PathFormer: A Transformer with 3D Grid Constraints for Digital Twin Robot-Arm Trajectory Generation
Ahmed Alanazi, Duy Ho, Yugyung Lee · 2025
Read PathFormer: A Transformer with 3D Grid Constraints for Digital Twin Robot-Arm Trajectory Generation by Ahmed Alanazi, Duy Ho, Yugyung Lee (2025) 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-2510-20161.txt.
Description
Ahmed Alanazi, Duy Ho, and Yugyung Lee build a transformer that plans how a robot arm should move. The model must make paths that are exact, fit the task, and stay in bounds. Their design stores each move on a grid with three parts: where, what, and when. Masked steps force each move to step to a next cell and stay in the work zone.
The model trains on more than 53,000 paths. It matches the true path step by step in most cases. Near all paths are legal by build. Sent to a real xArm Lite 6 arm with a depth camera twin, it picks up test items at high rates. On 60 spoken tasks in clutter, it ends well in most runs.
Catalog record
| CREATOR | Ahmed Alanazi, Duy Ho, Yugyung Lee |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-20161