Figure Collects Home Chores Video Data via Mobile App to Train Humanoids

Robotics startup Figure has launched a smartphone application called Index that pays individuals to record videos of routine housework. Users film everyday actions like cleaning, picking up items, and folding clothes. As detailed in Figure's product release, these footage submissions feed into a massive digital library designed to train humanoid machines how to interact smoothly with physical environments.

This strategy matters because teaching machines to navigate real homes requires exposure to diverse, messy environments that laboratories cannot replicate. If successful, crowdsourcing real world movements could accelerate the arrival of practical domestic robots that handle routine chores for homeowners. Additionally, it offers participants an easy route to earn money simply by documenting their daily routines.

At the same time, the reliance on intimate home footage brings notable privacy challenges. Figure has not fully detailed how it will protect private documents, faces, or sensitive surroundings captured in user uploads. Furthermore, it is unclear whether crowdsourced phone video offers sufficient precision to teach delicate spatial awareness, or if bad video quality will hamper robot learning over time.