Learn the loop.
Best for first datasets, tabletop pick-and-place, and learning how teleoperation, recording, and policy evaluation fit together.
Choose one leader + one follower, a bimanual four-arm system, or bundle the bimanual system with OpenBooth for a consistent two-camera workspace and access to its public skill catalog.
SO101 leader + follower kit.
The SO-101 options are organized by arm count: a standard pair has 1 leader + 1 follower; bimanual has 2 leaders + 2 followers. The bimanual OpenBooth bundle adds the booth frame and 2 cameras.
Bring a computer for setup, teleoperation, recording, and policy work. A GPU is not required to drive or record the arms; training requirements depend on the policy you select.
Use MakerMods Lab to connect and calibrate the arms, teleoperate the follower, record datasets, train policies, and evaluate results. Hugging Face publishes the official open-source SO-101 setup documentation.
Every configuration uses the same open-source SO-101 hardware and MakerMods Lab workflow. The difference is how many arms you control and whether the task happens inside a repeatable booth.
Best for first datasets, tabletop pick-and-place, and learning how teleoperation, recording, and policy evaluation fit together.
Use two leaders and two followers for tasks that need holding, stabilizing, opening, folding, or coordinated motion.
Add a consistent background and camera layout, then work from the same kind of setup used for public OpenBooth skills.
Read the full product story, the leader/follower workflow, hardware specifications, and official documentation links.
[ EXPLORE SO-101 → ]Explore the public SO-101 recordings already cataloged on Hugging Face before choosing a booth bundle.
[ BROWSE SKILLS → ]