Datasets:

riotu-lab
/

os-rfodg-outdoor-uav-synthetic-dataset-taif-saudi-arabia

---
license: mit
tags:
- UAV
- Digital_elevation_map
- DEM
- Terrain_navigation
---

# UAV Trajectory Simulation Dataset for Terrain-Based Localization

## Dataset Overview

This dataset contains simulated UAV flight data generated using ROS2, Gazebo, and PX4 autopilot system. The dataset features a quadcopter performing autonomous flight trajectories over realistic terrain imported from satellite imagery and Digital Elevation Model (DEM) maps of the Taif region in Saudi Arabia.

## Dataset Files

The dataset contains:
- **7 trajectory CSV files**: `traj_1_taif_map_data.csv` through `traj_7_taif_map_data.csv`
- **1 Digital Elevation Model**: `taif_DEM.tif` (GeoTIFF format)

### File Structure
```
traj_1_taif_map_data.csv
traj_2_taif_map_data.csv
traj_3_taif_map_data.csv
traj_4_taif_map_data.csv
traj_5_taif_map_data.csv
traj_6_taif_map_data.csv
traj_7_taif_map_data.csv
taif_DEM.tif
```

## Data Characteristics

### Flight Telemetry Data (CSV Files)
Each CSV file contains synchronized flight measurements with the following columns:

| Column | Description | Unit |
|--------|-------------|------|
| `timestamp` | Unix timestamp with nanosecond precision | ns |
| `tx`, `ty`, `tz` | 3D position coordinates in local frame | m |
| `vel_x`, `vel_y`, `vel_z` | Linear velocities | m/s |
| `orientation_x`, `orientation_y`, `orientation_z`, `orientation_w` | Quaternion orientation | - |
| `angular_vel_x`, `angular_vel_y`, `angular_vel_z` | Angular velocities | rad/s |
| `linear_acc_x`, `linear_acc_y`, `linear_acc_z` | Linear accelerations | m/s² |
| `latitude`, `longitude` | GPS coordinates | degrees |
| `altitude` | GPS altitude | m |
| `lidar_range` | LiDAR distance measurement | m |
| `AMSL` | Altitude above mean sea level | m |

### Digital Elevation Model (taif_DEM.tif)
- **Format**: GeoTIFF Digital Elevation Model
- **Coverage**: Taif region terrain corresponding to all flight trajectories
- **Source**: Generated from real satellite imagery and DEM data
- **Application**: Provides terrain elevation reference for localization algorithms

### Data Sample
```
timestamp,tx,ty,tz,vel_x,vel_y,vel_z,orientation_x,orientation_y,orientation_z,orientation_w,angular_vel_x,angular_vel_y,angular_vel_z,linear_acc_x,linear_acc_y,linear_acc_z,latitude,longitude,altitude,lidar_range,AMSL
993.128,-251.516708374023,-45.2020683288574,213.485137939453,5.11164286075962,0.045598402309674,0.220133052308578,0.023194692263158,0.003760659242595,-0.99578133623088,0.088699054215541,0.003482932690531,0.020824493840337,0.000891029543709,-0.57949709892273,0.020221803337337,9.83791255950928,21.2853756,40.2180381,1111.32048684451,213.142578125,1105.18774414063
```

## Simulation Environment

- **Flight Controller**: PX4 autopilot with realistic flight dynamics
- **Simulator**: Gazebo physics simulation
- **Framework**: ROS2 for data collection and sensor integration
- **Terrain Source**: Real-world satellite imagery and elevation data from Taif region
- **Vehicle**: Simulated quadcopter with onboard sensors

## Applications

This dataset is designed for research in:
- **Inertial Navigation Systems (INS)** development and testing
- **Terrain-based UAV localization** algorithms using elevation data
- **GPS-aided navigation** system validation
- **Multi-modal sensor fusion** techniques (IMU, GPS, LiDAR)
- **Flight trajectory analysis** and planning
- **Autonomous navigation** algorithm development

## Technical Specifications

### Flight Characteristics
- **7 distinct trajectories** with varying flight patterns
- **Precise time synchronization** across all sensor modalities
- **6DOF pose ground truth** available for algorithm benchmarking
- **Multi-sensor data fusion** including IMU, GPS, and LiDAR

### Terrain Characteristics
- **Desert-like terrain** with challenging navigation conditions
- **Realistic elevation profiles** from actual geographic data
- **Elevation-based localization** opportunities using DEM data

## Limitations

- **Sensor Coverage**: Current version includes IMU, GPS, and LiDAR data only
- **Terrain Type**: Limited to desert environment characteristics
- **Single Region**: Data collected over Taif region only

## Future Dataset Versions

### Planned Additions
- **Camera Images**: Complete image sequences from onboard cameras will be uploaded in the next version

## Usage Notes

1. **Data Alignment**: All measurements in each CSV file are time-synchronized
2. **Coordinate Systems**: Position data (tx, ty, tz) is in local simulation frame
3. **Terrain Context**: Use `taif_DEM.tif` for elevation reference and terrain-based localization
4. **Multiple Trajectories**: Each CSV represents a complete autonomous flight mission
5. **Future Updates**: Camera images will be added in a future version of this dataset

## Data Quality

- **Ground Truth Accuracy**: Complete 6DOF pose information for validation
- **Realistic Physics**: PX4 autopilot provides authentic flight dynamics
- **Synchronized Sensors**: All data streams precisely time-aligned
- **Validated Terrain**: Elevation model derived from verified satellite sources

## Citation

If you use this dataset in your research, please cite:
```
@article{JARRAYA2025106613,
title = {OS-RFODG: Open-source ROS2 framework for outdoor UAV dataset generation},
journal = {Results in Engineering},
volume = {27},
pages = {106613},
year = {2025},
issn = {2590-1230},
doi = {https://doi.org/10.1016/j.rineng.2025.106613},
url = {https://www.sciencedirect.com/science/article/pii/S2590123025026829},
author = {Imen Jarraya and Mohamed Abdelkader and Khaled Gabr and Muhammad Bilal Kadri and Fatimah Alahmed and Wadii Boulila and Anis Koubaa},
keywords = {UAV localization, Dataset generation, ROS2 framework, Geospatial mapping, Sensor data synchronization},
```