Dataset · UCLA-VAIL
Ready-to-use urban 3D scenes for the Isaac Sim simulator
UCLA-VAIL/UrbanVerse-Training-Scenes
Over 100,000 urban scenes in OpenUSD format with textures for robot training.
What's inside
A collection of 100,000 to 1 million simulation-ready urban 3D scenes in OpenUSD format for NVIDIA Isaac Sim and Isaac Lab platforms, released by the VAIL-UCLA laboratory (ICLR 2026). Each scene is a standalone USD file with all materials and textures, generated by the UrbanVerse-Gen pipeline based on urban walking video recordings.
What you can do with it
- Training autonomous robots and navigation agents in urban environments
- Generating synthetic visual data for computer vision in the simulator
- Testing perception algorithms and route planning in Isaac Lab
- Tasks
- Robotics
- Size
- 100K–1M rows
- Downloads this month
- 20.5K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("UCLA-VAIL/UrbanVerse-Training-Scenes")