Latent Adaptive Planner for Dynamic Manipulation
Donghun Noh, Deqian Kong, Minglu Zhao, Andrew Lizarraga, Jianwen Xie, Ying Nian Wu, Dennis Hong · 2025
Read Latent Adaptive Planner for Dynamic Manipulation by Donghun Noh, Deqian Kong, Minglu Zhao, Andrew Lizarraga, Jianwen Xie, Ying Nian Wu, Dennis Hong (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-2505-03077.txt.
Description
Donghun Noh and six co-authors teach a robot to catch and handle moving objects by planning in a compact hidden space. Their system, the Latent Adaptive Planner, treats each plan as a point in that space. It keeps updating its best guess as new camera views arrive, so the robot can adjust mid-motion. The planner learns from videos of people doing the task, which raises a mismatch problem: human arms and robot arms move differently.
To bridge that gap, the team uses a model-based mapping that rebuilds matching joint and object positions from the human demonstrations. In box-catching tests with varied objects, the planner caught more boxes than rival methods. It also moved more smoothly and used less energy, picking up the soft, giving motions people use. The same human videos could train the planner for different robot bodies.
Catalog record
| CREATOR | Donghun Noh, Deqian Kong, Minglu Zhao, Andrew Lizarraga, Jianwen Xie, Ying Nian Wu, Dennis Hong |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-03077