A Verilog-based Decision Tree-driven Adaptive Control System for Traffic Light Duration

As urbanization progresses, the challenge of traffic congestion has escalated, emphasizing the need for the optimization of traffic control systems. This research presents a Verilog-based, decision tree model driven adaptive control system for traffic light duration, which involves the construction and refinement of a decision tree utilizing a traffic flow dataset, followed by optimization through a post-pruning technique. The optimized decision tree model, along with the adaptive traffic light system, is subsequently implemented using Verilog hardware description language. Results from simulation validation and testing on an experimental box demonstrate that the system accurately predicts current traffic flow based on varying inputs and adjusts traffic light durations in real time. This study underscores the potential for integrating machine learning algorithms with programmable logic devices for dynamic traffic signal control and related applications, while also demonstrating opportunities for future improvements.