A Rapid Deployment Pipeline for Autonomous Humanoid Grasping Based on Foundation Models
Yifei Yan 1∗ , Yankai Liao 2∗ , and Linqi Ye 1 · 2026
Read A Rapid Deployment Pipeline for Autonomous Humanoid Grasping Based on Foundation Models by Yifei Yan 1∗ , Yankai Liao 2∗ , and Linqi Ye 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-17258.txt.
Description
Yan, Liao, and Ye present a pipeline that cuts onboarding for a new object from one to two days to about 30 minutes. Roboflow annotation trains a YOLOv8 detector, SAM 3D rebuilds the object without a scanner, and FoundationPose tracking uses the generated mesh as its template. The pose drives a Unity planner whose joint commands stream via UDP to a Unitree G1 humanoid. Tests show detection mAP of 0.995, pose precision under about 1 mm, and successful grasps at five positions, with generality confirmed on an automobile-window glue task.
Catalog record
| CREATOR | Yifei Yan 1∗ , Yankai Liao 2∗ , and Linqi Ye 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20404 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.17258