a laboratory of human action
Teach robots to act
like humans.
Human motion, objects, and space captured from any camera. Training data for world models and robots that work with and around people.
THE PLATFORM
Capture the act. Structure it.
Teach a machine to do it.
Zendo captures human action and turns it into training data. One pipeline, from live capture to datasets your robots and world models can learn from.
Learn more →High-fidelity RGB capture
Multi-camera capture from ordinary RGB, with one-click calibration and no markers or suits.
Footage you already have
Import existing video of people working and run the full pipeline on it, no recapture needed.
Advanced teleoperation
Full-body humanoid control and other high-DOF tasks where handheld controllers and leader arms fail.
TALK TO US
Give your robot a sense of us.
Real-time capture, training data structured from your own video, or control policies designed around people — tell us what you are building.