DYNAMIC NPC AI USING REINFORCEMENT LEARNING FOR AN ENHANCED GAMING EXPERIENCE

Non-Playable Characters (NPCs) are vital to interactive gaming environments, yet traditional modeling techniques, such as Finite State Machines (FSMs), have a tendency of producing rigid and predictable behavior. The study explores the use of Machine Learning (ML) and Artificial Intelligence (AI) algorithms for enhancing NPC behavior modeling with the aim of predicting NPC actions and movement with a minimum accuracy of 75%. The research employs a hybrid AI model that combines Reinforcement Learning (RL), Deep Learning (DL), Transfer Learning, and Explainable AI (XAI) to improve NPC decision-making, adaptability, and realism. A prototype game is developed for experimentation and validation of such models in dynamic gaming environments. Results indicate that the hybrid AI model improve NPC adaptability and engagement, reducing limitations linked to rule-based approaches. The study contributes to AI-driven NPC modeling, game design standards, and understanding of realtime AI implementation in games. Future research opportunities may be explored in terms of computational efficiency optimization and ethical considerations for AIdriven NPC behavior.