Алгоритми поведінки NPC противників в іграх жанру Metroidvania

Abstract

In this work, an intelligent NPC behavior algorithm for Metroidvania games is proposed based on Behavior Trees and Raycasting. The system integrates dynamic environment analysis to enable autonomous navigation and adaptive interaction with complex geometry, significantly enhancing the nature of AI. The research covers a comprehensive cycle of NPC states, ensuring efficient state management, scalability, and dynamic response to player actions. A practical implementation of the proposed modular architecture is demonstrated within the Unreal Engine 5 using Blueprints.