A UAV-Based Framework for Vehicle Trajectory Extraction and Kinematic Validation in Heterogeneous Highway Work Zones

This repository contains UAV- and instrumented vehicle–based trajectory data for validation, along with representative video clips used for vehicle detection, trajectory extraction, stitching, and kinematic conditioning under heterogeneous highway work-zone conditions. The data were collected along NH-65 in Sangareddy, Telangana using a DJI Mavic 3 Classic Drone hovered at an altitude of 120 m. The uploaded materials correspond to the study titled “A UAV-Based Framework for Vehicle Trajectory Extraction and Kinematic Validation in Heterogeneous Highway Work Zones” and support the reproducibility of the proposed framework, including trajectory extraction, post hoc stitching to address ID fragmentation, and kinematic conditioning of vehicle motion. The files labeled as Pipeline extracted_L1 and similar correspond to UAV-extracted trajectories associated with synchronized instrumented vehicle experiments, while files such as V-Box_L1 represent the corresponding instrumented vehicle trajectory data collected simultaneously for validation purposes. Additionally, files such as DFS extraction_L1 (and similar) contain trajectories extracted using the DataFromSky platform from the same UAV recordings, enabling comparative evaluation with the proposed framework. The repository further includes sample trajectory outputs, before-and-after stitching examples, representative acceleration profiles, statistical analysis outputs, and validation plots associated with the proposed processing pipeline. The first validation dataset has been collected using a UAV at an altitude of 120 m on a campus midblock section, along with an instrumented vehicle equipped with a V-Box system. Validation of the proposed framework was also performed using synchronized UAV-extracted and V-Box trajectory datasets collected during multiple work-zone laps, with the matched pairs comprising L2–Lap1_Vbox, L3–Lap2_Vbox, L5–Lap5_Vbox, and L6–Lap6_Vbox. These synchronized trajectory pairs form the basis of the quantitative validation presented in the study. Additional datasets collected in this study may be made available upon reasonable request for research purposes.