AI-DRIVEN GAME BOT USING REINFORCEMENT LEARNING

This technical report presents the design and implementation of an AI-driven game bot based on Reinforcement Learning. The system applies the Q-Learning algorithm to provide real-time gameplay assistance and autonomous decision-making across three games: Tic-Tac-Toe, Snake, and Memory Match. Unlike traditional rule-based bots, the proposed solution learns optimal strategies through reward-based interaction with each game environment. The platform is implemented using Python with Flask and Flask-SocketIO to enable real-time communication between the reinforcement learning agent and the web interface. Experimental results demonstrate that a unified tabular Q-Learning framework can effectively adapt to multiple structured game environments while supporting interactive hint-based gameplay.