Low Cost
Accessible dual-arm research with SO-101 hardware.
BREE
A reproducible bimanual benchmark.
Overview
Get started
Build the frame
No assembly experience needed. Follow the video to quickly build your BREE frame.
Core advantages
Why BREE
Standardized hardware. Shared data. Reproducible robot learning.
Reproducible
One hardware standard across BREE setups.
Scalable
Collect and share data across BREE setups.
Core feature
Cross-platform transfer
Policies work out of the box on any BREE setup without retraining.
Multi-instance scaling
Data from multiple robots merges into one training set for better performance.
Authors
Suggested citation
The BREE Team. “BREE: A Reproducible and Scalable Real-World Benchmark for Low-Cost Bimanual Robot Learning.” 2026.
Community
QQ group number:—
Group number to be added