DRAGON: diversity regulated adaptive generator online

Abstract

Approaches based on Procedural Content Generation (PCG) are more and more diffused among video game developers. They offer many advantages, among which two of the most notables are the opportunity to lighten the burden of level designers and the possibility to produce personalized experiences for the players. In the present work we focus especially on the second aspect, while the first one is addressed as a side effect. In particular, we present DRAGON (Diversity Regulated Adaptive Generator Online), an algorithm for procedurally generating "monster" archetypes for multiplayer games basing also on the players' preferences. The generation process exploits the genetic algorithm paradigm, opportunely adapted, and modified in order to guarantee enough flexibility to the game or level designers. Ideally, the archetypes produced by DRAGON can be employed for any game genre and setting. DRAGON has been implemented as a plugin for one of the state-of-the-art game engines and tested with game developers. Moreover, a simulation has been conducted for the end-users.Keywords Genetic algorithms (GAs) . Procedural content generation (PCG) . Personalized userexperience . Video games design Multimedia Tools and Applications