How We Trained a Robot to Fold Shirts With LeRobot
A Hugging Face presenter discusses building an autonomous robot that folds T-shirts, from choosing hardware and collecting demonstrations to training and evaluating a model.
In this talk
- 0:00
A robot that handles different shirts
The team is testing an autonomous system that folds T-shirts. It can handle varied colors and sizes, and can recover a crumpled shirt by unfolding it before folding it.
- 0:32
Building the project from scratch
The project began the previous year as an effort to teach a robot a challenging task while sharing the practical steps involved. The team selected a robot and sensors, then focused on gathering training data.
- 0:47
Collecting demonstrations at scale
Because shirt folding required substantial data collection, the team hired about 10 teleoperators. They worked in pairs and took shifts, producing demonstrations of T-shirt folds.
- 1:07
Training toward end-to-end automation
The team used those demonstrations to train a model to reproduce folds and fold shirts independently. They describe this as an initial step toward automating a complex task with open-source tools.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.