Архітектура адаптивного модуля ai на базі Goal-Oriented Action Planning (GOAP) для динамічних середовищ

Abstract

The report explores the architectural features and advantages of using the Goal-Oriented Action Planning (GOAP) algorithm for creating intelligent agents in modern gaming systems. The mechanisms of the planner's operation, the structure of world states, and the description of actions are considered. Methods for adapting NPC behavior to dynamic changes in the game environment through the manipulation of action weight coefficients are analyzed. The effectiveness of using GOAP compared to Behavior Trees for creating emergent character behavior is determined.