CityArtery

High-resolution trajectory datasets are essential for advancing traffic flow modeling, yet existing open-source datasets suffer from limited spatiotemporal coverage, insufficient urban interaction complexity, and incomplete semantic map data. To address these gaps, we present a UAV-collected vehicle trajectory dataset comprising 14,490 complete trajectories spanning a 1-km arterial road with 2-3 signalized intersections, capturing diverse urban traffic dynamics. The dataset features two subsets: single- and multi-intersection scenarios. Using three synchronized UAVs during peak/off-peak hours over four days, we recorded dense traffic conflicts and signal interactions. Trajectories were extracted via SLAM technology and spatially aligned in the SUMO simulation platform. Structured CSV files provide vehicle kinematics (speed, acceleration), spatiotemporal coordinates (pixel/UTM), and embedded semantic data with full-day temporal granularity. This dataset supports autonomous driving algorithm validation, safety-critical scenario generation, and signal coordination studies. The publicly available repository includes trajectory data, simulation calibration parameters, and signal timing records.

The Dataset directory includes two folders:
Dataset/Intersection_trajectory: Contains single-intersection data with different time periods for three intersections.
Dataset/Processed Data: Holds partially processed data.
Additionally, the Signal Timing Records table contains phase-based signal timing records for three scenarios.