Image, speech, and ADS-B trajectory datasets for terminal airspace operations.

Area of Science:

• Aerospace Engineering
• Computer Science
• Data Science

Background:

• Air traffic management relies on diverse data streams.
• Integrating AI into air traffic control (ATC) requires comprehensive datasets.
• Autonomous aircraft development necessitates advanced operational data.

Purpose of the Study:

• Introduce TartanAviation, an open-source, multi-modal dataset for terminal-area airspace operations.
• Provide a holistic view of airport environments using synchronized image, speech, and ADS-B trajectory data.
• Facilitate AI and machine learning integration into ATC systems and autonomous aviation.

Main Methods:

• Collected concurrent image, speech, and ADS-B trajectory data from inside airport boundaries.
• Gathered data across diverse airfields, seasons, aircraft types, and weather conditions over multiple months.
• Provided raw and post-processed data, including synchronized, filtered, and interpolated versions.

Main Results:

• Dataset comprises 3.1 million images, 3374 hours of Air Traffic Control speech, and 661 days of ADS-B trajectory data.
• Includes open-sourced code for data collection and pre-processing, enhancing accessibility.
• Offers diverse and synchronized data crucial for advanced aviation research.

Conclusions:

• TartanAviation is a vital resource for advancing AI in ATC and autonomous flight.
• The dataset's multi-modal nature and open-source availability accelerate research and development.
• Enables novel applications in airspace monitoring, safety, and operational efficiency.