SoCTA 2024

Traffic congestion is exacerbated by inefficient traffic light control systems, causing delays and associated problems. This study introduces a reinforcement learning-based approach to optimize traffic light control in various scenarios. A customizable crossroads environment was developed to simulate different traffic conditions by adjusting key parameters. Using the Q-learning algorithm, the reinforcement learning agent was trained to make optimal traffic signal decisions. Validation tests in moderate and low traffic environments demonstrated the effectiveness of this approach, yielding promising results in reducing congestion and improving traffic flow.

Dec 28, 2024 10:00 AM — 4:00 PM