Total site visits
...
Total visits
...
Visitors
Displayed using a lightweight public visitor counter for static sites.
Off-road multimodal dataset
The Great Outdoors Dataset
A multimodal perception dataset for autonomous navigation in unstructured outdoor terrain, combining LiDAR, RGB imagery, thermal imaging, radar, INS, and RTK GPS with semantic annotations.
64
channel Ouster LiDAR
3
RGB camera streams
23
semantic classes
17
raw ROS bag trajectories
Developed by Texas A&M University and DEVCOM Army Research Laboratory
Visitor analytics
Public aggregate traffic signals.
Privacy-friendly counts help the community understand dataset interest without exposing precise individual visitor activity.
Aggregate visitor regions
- Country/region stats Setup pending Configure location_stats_url after enabling a public aggregate report.
Country or region statistics load only from a configured public aggregate endpoint.
Static-site friendly
The website does not store analytics data itself and does not include private API keys. Configure only public service identifiers or public aggregate endpoints.
Add a public dashboard URL in _data/analytics.yml after enabling shared analytics.
Research focus
Built for perception beyond paved roads.
The Great Outdoors Dataset supports robust autonomy research in complex off-road environments where terrain, lighting, vegetation, and surface conditions vary constantly.
Collected with an unmanned ground vehicle designed for unstructured terrain, the dataset brings together high-density 3D point clouds, high-resolution RGB imagery, long-wave infrared imaging, radar, inertial navigation, and precise geolocation. It extends the foundation of RELLIS-3D with additional terrain and object categories for outdoor autonomy.
LiDAR
Detailed point clouds for geometry, mapping, and 3D semantic segmentation.
RGB
Multiple color camera views for visual perception and sensor fusion.
LWIR
Thermal imagery for low-light and adverse visibility conditions.
Radar
2D mmWave radar data for complementary perception in challenging weather.
Dataset preview
Multimodal scenes with semantic structure.
The data is tailored for machine learning models that need to understand vegetation, trails, mud, rubble, water, barriers, vehicles, and other off-road scene elements.
Why it matters
A practical benchmark for off-road autonomy.
Urban datasets rarely capture the perception problems found in unstructured terrain. This dataset is organized around those harder outdoor conditions.
Sensor fusion
Combine LiDAR, camera, thermal, radar, INS, and RTK GPS data for multimodal perception pipelines.
Semantic understanding
Train and evaluate models across terrain, vegetation, people, objects, vehicles, structures, and void regions.
Outdoor complexity
Work with unstructured scenes that include gravel, mulch, mud, puddles, rubble, grass, trees, and water.
Citation
Cite the dataset.
Please cite the Great Outdoors Dataset paper if this data supports your research.
@misc{jiang2025gogreatoutdoorsmultimodal,
title={GO: The Great Outdoors Multimodal Dataset},
author={Peng Jiang and Kasi Viswanath and Akhil Nagariya and George Chustz and Maggie Wigness and Philip Osteen and Timothy Overbye and Christian Ellis and Long Quang and Jia Huang and Srikanth Saripalli},
year={2025},
eprint={2501.19274},
archivePrefix={arXiv},
primaryClass={cs.RO},
url={https://arxiv.org/abs/2501.19274},
}
Collaborators
Research contributors.
Texas A&M University
Peng Jiang, Kasi Viswanath, Akhil Nagariya, George Chustz, Jia Huang, Srikanth Saripalli
DEVCOM Army Research Laboratory
Maggie Wigness, Philip Osteen, Tim Overbye, Christian Ellis, Long Quang
All datasets and code on this page are copyrighted by the authors and published under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License.