RoboArena2026MIT
RoboArena Dataset Snapshot — 2026-07-17
A collection of autonomous policy rollouts, task-success scores, and policy-preference annotations from the RoboArena benchmark evaluation platform, containing 3,883 evaluation sessions with 10,783 policy episodes and associated video and proprioception data.
Downloads6K
Episodes10783
Likes1
Why This Matters for Physical AI
RoboArena provides large-scale distributed evaluation data for generalist robot policies across diverse embodiments and tasks, enabling systematic benchmarking and comparison of policy performance in real-world robotic settings.
Technical Profile
- Modalities
- rgbproprioceptionaction
- Episodes
- 10783
- Data Format
- NPZ
- Annotation Types
- language_instructionspreference_labelsreward_labelslong_form_feedback
- License
- MIT
Community Signals
Top 10% by downloads
Access
Need custom rgb data?
Claru builds purpose-built datasets for any environment applications with dense human annotations and quality assurance.
Request a Sample Pack