Universal Manipulation Interface is a data collection and policy learning framework that allows direct skill transfer from in-the-wild human demonstrations to deployable robot policies.
| UMI | |
|---|---|
| Name | UMI |
| Full Name | Universal Manipulation Interface: In-The-Wild Robot Teaching Without In-The-Wild Robots |
| Arxiv | Link |
| Publication Date | February 2024 |
| Authors | Cheng Chi, Zhenjia Xu, Chuer Pan, Eric Cousineau, Benjamin Burchfiel, Siyuan Feng, Russ Tedrake, Shuran Song |
The UMI paper was novel for several reasons:
- It completely avoids robot teleoperation and the associated latency. This lets the robot do things like reliably tossing balls.
- It provides a low-cost, scalable way to collect lots of data in the wild