Створення ші-агента для 4X стратегій на основі системи динамічних особистостей

Abstract

The goal of this paper is to create an AI-agent for a 4X strategy game that avoids the shortcomings of typical AI behavior in this genre, such as lack of individuality and low adaptability, using an alternative dynamical pesonality driven system. The proposed architecture for AI agent consists of three layers: the personality archetype layer, that represents who they are, the mood layer, that determines current goals and strategic posture, and memory layer, that represents the information about previous events. Each game turn, the information from all three layers and current game state are analyzed and transformed into actions using response curves. This allows the creation of AI opponents that have specific personalities, but can change their strategy in response to current game events.