Aplicación de técnicas de aprendizaje automático supervisables por el diseñador al desarrollo de agentes inteligentes en videojuegos

  1. SAGREDO OLIVENZA, ISMAEL
Supervised by:
  1. Marco A. Gómez-Martín Director
  2. Pedro González Calero Director

Defence university: Universidad Complutense de Madrid

Fecha de defensa: 21 June 2017

Committee:
  1. María Belén Díaz Agudo Chair
  2. Antonio Alejandro Sánchez Ruiz-Granados Secretary
  3. Antonio José Fernández Leiva Committee member
  4. David Llansó García Committee member
  5. David Camacho Fernández Committee member

Type: Thesis

Abstract

Creating the behavior for non-player characters in video games is a complextask that requires the collaboration among dierent roles in the gameindustry with dierent knowledge and motivations, mainly artists, programmers,and game designers. This process has tried to be simplied with amultitude of computational models during years. Nowadays, in the industry,there is a consensus on the use of behavior trees as a standard model, especiallywhen the behavior is complex or some coordination among charactersis needed.A number of graphic editors have been created to simplify the behaviorcreation, allowing systems increasingly powerful and robust. Even so, in theindustry the behaviors are created by programmers. This happens becausethe programmers do not trust the designers, furthermore, the designersfeel overwhelmed when they must think in the language of behavior trees.For that reason, the designers typically only design the behavior while theprogrammers write the code...