foxmirko/agri-gs-slam-dataset-demo

🚜 AgriGS-SLAM Dataset: Orchard Mapping Across Seasons [IEEE RA-L 2026]

⚠️ PREVIEW DEMO — This is a preview demonstration of the AgriGS-SLAM dataset. The complete dataset will be released soon. For updates and project details, visit the project website.

This dataset accompanies the AgriGS-SLAM paper. For the source code and methodology, please visit the main repository.

Citation - IEEE Robotics and Automation Letters (RA-L 2026)

@article{usuelli2026agrigsslam,
    author={Usuelli, Mirko and Rapado-Rincon, David and Kootstra, Gert and Matteucci, Matteo},
    journal={IEEE Robotics and Automation Letters}, 
    title={AgriGS-SLAM: Orchard Mapping Across Seasons via Multi-View Gaussian Splatting SLAM}, 
    year={2026},
    volume={11},
    number={6},
    pages={7102-7109},
    keywords={Central Processing Unit;Feedback;Circuits;Electronic circuits;Location awareness;Protocols;Mobile communication;Pixel;Electronic mail;Communication systems;Agricultural automation;SLAM;sensor fusion;field robots;RGB-D perception},
    doi={10.1109/LRA.2026.3685453}}

🍎 Overview

The AgriGS-SLAM Dataset provides a comprehensive benchmark for visual–LiDAR SLAM research in agricultural orchard environments. Collected using a mobile field platform equipped with multi-view cameras, LiDAR sensors, and GPS/odometry, this dataset captures the unique challenges of orchard scenes including repetitive row geometry, seasonal appearance changes, and wind-driven foliage motion.

🌳 Dataset Description

This dataset was collected in apple and pear orchards across multiple growing seasons:

• Dormancy phase (winter, minimal foliage)
• Flowering phase (spring, blooms)
• Harvesting phase (summer, fruit)

The standardized trajectory protocol enables evaluation of both training-view synthesis and novel-view geometry for robust assessment of SLAM and 3D reconstruction performance.

📊 Data Structure

├── train/                          # Training split
│   ├── groundtruth.csv             # SLAM trajectory ground truth
│   ├── groundtruth_cam_*.csv       # Per-camera pose ground truth
│   ├── groundtruth_lidar.csv       # LiDAR pose ground truth
│   ├── fixposition/
│   │   └── odometry/
│   │       ├── gps/                # GPS/RTK measurements
│   │       ├── poses/              # LiDAR-based odometry poses
│   │       └── twists/             # LiDAR-based odometry velocities
│   ├── ouster/
│   │   └── points/                 # LiDAR point clouds (3D)
│   └── zed_multi/
│       ├── cam_1/                  # Front-left camera
│       │   ├── depth/              # Ultra depth maps
│       │   ├── depth_anything/     # Zero-shot monocular depth estimates
│       │   ├── point_cloud/        # RGB-D point clouds
│       │   └── rgb/                # RGB images
│       ├── cam_2/                  # Front-center camera
│       │   └── [same as cam_1]
│       └── cam_3/                  # Front-right camera
│           └── [same as cam_1]
│
└── val/                            # Validation split (same structure as train/)

🔧 Sensor Suite

Sensor	Model	Description	Modality
Multi-View Cameras	ZED X	3× RGB cameras with synchronized stereo pairs	Visual (RGB)
Depth Sensors	ZED X Ultra	Ultra depth sensing for stereo depth estimation	Visual (Depth)
Monocular Depth	Depth Anything v2	Zero-shot monocular depth for all images	Visual (Depth)
LiDAR	Ouster OS0	32-channel mechanical LiDAR	Range (3D Points)
GPS/RTK	FixPosition	Real-time kinematic positioning	GNSS
Odometry	FixPosition	Dead reckoning fusion	Proprioceptive

📋 File Formats

Images

• Format: PNG
• Resolution: Camera-dependent (ZED X: 1920×1200 per view)
• Location: zed_multi/cam_*/rgb/

Depth Data

• Ultra Stereo Depth: 16-bit PNG depth maps (ZED X Ultra)
  • Location: zed_multi/cam_*/depth/
• Monocular Depth (Depth Anything): Floating-point estimates
  • Location: zed_multi/cam_*/depth_anything/

Point Clouds

• Format: PCD (Point Cloud Data format)
• Type: RGB-D fusion data
• Location: zed_multi/cam_*/point_cloud/ and ouster/points/

Trajectories & Poses

• Format: CSV
• Columns (typical): [timestamp, tx, ty, tz, qx, qy, qz, qw] or velocity equivalents
• Locations:
  • Ground truth: groundtruth.csv, groundtruth_cam_*.csv, groundtruth_lidar.csv
  • Odometry: fixposition/odometry/poses/, fixposition/odometry/twists/
  • LiDAR-based odometry derived from Ouster OS0

GPS/RTK Measurements

• Format: CSV
• Location: fixposition/odometry/gps/

💾 Data Statistics

• Training Sequences: Multiple trajectories in train/
• Validation Sequences: Multiple trajectories in val/
• Temporal Resolution:
  • Cameras: ~5 fps (ZED X synchronized capture)
  • LiDAR: 10 Hz (Ouster OS0 standard)
  • GPS/Odometry: 10 Hz (Fixposition VIO-RTK-GNSS)
• Seasonal Coverage: Dormancy, Flowering, Harvesting phases

📚 Related Work

For context on SLAM and 3D Gaussian Splatting in outdoor environments, see the main AgriGS-SLAM repository.

👨‍🌾 Authors

• Mirko Usuelli¹* and Matteo Matteucci¹
  Dipartimento di Bioingegneria, Elettronica e Informazione, Politecnico di Milano, 20133 Milano, Italy
  {mirko.usuelli, matteo.matteucci}@polimi.it

• David Rapado-Rincon² and Gert Kootstra²
  Agricultural Biosystems Engineering, Wageningen University & Research, 6708 PB Wageningen, The Netherlands
  {david.rapadorincon, gert.kootstra}@wur.nl

*Corresponding author

🙏 Acknowledgments

The authors thank the Fruit Research Center (FRC) in Randwijk for access to the orchards.

Mirko Usuelli's work was carried out within the Agritech National Research Center and funded by the EU Next-GenerationEU (PNRR – M4C2, Inv. 1.4 – D.D. 1032 17/06/2022, CN00000022). This work reflects only the authors' views; the EU and Commission are not responsible.

Contributions from Matteo Matteucci, Gert Kootstra, and David Rapado-Rincon were co-funded by the EU Digital Europe Programme (AgrifoodTEF, GA Nº 101100622).

📝 License

This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

You are free to:

• Share and adapt the dataset
• Use it for commercial and non-commercial purposes

Provided you:

• Give appropriate credit to the authors
• Link to the license
• Indicate if changes were made

---

🔗 Resources & Links

🌐 Follow Us

🏛️ Funding & Institutions

Agritech Center	Next Generation EU	AgrifoodTEF	Politecnico di Milano	Wageningen University

---