OS-RFODG: Open-source ROS2 framework for outdoor UAV dataset generation
Accurate localization is critical for Unmanned Aerial Vehicles (UAVs) in applications such as military operations, search and rescue, and environmental monitoring. However, existing open-source UAV datasets often lack the synchronization and detail necessary for high-precision localization in comple...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025026829 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Accurate localization is critical for Unmanned Aerial Vehicles (UAVs) in applications such as military operations, search and rescue, and environmental monitoring. However, existing open-source UAV datasets often lack the synchronization and detail necessary for high-precision localization in complex outdoor environments. To address this, the Open Source ROS2 Framework for Outdoor UAV Dataset Generation (OS-RFODG) is proposed as a cost-effective tool for generating multi-sensor UAV localization datasets. These datasets support the benchmarking of various localization methods, including learning-based models and sensor fusion techniques. OS-RFODG integrates Robot Operating System 2 (ROS2) with the PX4 autopilot, Gazebo simulator, and QGroundControl (QGC) to collect synchronized data from Light Detection and Ranging (LiDAR), Global Positioning System (GPS), Inertial Measurement Unit (IMU), camera, and barometer sensors. Geospatial data from Quantum GIS (QGIS) and Blender is used to create detailed 3D digital maps, improving terrain realism and spatial accuracy. Validation was conducted in the Makkah region of Saudi Arabia, featuring six UAV flights across diverse terrains, with distances ranging from 3.23 km to 14.70 km. In the first evaluation, GPS tracks were exported as Keyhole Markup Language (KML) files and overlaid onto Google Earth imagery, showing strong alignment with real-world terrain. In the second evaluation, the framework achieved RMSE values between 13.59 m and 19.50 m in trajectory alignment relative to a digital 3D GeoTIFF map. Additionally, image-based localization using SIFT keypoint matching reached precision scores up to 0.8964 and spatial RMSEs below 1.34 m. OS-RFODG open-source architecture ensures reproducibility and easy integration into UAV research workflows. |
|---|---|
| ISSN: | 2590-1230 |