ANALISIS METODE FINITE STATE MACHINE PADA ARTIFICIAL INTELLIGENCE MUSUH DALAM LINGKUNGAN GAME GENRE METROIDVANIA
Abstract
This study aims to analyze the implementation of the Finite State Machine (FSM) algorithm in enemy AI systems within Metroidvania games and to measure its effectiveness regarding the player experience. The research employs a mixed-methods approach—combining qualitative and quantitative analysis—through case studies of Hollow Knight, Dead Cells, and Castlevania: Symphony of the Night, alongside a questionnaire distributed to 36 experienced respondents. The results indicate that while FSM effectively represents enemy behavior in a structured, developer-friendly manner, it faces limitations regarding behavioral variability when encountering complex situations. The use of Hierarchical FSM (HFSM) proved capable of overcoming the weaknesses of conventional FSM by generating more dynamic and adaptive attack patterns. Questionnaire analysis revealed positive respondent perceptions, with 66.29% agreeing or strongly agreeing regarding the AI's responsiveness, consistency, and contribution to gameplay satisfaction; however, 25.19% provided neutral responses, suggesting a need for improved adaptability. The study concludes that FSM remains an effective foundation for Metroidvania enemy AI, recommending the use of hierarchical approaches or the integration of supporting systems to handle more complex challenges. These findings are intended to serve as a practical reference for indie developers designing efficient and immersive AI systems.





