Get in touch

Teaching robot hands from human hands

dext builds a forearm band that reads muscle signals and turns them into hand motion and grip-force data, so robots can learn the skills people use every day.

Live acquisitionSurface electromyography, 8 channels, forearm

Raw surface electromyography Band-pass 20–450 Hz, 50 Hz notch Channel 1Channel 2Channel 3Channel 4Channel 5Channel 6Channel 7Channel 8−5−4−3−2−10Time (seconds) 200 µV
Illustration of the data our forearm band records.

How it works

From the forearm to the robot hand.

Nothing is worn on the hand, so people work exactly as they normally would.

  1. Wear

    A soft band on the forearm, where the muscles that move the fingers sit.

  2. Perform

    Do a task such as picking, assembling or any other dexterous work while the band records.

  3. Decode

    Software turns eight channels of muscle signals into finger motion and grip force.

  4. Train

    Robotics teams use the data to teach robot hands that same task.

Built around your robots.

We work directly with robotics companies to develop a data collection solution tailored to their robots, tasks and workflows.

Get in touch