Drone-Based Pollution Detection Using IoT Sensors and Machine Learning

Abstract

Air pollution remains one of the most critical environmental and public health challenges worldwide, driven by rapid urbanization, industrialization, and increasing vehicular emissions. Accurate and high-resolution air quality monitoring is essential for understanding pollution dynamics and implementing effective mitigation strategies. Conventional air quality monitoring systems primarily rely on fixed ground-based stations, which are costly, provide limited spatial coverage, and fail to capture vertical and localized variations in pollutant concentrations. To address these limitations, this study proposes a drone-based pollution detection system that integrates Internet of Things (IoT)–enabled sensors with machine learning–based data analytics for real-time and adaptive air quality monitoring.The proposed system employs an unmanned aerial vehicle (UAV) equipped with lightweight environmental sensors capable of measuring key air pollutants, including particulate matter (PM2.5 and PM10), carbon monoxide (CO), nitrogen dioxide (NO₂), along with temperature and relative humidity. The UAV collects geo-referenced pollution data at varying altitudes and locations, enabling three-dimensional air quality assessment. Sensor data are transmitted wirelessly to a cloud-based platform using IoT communication protocols, where preprocessing techniques such as noise filtering, normalization, and outlier removal are applied to enhance data reliability.Machine learning algorithms are utilized to analyze both historical and real-time sensor data for pollution prediction and hotspot identification. Multiple supervised learning models, including linear regression, support vector machines, and random forest regression, are evaluated to assess predictive performance. Experimental results indicate that the random forest model achieves superior accuracy due to its ability to model nonlinear relationships among environmental parameters and handle sensor variability. The trained model enables short-term pollution forecasting and supports real-time alert generation when pollutant concentrations exceed predefined safety thresholds.Field experiments conducted in urban and industrial environments demonstrate that the proposed UAV-based system effectively captures spatial and vertical pollution variations that are often overlooked by conventional monitoring stations. The system provides higher spatial resolution, rapid deployment capability, and improved adaptability compared to static air quality monitoring infrastructure. Visualization of processed data through pollution heat maps and dashboards facilitates intuitive interpretation and decision-making for environmental authorities and policymakers.The proposed drone-based pollution detection framework offers a cost-effective, scalable, and flexible solution for modern air quality monitoring applications. Potential use cases include smart city environmental management, industrial emission surveillance, disaster response during hazardous gas leaks, and environmental impact assessment. The integration of UAV technology, IoT architecture, and machine learning analytics represents a significant step toward intelligent and responsive environmental monitoring systems. Future work will focus on multi-drone coordination, adaptive flight path optimization, and the integration of deep learning models to further enhance prediction accuracy and system autonomy.