AI-Driven Frameworks and Sensor-Robotic Networks for Urban Waste Management: A Systematic Review Toward Sustainable Smart Cities

Abstract

This study presents a systematic and quantitative review of artificial intelligence (AI) applications in modern waste management systems, with a focus on environmental chemistry, process
optimization, and sustainability outcomes, supported by quantitative performance comparisons from recent literature. The review systematically evaluates AI-driven approaches across the entire waste lifecycle, including generation, monitoring, sorting, transportation, recycling, energy recovery, and final disposal. Emphasis is placed on machine learning (ML), computer vision,
robotics, and wireless sensor networks for improving operational efficiency, cost reduction, and environmental performance. The analysis highlights recent advances in AI-assisted waste
chemistry, including carbon emission prediction using FTIR-ML hybrid models, optimization of anaerobic digestion and biogas generation, and predictive modeling of plastic pyrolysis processes for hydrogen and biofuel recovery. AI-based surveillance systems using Convolutional Neural Networks (CNNs), Unmanned Aerial Vehicles (UAVs), and satellite data demonstrate high accuracy in illegal dumping detection and spatial waste monitoring, with reported classification accuracies exceeding 90-98% in several studies. Quantitative comparisons indicate improvements in collection efficiency, recycling accuracy, and cost reduction relative to conventional methods, with reported efficiency gains of approximately 20-40% and operational cost reductions ranging from 15-30%, depending on system scale. However, the review identifies critical limitations, including data scarcity, model interpretability challenges, high computational cost, and the “reality gap” between laboratory datasets and real-world waste conditions. The findings indicate that isolated AI solutions often fail to achieve scalable and context-adaptive performance due to fragmentation, limited data integration, and a lack of environmental feedback mechanisms. This study uniquely proposes a unified cyber-physical-environmental framework that integrates sensing, AI analytics, and environmental feedback mechanisms across the urban waste management lifecycle. The study concludes that integrated, explainable, and hybrid AI frameworks are essential to enable scalable, context-aware, and policy-relevant waste management systems. Such systems can significantly support circular economy strategies, climate change mitigation, and public health protection in rapidly urbanizing regions.