A Dynamic Traffic Light Control Algorithm to Mitigate Traffic Congestion in Metropolitan Areas.

Area of Science:

• Artificial Intelligence
• Transportation Engineering
• Computer Science

Background:

• Traffic congestion is a significant issue in urban areas, leading to increased travel times and emissions.
• Current traffic signal control systems often struggle to adapt dynamically to changing traffic conditions.
• Optimization of traffic signal timing is crucial for improving urban mobility and reducing environmental impact.

Purpose of the Study:

• To propose a novel Convolutional Neural Network (CNN) model for the Signal Distribution Control Algorithm (SDCA).
• To maximize dynamic vehicular traffic signal flow at each junction phase.
• To enhance traffic signal timing optimization using Deep Q-learning.

Main Methods:

• Developing a CNN model integrated with the Signal Distribution Control Algorithm (SDCA).
• Deconstructing the Multi-Directional Queuing System (MDQS) architecture to identify optimal routing policies.
• Utilizing Deep Q-learning methodology with a quad agent for enhanced decision-making.

Main Results:

• The proposed algorithm successfully determines optimal reward values and new states for traffic scenarios.
• The CNN-SDCA model, combined with Deep Q-learning, demonstrates superior performance in optimizing traffic signal timing.
• The developed method significantly outperforms traditional traffic signal control approaches.

Conclusions:

• The CNN-based SDCA model offers an effective solution for dynamic traffic signal optimization.
• Deep Q-learning enhances the adaptability and efficiency of traffic signal control systems.
• This research contributes to improving urban traffic flow and reducing congestion through intelligent signal management.