MiTra: A Drone-Based Trajectory Data for an All-Traffic-State Inclusive Freeway with Ramps.

Area of Science:

• Transportation Engineering
• Urban Planning
• Computer Vision

Background:

• High-quality traffic trajectory data is vital for transportation engineering and urban planning.
• Existing datasets often lack comprehensive coverage, especially for complex scenarios like ramp merging and diverging maneuvers.
• Traditional data collection methods struggle in urban environments.

Purpose of the Study:

• To present MiTra (Milan Trajectories), a novel, high-resolution traffic trajectory dataset.
• To capture detailed vehicle behavior across various traffic states, including free flow and congestion.
• To provide data specifically covering critical on-ramp and off-ramp interactions.

Main Methods:

• Utilized six drones to collect data over a 900 m section of the A50 urban freeway in Milan, Italy.
• Conducted nine flight campaigns totaling 135 minutes of data collection.
• Employed advanced techniques for stitching trajectories, ensuring data quality and completeness.

Main Results:

• Collected 124,641 vehicle trajectories, averaging 650 m in length.
• Captured detailed kinematic information including position, speed, and acceleration for each vehicle.
• Documented complex behaviors such as lane changes, with nearly half of the vehicles performing them.

Conclusions:

• The MiTra dataset provides a reliable resource for traffic modeling, autonomous driving research, and computer vision applications.
• The dataset's comprehensive nature, including raw videos and tracking logs, facilitates in-depth analysis.
• MiTra addresses the limitations of existing datasets by offering full coverage of traffic states and ramp dynamics.