donaldAitchedji/AIRA-project

AIRA (Artificial Intelligence for Road Automation) is an intelligent adaptive traffic control system tailored to the urban context of Benin (Cotonou, Abomey-Calavi). # AIRA-project **AIRA (Artificial Intelligence for Road Automation)** is an intelligent adaptive traffic control system tailored to the urban context of Benin (Cotonou, Abomey-Calavi). By replacing traditional fixed traffic light cycles with a Deep Reinforcement Learning (Deep RL) approach, AIRA dynamically adjusts phase durations in real time. The system relies on a multimodal perception pipeline combining visual detection using YOLOv8 to estimate traffic density and a CNN model (MFCC) trained on a local audio dataset for acoustic emergency vehicle detection (ambulances, fire trucks). The entire system is tested and validated using a SUMO-Python co-simulation setup. **Key Features**: 🚦 **Dynamic Control**: Deep Reinforcement Learning 👁️ **Visual Perception**: YOLOv8 (vehicle detection & counting) 🎙️ **Acoustic Detection**: MFCC + CNN (emergency vehicles) 🚘 **Co-simulation**: SUMO & Python (TraCI)