PhysicalAI-Autonomous-Vehicles
Large multi-sensor dataset with 1700 hours of autonomous driving data.
Open dataset Updated 2026-06-18
What is PhysicalAI-Autonomous-Vehicles?
This dataset consists of multi-sensor recordings from vehicle operation, totaling 1700 hours, with both collection and labeling handled automatically by sensors.
It supports AV researchers developing Physical AI end-to-end driving systems and is usable for commercial or non-commercial autonomous vehicle projects per the license.
What you can build with PhysicalAI-Autonomous-Vehicles
Train multi-modal perception models
Use the 1700 hours of synchronized sensor streams to train object detection, lane segmentation, and traffic-sign recognition networks for autonomous driving.
Develop sensor-fusion pipelines
Combine data from multiple onboard sensors to build and evaluate fusion algorithms that improve robustness in varied weather and lighting.
Benchmark AV algorithms at scale
Run large-scale evaluation of planning and control modules on geographically diverse recordings with automatically generated ground truth.
Load PhysicalAI-Autonomous-Vehicles
Python
from datasets import load_dataset
ds = load_dataset("nvidia/PhysicalAI-Autonomous-Vehicles")- 1pip install datasets
- 2from datasets import load_dataset
- 3dataset = load_dataset('nvidia/PhysicalAI-Autonomous-Vehicles')
- 4Inspect dataset features and load individual sensor recordings
- 5Apply custom preprocessing before feeding data into training loops
PhysicalAI-Autonomous-Vehicles: pros & cons
Pros
- +1700 hours of driving data
- +Multi-sensor recordings with automatic labels
- +Geographically diverse coverage
- +Permits both commercial and non-commercial AV use
Cons
- –Automatic labeling may contain sensor-induced errors
- –No public details on exact sensor suite or data formats
- –Large volume requires substantial storage and compute
Did you find this helpful?
Frequently asked questions
It is a collection of 1700 hours of multi-sensor driving recordings with labels generated automatically via sensors, released by NVIDIA for autonomous-vehicle research.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar datasets
Other ai & machine learning options worth comparing.