ForcePoseTalk to us

Hand data without the glove.

EMG hand tracking for robot learning · physical AI

ForcePose reads forearm muscle signals and outputs finger joint angles and fingertip force from a bare hand.

No glove.

Just the band.

gForce Ultra · wrist

Built for robot learning and physical AI. A gForce band reads the muscles that move your fingers, and our model turns them into finger joint angles and fingertip force.

The ForcePose SystemAvailable now

Record your own hand data.

gForce bands, cameras, data gloves, fingertip force sensors and recording software in one package. Each session records muscle signals, video, joint angles and fingertip force together. For teams training their own models.

The ForcePose ModelEarly access

Track a hand with just the band.

Record about 20 minutes with a gForce band, and we fine-tune our model on your data. After that, just wear the band to get finger joint angles and fingertip force.

Built for

Robotics

Train robots on human demonstrations. Every recording has joint angles and fingertip force, so a policy learns how hard to grip as well as where the fingers go.

Physical AI

Train models on how people handle objects. The band keeps tracking the hand when a camera can't see it.

Used by teams at

  • Stanford
  • Xi'an Jiaotong University
  • UC Berkeley
  • University of Bologna
  • Shanghai Jiao Tong University
  • The University of Hong Kong
  • Sharpa

In the research

Sharpa's team drove a 22-DoF robot hand from a gForce armband, with no cameras on the hand.

DexEMG: Towards Dexterous Teleoperation System via EMG2Pose Generalization. Sharpa and Shanghai Jiao Tong University, 2026.

Read the paper ↗

Talk to us

What should your model learn from human hands?

Buy the ForcePose System, join early access for the model, or have us collect the data for you.