Other
BridgeData V2
Scalable language/goal-conditioned robot manipulation learning
A WidowX-arm collection contains 60096 expert and scripted trajectories across 24 tabletop and toy-kitchen environments.
Access links
Primary places to inspect, download, read about, or reproduce the dataset.
Quick facts
- Vehicle type
- Other
- Environment
- Indoor
- Data origin
- Real-world
- Sensor count
- 5
- Ground truth
- Not reported
- Calibration
- Not reported
- Annotations
- Available
Sensor overview
High-level sensor availability before the detailed sensor records below.
Ground truth
Reference scope, method, rate, coverage, and provenance.
Not reported
Calibration and synchronization
Reported calibration level, reproducibility signals, and supporting notes.
- Level
- Not reported
- Processed parameters
- Not reported
- Raw calibration data
- Not reported
- Calibration targets
- Not reported
Calibration data
Calibration reproducibility details are not reported.
Notes
Not reportedKnown or intentional limitations
Reported constraints and characteristics to check before using the dataset.
The 60,096 trajectories comprise 50,365 teleoperated and 9,731 scripted examples across 24 environments and thirteen skills; trajectories, environments and skills are different units.
Most trajectories contain only the primary fixed view. Two randomized views, wrist imagery and depth have different subset coverage; depth comes from the same primary RGB-D system.
Five hertz is the robot control frequency, not established native camera sampling. Calibration completeness, achieved synchronization and independent geometric reference accuracy remain unknown.
Raw JPEG/PNG/pickle archives and downsampled converted RLDS/TFDS data have different formats/resolutions. Manipulation and language labels do not establish navigation pose ground truth.
Cameras
Camera 1
- Model
- Intel RealSense D435
- Setup
- Rgbd
- Count
- 1
- Effective cameras
- 2
- Modality
- Rgbd
- Resolution
- 640 x 480
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Primary fixed over-the-shoulder view. Depth is available only for some trajectories; 5 Hz is stated as control frequency, not explicitly camera sampling.
Camera 2
- Model
- Logitech C920
- Setup
- Multi Camera
- Count
- 2
- Effective cameras
- 2
- Modality
- Rgb
- Resolution
- 640 x 480
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Two alternative views with randomized poses; subset availability only.
Camera 3
- Model
- Raspberry Pi camera module
- Setup
- Mono
- Count
- 1
- Effective cameras
- 1
- Modality
- Rgb
- Resolution
- 640 x 480
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Wrist/gripper-mounted view; exact module model and subset rates are unspecified.
IMUs
Not reported
GNSS
Not reported
LiDAR
Not reported
Radar
Not reported
Additional sensors
Not reported
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- Crowdsourced natural-language task descriptions per trajectory
- Class count
- Not reported
Semantic segmentation
Not reported
Instance segmentation
Not reported
Object detection
Not reported
Optical flow
Not reported
Depth ground truth
Not reported
Citation
Citation key and BibTeX kept at the end of the page for reference.
@misc{walke2023bridgedatav2,
author = {{Homer Walke} and {Kevin Black} and {Abraham Lee} and {Moo Jin Kim} and {Max Du} and {Chongyi Zheng} and {Tony Zhao} and {Philippe Hansen-Estruch} and {Quan Vuong} and {Andre He} and {Vivek Myers} and {Kuan Fang} and {Chelsea Finn} and {Sergey Levine}},
title = {BridgeData V2: A Dataset for Robot Learning at Scale},
year = {2023},
eprint = {2308.12952},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2308.12952v3}
}Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy