LLM-based Interactive Imitation Learning for Robotic Manipulation
Jonas Werner, Kun Chu *, Cornelius Weber, Stefan Wermter · 2025
Read LLM-based Interactive Imitation Learning for Robotic Manipulation by Jonas Werner, Kun Chu *, Cornelius Weber, Stefan Wermter (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-2504-21769.txt.
Description
Jonas Werner and co-authors replace the human teacher of a robot with a language model in a 2025 paper. Robots that learn by imitation copy human demonstrations, but pure copying breaks when the robot drifts into states the demos never covered. Interactive imitation fixes that by letting a teacher correct the learner live. The bottleneck is the teacher's time.
Their framework, LLM-iTeach, puts a large language model in the teacher's chair. The model watches the robot's attempts, reasons about them in words, and gives corrective feedback in the same loop a human would. The robot still learns from guidance, but the guidance no longer needs a person on call. Tests judge how close the model teacher comes to the gains of human feedback on manipulation tasks.
Catalog record
| CREATOR | Jonas Werner, Kun Chu *, Cornelius Weber, Stefan Wermter |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-21769