A Distributional Treatment of Real2Sim2Real for Object-Centric Agent Adaptation in Vision-Driven Deformable Linear Object Manipulation
Georgios Kamaras, Subramanian Ramamoorthy · 2025
Read A Distributional Treatment of Real2Sim2Real for Object-Centric Agent Adaptation in Vision-Driven Deformable Linear Object Manipulation by Georgios Kamaras, Subramanian Ramamoorthy (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-2502-18615.txt.
Description
A robot that must bend a cable or rope faces a guessing problem: no two cables flex alike. Georgios Kamaras and Subramanian Ramamoorthy build a loop that starts and ends in the real world. First, a camera watches the real object move. Statistical inference then estimates the object's physical properties as a probability distribution, not a single guess.
Those distributions seed randomized simulations. Inside them, reinforcement learning trains a manipulation policy tailored to the object family. The trained policy returns to the real robot with no further tuning. The authors test how well the inference sorts look-alike objects from vision and touch alone. They then trace how the quality of that guess shapes real-world success.
Catalog record
| CREATOR | Georgios Kamaras, Subramanian Ramamoorthy |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18615