A constructive approach to strategy game map generation

Abstract

Designing strategy games poses unique challenges due to the complicated nature of the genre where balancing game elements is a major concern. Maps in strategy games commonly contain elements such as resources to gather and buildings with different capabilities, playing a crucial role in the game life cycle. Resource and base placements must be done carefully. It is difficult to design maps manually and even harder to design multiple variations. This study presents a method for procedurally creating strategy game maps. A strategy game was developed and used as a test-bed for the study. A constructive approach is followed in the main steps of the algorithm design and a novel procedural noise function is used together with a uniform stratified sampling method to create and distribute the resources. Then eligible base locations are detected and scored according to a fitness function that defines how favorable a location is. These scores are used to place bases appropriately. The algorithm is evaluated using simulations and a small-scale user study is conducted at the end. The proposed method is capable of creating playable, diverse, and fair maps and is fast enough to work at runtime.