DexSynRefine: Synthesizing and Refining Human-Object Interaction Motion for Physically Feasible Dexterous Robot Actions
Hyesung Lee, Hyunwoo Jung, Si-Hwan Heo, Sungwook Yang · 2026
Read DexSynRefine: Synthesizing and Refining Human-Object Interaction Motion for Physically Feasible Dexterous Robot Actions by Hyesung Lee, Hyunwoo Jung, Si-Hwan Heo, Sungwook Yang (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-2605-05925.txt.
Description
Hyesung Lee and three co-authors address a hard problem in robot learning: teaching dexterous hands from human motion data. Human object interaction records are sparse and purely kinematic, so copying them onto a robot often fails. Body mismatch and rich contact forces make direct retargeting unreliable in the real world.
Their framework, DexSynRefine, treats human motion as a guide rather than as ready robot action. It first builds hand and object paths from a learned motion prior, then grounds them with reinforcement learning in task space. A final step reads contact clues from the robot's own sensor history. Across five tasks, real world success rose by 50 to 70 percentage points over kinematic copying.
Catalog record
| CREATOR | Hyesung Lee, Hyunwoo Jung, Si-Hwan Heo, Sungwook Yang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05925