Adaptive traffic lights management system using reinforcement learning to reduce traffic congestion in Nairobi

Abstract

Traffic congestion has become a significant issue in urban areas, especially in rapidly grow-ing cities like those in Kenya. The inefficient management of traffic lights contributes to increased travel times, air pollution, and overall frustration among commuters. Traditional traffic light sys-tems, which follow fixed schedules, are often ill-equipped to handle the dynamic nature of traffic flow. This study developed the Adaptive Traffic Lights Management System Using Reinforcement Learning to address these challenges by optimizing traffic light control to reduce congestion and improve traffic flow. The simulation environment was developed using a custom-built web application designed to replicate real-world traffic conditions dynamically. This web-based simulator modeled a com-plex urban intersection, simulating vehicle movements across multiple lanes with varying traffic densities and arrival rates. It allowed the reinforcement learning (RL) agent to interact with realistic traffic scenarios by processing live and historical traffic data. The system integrated Google Maps data to enhance accuracy, ensuring that congestion levels and intersection dynamics reflected real-world conditions. This tailored simulation platform provided an interactive and scalable environ-ment for evaluating the performance of the Adaptive Traffic Lights Management System Using Reinforcement Learning to Reduce Traffic Congestion in Kenya. The system demonstrated a potential to reduce traffic congestion based on simulation re-sults. It showed a significant improvement in traffic flow and a reduction in waiting times at inter-sections, especially during peak hours. The RL agent effectively optimized traffic light timings, leading to smoother traffic movement and less congestion in the simulated urban environment. Keywords: Efficient traffic management, Reinforcement machine learning, Deep Q-Learning, adaptive traffic lights