A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning

Urban vehicular emissions, a major contributor to environmental degradation, demand accurate methodologies that reflect real-world driving conditions. This study presents a telemetric data-driven framework for assessing emissions of Carbon Monoxide (CO), Hydrocarbons (HCs), and Nitrogen Oxides (NOx) in real-world scenarios. By utilizing Vehicle Specific Power (VSP) calculations, Gaussian Mixture Models (GMMs), and Ensemble Isolation Forests (EIFs), the framework identifies high-risk driving behaviors and maps high-emission zones. Achieving a Silhouette Score of 0.72 for clustering and a precision of 0.88 in anomaly detection, the study provides actionable insights for policymakers to mitigate urban emissions. Spatial–temporal analysis highlights critical high-emission areas, offering strategies for urban planners to reduce environmental impacts. The findings underscore the potential of interventions such as speed regulation and driving behavior modifications in lowering emissions. Future extensions of this work will include hybrid and electric vehicles, alongside the integration of granular environmental factors like weather conditions, to enhance the framework’s accuracy and applicability. By addressing the complexities of real-world emissions, this study contributes to bridging significant knowledge gaps and advancing sustainable urban mobility solutions.