AI-Assisted IoT Monitoring for Tracking Construction and Demolition Waste

Integrating Internet of Things (IoT) solutions in managing construction and demolition waste is critical to advancing sustainability in the built environment. This paper presents a robust, energy-independent IoT unit designed for in-situ monitoring of waste containers during the filling process. The device captures images at regular intervals—adjustable via remote downlink messages—and operates on a lithium-ion battery recharged by a solar panel, ensuring uninterrupted functionality in harsh environments without human intervention. A key objective is to gather training data for TinyML-based on-end-node image analysis, leveraging lightweight machine learning algorithms that significantly reduce data transmission requirements. This approach addresses the inherent bandwidth limitations of low-power wide-area networks (LPWAN) while preserving actionable insights.Furthermore, we explore integrating this IoT unit with both LoRaWAN and NB-IoT technologies, proposing a hybrid communication model that balances energy efficiency and performance. End-node computation determines whether to store, transmit, or capture higher-resolution images, enabling detailed analysis only when necessary. Ultimately, this research aims to provide an effective pre-sorting tool for construction and demolition waste at the point of generation.