FLUID: Fine-Grained Drone-Captured Traffic Trajectories at Urban Intersections

FLUID is a dataset of fine-grained trajectories with a focus on dense traffic conflicts at typical urban signalized intersections. It contains approximately 5 hours of drone-captured data covering over 20,000 traffic participants across three distinct intersection types. The dataset was authored by Yiyang Chen and last updated on 2026-05-02.

Use Cases

• Modeling vehicle-pedestrian interaction dynamics based on fine-grained trajectory data.
• Training and validating autonomous vehicle prediction algorithms based on dense conflict scenarios.
• Analyzing traffic signal efficiency and safety based on synchronized trajectory and traffic light data.
• Developing simulation environments for urban driving based on provided maps and raw videos.

Strengths

• Contains over 20,000 traffic participant trajectories.
• Captures a high conflict rate, averaging 2.8 vehicle conflicts per minute.
• Provides a complete package including trajectories, traffic signals, maps, and raw videos.
• Dataset size is 16.5 GB, indicating substantial raw and processed data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Geographic and temporal coverage of the three intersections is not specified.

Provenance

Source

Yiyang Chen

Collection Method

Captured by drones at urban signalized intersections.

Freshness

Last updated 2026-05-02 14:01:15.