EUVIMEDEUVIMED entdecken
Crossref

From Experience to 3D Platformer Levels via Topological Graphs

Kaijie Xu, Clark Verbrugge

Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · 08.11.2026 · 22 (1) · 434-443

DOI: 10.1609/aiide.v22i1.43601 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, online erschienen am 08.11.2026. Verfasst von Kaijie Xu und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Research on 3D game level generation remains limited, especially for methods that combine structure and topological constraints on playability during spatial realization. We address this gap by introducing the Experience Topological Graph (ETG), a compact representation of progression order, branching, and gating semantics, and by building an ETG-to-level generation pipeline grounded in a five-dimensional state model (3D space, time, and progression phase). The pipeline integrates incremental construction with a local topological validator to check progression consistency, prevent unintended shortcuts, and maintain experience-level structural fidelity during generation. We evaluate the approach in a Mario-like 3D platformer simulation under strict topology validation and compare against constructive, lane-based, evolutionary, and constraint-programming baselines in a multi-scale benchmark. Results show that the proposed method achieves the highest playability, controllability, and topology scores in our benchmark, while remaining competitive on pacing variation. An additional case study in a Fall Guys Simulation environment shows that our framework can be adapted to a distinct 3D platformer-like obstacle-course domain. These results support topology-aware experience modeling as a practical approach to controllable 3D level generation while highlighting remaining challenges in robust 3D PCG.

Zitieren

APA

Xu, K., & Verbrugge, C. (2026). From Experience to 3D Platformer Levels via Topological Graphs. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 22(1), 434-443. https://doi.org/10.1609/aiide.v22i1.43601

Vancouver

Xu K, Verbrugge C. From Experience to 3D Platformer Levels via Topological Graphs. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment. 2026;22(1):434-443. doi:10.1609/aiide.v22i1.43601

BibTeX
@article{Xu2026,
  author = {Xu, Kaijie and Verbrugge, Clark},
  title = {From Experience to 3D Platformer Levels via Topological Graphs},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment},
  year = {2026},
  volume = {22},
  number = {1},
  pages = {434--443},
  doi = {10.1609/aiide.v22i1.43601}
}
Kostenlos starten →