also exploitsthe possibilities of the medium, which provides an environment for the simulation of problems thatare analogous to those of the real world.Until now, research has been done mainly using comercial videogames that weren't developedwith the purpose of being used this way. This project aims to develop a platform that has di erentcharacteristics from those o ered by the games that have been used in research until now and thatprovides a convenient interface for the users so that they can implement Arti cial Intelligence agentsthat can be tested within the game.The platform has been developed using the Unity3D game engine and is formed by a mapgenerator, a tactical strategy team-based multiagent 3D game, an API that allows the agents tointeract with the game and a few demonstration agents to show the use of the platform. The game hasbeen designed with the intention of allowing the exploration of aspects of gameplay such as terrainanalysis, cooperation between independent, heterogeneous agents, the formation of hierarchies withinthe teams and the communication of information between agents.