Game level layout generation using evolved cellular automata

Abstract

Design of level layouts typically involves the production of a set of levels which are different, yet display a consistent style based on the purpose of a particular level. In this paper, a new approach to the generation of unique level layouts, based on a target set of attributes, is presented. These attributes, which are learned automatically from an example layout, are used for the off-line evolution of a set of cellular automata rules. These rules can then be used for the real-time generation of level layouts that meet the target parameters. The approach is demonstrated on a set of maze-like level layouts. Results are presented to show the effect of various CA parameters and rule representation. © 2016 Taylor \u26 Francis

Procedural Dungeon Generation: A Survey

“…Dormans and Bakkes (2011) and Lavender and Thompson (2017) generated mis sions from Graph Grammars and then generated the dungeon levels with Shape Grammars. Similarly, Smith and Bryson (2014) Pech et al (2015), Pech et al (2016) and Kreitzer et al (2019) used GA to evolve CA rules which generate the dungeon levels. More specifi cally, CA rules generate the mazelike dungeon levels, while the GA evolved the CA rules to satisfy some level's con straints.…”

“…Within these 29 pa pers, 17 papers are solely searchbased generative processes -e.g., Pereira et al (2018). Nine of the generateandtest works are hybrid, with emphasis on the works of Pech et al (2015), Ashlock (2015), Pech et al (2016) and Kreitzer et al (2019) that hybridize a searchbased approach (GA) with a constructive approach (CA). One of the hybrid works pre sented a MLbased solution for dungeon generation together with a CAinspired approach.…”