Urban traffic congestion at signalized intersections is a significant challenge due to the inflexibility of fixed-time control systems. While Reinforcement Learning (RL) offers optimization, it comes with complexities in data, interpretability, and deployment. This report presents a heuristic-based adaptive traffic signal control (ATSC) system within a microscopic Pygame simulation. The system uses dynamic green time calculation and phase prioritization based on a weighted score of maximum waiting time, queue length, and an aging mechanism. Comparative analysis against a fixed-time baseline showed improvements, including a 36.9% reduction in average vehicle waiting time and a 17.6% increase in intersection throughput. The architecture supports integration with computer vision (YOLO/CNN) for realworld input.