Skip to main content
Datasetroboticsv1.0

RoboNet

by Berkeley AI Research (BAIR) · open-source · Last verified 2026-03-17

RoboNet is a large-scale dataset of 15 million video frames of robot-arm manipulation demonstrations collected from multiple robot platforms across different labs and environments. It is designed to benchmark and train self-supervised visual robot learning models that generalize across different robot morphologies and workspaces without task-specific supervision.

https://www.robonet.wiki
B
BAbove Average
Adoption: BQuality: AFreshness: BCitations: B+Engagement: F

Specifications

License
CC BY 4.0
Pricing
open-source
Capabilities
visual-robot-learning, self-supervised-learning, video-prediction
Integrations
Use Cases
robot-policy-pretraining, video-prediction, multi-robot-generalization
API Available
No
Tags
robotics, video, manipulation, multi-robot, self-supervised
Added
2026-03-17
Completeness
100%

Index Score

60.3
Adoption
62
Quality
80
Freshness
65
Citations
78
Engagement
0

Put AI to work for your business

Deploy this dataset alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.

Explore the full AI ecosystem on Agents as a Service