AirSim360 & Omni360-X Datasets

AirSim360 is a high-fidelity simulation platform optimized for panoramic vision, closed-loop reconstruction, and autonomous system training. This Hugging Face repository hosts AirSim360 Air/Pro beta builds and the Omni360-X synthetic data series (multi-hundred GB total across folders). For source code and development updates, see the AirSim360 GitHub repository.

---

📂 Repository Structure

Top-level layout:

Path	Role
AirSim360_Air_Beta/	Air edition scene packages (ready-to-use simulation environments).
AirSim360_Pro_Beta/	Pro edition scene packages (research/developer builds with extended assets).
Omni360-Scene/	Large-scale panoramic scene datasets, organized by collection.
Omni360-WayPoint/	Waypoint / trajectory data for navigation-oriented experiments.

1. AirSim360_Air_Beta (Air)

Standalone environment archives for immediate use. The following five scenes are available:

Scene	Filename	Size	Notes
Atmospheric House	AtmosphericHouse.zip	2.36 GB	Indoor lighting-focused scene.
City Downtown	CityDowntown.zip	1.39 GB	Dense urban streetscape.
Dekogon Gym	DekogonGym.zip	1.11 GB	Industrial gym interior.
Factory	Factory.zip	966 MB	Factory environment.
Spanish Courtyard	SpanishCourtyard.zip	917 MB	Outdoor courtyard.

2. AirSim360_Pro_Beta (Pro)

Pro builds ship as separate archives (naming includes the AirSim360_Pro_ prefix). Current uploads:

Scene	Filename	Size
Atmospheric House	AirSim360_Pro_AtmosphericHouse.zip	2.36 GB
City Downtown	AirSim360_Pro_CityDowntown.zip	1.39 GB
Dekogon Gym	AirSim360_Pro_DekogonGym.zip	1.11 GB
Factory (exterior + interior bundle)	AirSim360_Pro_FactoryExteriorInteriorBundle.zip	795 MB
Spanish Courtyard (Cattus)	AirSim360_Pro_SpanishCourtyard_Cattus.zip	919 MB

3. Omni360-Scene

Scene data is grouped under per-collection folders. Omni360-Scene/ currently contains CityPark/, DTW/, and NYC/ (each holds that collection’s assets on Hugging Face).

Detailed layout for Omni360-Scene/NYC/ (multimodal panoramic releases):

Asset	Filename	Approx. size	Description
Raw panoramas	nyc_Raw.zip	12.2 GB	Raw panoramic imagery for the scene.
Depth	nyc_Depth.zip	8.19 GB	Panoramic depth maps.
Instance segmentation	nyc_instance_panorama.zip	237 MB	Instance masks in panoramic layout.
Semantic segmentation	nyc_seg_panorama.zip	158 MB	Semantic masks in panoramic layout.
Class list	semantic_lists_nyc.txt	324 B	Semantic category listing for NYC.

---

🛠 Quick Start: Download Air builds

1. Open the AirSim360_Air_Beta directory in this repository.
2. Download the desired *.zip scene file(s) from the table above and extract them locally to initialize the Air simulation content.

For Pro content, download the matching archives under AirSim360_Pro_Beta using the Pro filenames listed above.

---

🗓 Release Roadmap

• April 15, 2026: Monthly map release cadence begins.
• April 17, 2026: AirSim360 Pro beta packages and Omni360-X scene drops (ongoing uploads under AirSim360_Pro_Beta, Omni360-Scene, and related folders).

---

⚖️ License & Terms of Use

This project (including software and datasets) is released under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license.

• ✅ Academic Use: You are free to use, modify, and build upon this work for non-commercial research and educational purposes.
• ❌ Commercial Use: Any form of commercial exploitation, including but not limited to using the data to train commercial models or integrating the software into proprietary products, is strictly prohibited without prior written consent from Insta360-Research.

📜 Citation

If you utilize AirSim360 or Omni360-X in your research, please cite:

@misc{AirSim360_2026,
  author = {Zhang Yuhang, Insta360 Research},
  title = {AirSim360: A High-Fidelity Panoramic Simulation Platform},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/Insta360-Research/AirSim360}}
}