Improved Non-Player Character (NPC) behavior using evolutionary algorithm—A systematic review

Abstract

Games, once solely intended for entertainment, have emerged as a significant research focus in recent years, with the primary goal of enhancing the gaming experience. Research in the gaming domain has expanded to encompass a wide range of topics, spanning from game theory to artificial intelligence. Within the realm of artificial intelligence itself, Non-Player Characters (NPCs) play a crucial role in shaping the overall gaming experience. The quality of NPC behavior directly influences player satisfaction. Evolutionary algorithms stand out as a key algorithm for optimizing NPC behavior and interactions. This review paper extensively explores the intricate relationship between evolutionary algorithms and NPC behavior, proposing six categories (planning, user interaction, position modification, parameter modification, character state modification, and target assignment strategy), each delineating a distinct role for evolutionary algorithms. Ultimately, the paper draws three main conclusions: the pervasive use of evolutionary algorithms in gaming research, the diversity in game selection for research trials, and the varying strategies employed by researchers in selecting testing techniques. This comprehensive review aims to serve as a valuable reference for future research, particularly in the domain of evolutionary algorithms applied to NPC behavior.