Traffic Emission Estimation and Modelling at Urban Intersections Using High-Resolution UAV-Based Video

Swiftly assessing emissions at urban traffic intersections without installed monitoring devices poses a challenge. Traditional methods of collecting vehicle emission data are constrained by costs or limitations in traffic road infrastructure, lacking a cost-effective approach. To address this challenge, this paper proposes a method for estimating vehicle emissions at intersections based on UAV videos by taking advantage of the fact that UAVs are easy to deploy, low-cost, and unaffected by ground traffic conditions. This method enables the quick estimation of greenhouse gas and air pollutant emissions, such as CO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> , NO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</inf> , and PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">10</inf> emissions at intersections. Initially, the YOLO-SD algorithm is employed to extract accurate vehicle trajectories from UAV videos. Subsequently, this data is input into the VERSIT+ micro emission model to estimate vehicle emissions. Subsequently, based on the actual traffic condition at the intersection, a reliable microscopic traffic simulation model is developed using VISSIM to estimate vehicle emissions. Comparative analysis of the drone-based intersection emission estimation results and the high-fidelity intersection model results reveal a relative error of 9% for CO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> emissions, indicating good consistency of the models. However, noticeable differences exist in NO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</inf> and PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">10</inf> emissions. Through experimental result comparisons, this study validates the feasibility and accuracy of the proposed method based on drone video for estimating vehicular emissions at intersections, particularly in estimating CO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> at intersections.