We build robots
that learn.

Join View the code

Drag to move the arm · tap a joint to learn what it does

Scroll for more
01HOW IT LEARNS

Show it. Record it. Train it. Repeat.

1Teleop 2Record 3Train 4Eval 5Deploy
01 / 05

Teleoperate

Drive the follower by moving the leader arm by hand.

POLICY ZOO
ACTDiffusion PolicySmolVLAπ₀GR00T · soon
02THE STACK · ONE OPEN REPO
github.com/physical-hardware-intelligence/phi Apache-2.0

Everything we build, in one place.

Hardware→Data→Training→Evaluation→Deployment
Open the repo
THE MEMBER LADDER
  • OnboardRun a pretrained policy in replay.
  • OperatorCalibrate, teleoperate, record a dataset.
  • TrainerTrain ACT on your data and evaluate it.
  • ContributorAdd a task or config; close a good-first-issue.
  • ResearcherRun a new policy / edge / RL experiment and write it up.
  • MaintainerOwn a module, review pull requests.
03JOIN Φ

Come make a robot move.

Any program, any level. There's a rung for you.

Sign up to join Follow on LinkedIn
FOUNDING TEAM
Yashvardhan GuptaCo-founder
Parv PatodiaCo-founder
Sai Krishna MaligireddyCo-founder
Vineeth SakhamuruCo-founder

Faculty sponsor · Prof. Ilmi Yoon, Director of Computing Programs & co-founder of the NEURAI Lab.