Hexagonal–Tile Map generation using Reinforcement Learning-based Wave Function Collapse

Key Points

• The research aims to improve the visual quality and structural consistency of hexagonal tile maps generated via procedural content generation techniques.
• Developed a reinforcement learning-based Wave Function Collapse (WFC) algorithm.
• Implemented intermediate rewards for local pattern quality and a final reward for global aesthetic quality.
• Utilized Proximal Policy Optimization (PPO) for policy training.
• Achieved significant improvements in structural consistency of the generated maps.
• Enhanced visual quality of hexagonal tile maps compared to traditional WFC methods.

Abstract