Multimodal dataset for indoor 3D drone tracking.

Area of Science:

• Robotics and Computer Vision
• Sensor Fusion and Data Acquisition

Background:

• Accurate 3D drone localization is critical for autonomous navigation and aerial robotics.
• Existing datasets often lack the multimodal richness required for robust 3D tracking algorithm development.

Purpose of the Study:

• To introduce the Drone Pose and Jittering with Accurate Jittering Tracking (DPJAIT) multimodal dataset.
• To provide a comprehensive resource for developing and evaluating 3D drone tracking algorithms.
• To demonstrate novel applications of the dataset in drone pose estimation.

Main Methods:

• Dataset creation using both simulation and real-world measurements (Vicon system).
• Integration of synchronized multi-camera video sequences and reference 3D drone positions.
• Inclusion of ArUco markers with known 3D positions and drone-mounted RGB camera data.

Main Results:

• The DPJAIT dataset offers dual variants: simulation-based and real-measurement data.
• Demonstration of three 3D tracking applications: camera projection, particle swarm optimization, and ArUco marker-based extrinsic matrix determination.
• Validation of drone tracking accuracy using gold-standard motion capture techniques.

Conclusions:

• The DPJAIT dataset is a valuable resource for advancing research in drone 3D tracking and localization.
• The demonstrated applications highlight the dataset's utility for algorithm development and performance evaluation.
• This work facilitates the creation of more robust and accurate autonomous drone systems.