AI-Driven Traffic Light Prototype Enhances Urban Flow Coordination in Yaoundé

Researchers in Yaoundé, Cameroon, have developed a prototype traffic light control system that aims to enhance urban traffic coordination. The system integrates an artificial neural network (ANN) for predicting mean travel speeds and a reinforcement learning (RL) model for adaptive traffic signal control. This novel approach offers an alternative to traditional sensor-based data, using artificial intelligence to analyze and respond to traffic conditions.

The team collected field data from Google Maps API for a specific road section in Yaoundé, linking the Messassi junction to the Nlongkak roundabout, to simulate traffic conditions. The ANN for speed prediction demonstrated high accuracy, with a mean square error of 1.095 and an R² of 0.908 for real speeds. The subsequent reinforcement learning model, simulated in SUMO software and guided by rewards based solely on mean travel speeds, successfully generated adaptive signal policies that outperformed fixed-cycle traffic light control systems.

Simulation results unveiled substantial improvements in traffic coordination. The average coefficient of variation of speeds was reduced by 47.75% in one direction and 18.85% in the other, indicating improved coordination. These reductions were attributed, in part, to left-turn distributions along the axis and between directions.

As lead author Eric notes in the paper, "These findings demonstrate the potential of AI-driven approaches combined with mean travel speeds as an alternative to the use of physical sensor-based data, to optimize urban traffic flow and reduce congestions." This research, focusing on a prototype system, suggests a path toward more intelligent urban traffic management systems.