Design and Validation of a Smart Waste Management System Integrating Internet of Things (IoT) and Artificial Intelligence (AI)

This paper presents the design, fabrication, and validation of an end-to-end smart waste management system that integrates AI, embedded IoT, and mechanical automation for real-time waste classification, sorting, and environmental monitoring. The system introduces a novel dual-stage AI pipeline, combining YOLOv5 object detection with a fine-tuned ResNet50 convolutional neural network architecture classifier that categorises waste into six classes: Plastic, Paper, Metal, Glass, Cardboard, and Organics. The AI model, deployed on an Intel NUC Mini-PC (Core i3-3217U), also functions as an IoT gateway, transmitting sensor data from an Arduino MEGA–controlled sorting module to a cloud dashboard. Integrated IoT components include an ultrasonic sensor for fill-level detection, a DHT22 sensor for temperature and humidity monitoring, and a GPS module for real-time geolocation data is extracted from $GPGGA NMEA sentences, ensuring precise tracking of system deployment. The proposed system achieved 90.44% accuracy at a fabrication cost of £1231.17, delivering a cost-effective, real-time AI-IoT waste management solution with strong performance across all categories for scalable smart city deployment. Future work will expand waste categories, enhance sorting speed, and integrate additional environmental sensors to improve scalability and adaptability in diverse municipal contexts