MAP-Elites algorithm generates higher-quality FPS game maps
New point-line and spatial-layout representations boost map diversity by 40%.
A team led by Simone de Donato, Pier Luca Lanzi, and Daniele Loiacono (Politecnico di Milano) has published a new paper on arXiv demonstrating how the MAP-Elites algorithm can procedurally generate superior FPS game maps. The researchers compared traditional map representations (All-Black and Grid-Graph) with two novel formulations they introduced: Point-Line and Spatial-Layout. These new representations encode both topological features (like corridors and rooms) and emergent properties that require actual gameplay simulation to evaluate.
The team applied MAP-Elites with Sliding Boundaries (MESB) to evolve populations of FPS maps over multiple generations. Their key finding is that both Point-Line and Spatial-Layout representations achieve higher diversity and quality scores than the older methods. The system can automatically generate a wide range of playable map layouts, potentially reducing manual level design effort. The research also defines specific metrics for topological and emergent properties, making the approach reproducible for game studios and AI researchers.
- Two novel map representations (Point-Line and Spatial-Layout) outperform traditional All-Black and Grid-Graph methods.
- MAP-Elites with Sliding Boundaries (MESB) algorithm evolves diverse, high-quality FPS maps automatically.
- Emergent gameplay properties are evaluated through actual simulation, not just static topology metrics.
Why It Matters
Automated map generation could slash level design costs and create more varied, replayable FPS games.