Conditional Multi-Stage Failure Recovery for Embodied Agents
Youmna Farag, Svetlana Stoyanchev, Mohan Li, Simon Keizer, Rama Doddipatla · 2025
Read Conditional Multi-Stage Failure Recovery for Embodied Agents by Youmna Farag, Svetlana Stoyanchev, Mohan Li, Simon Keizer, Rama Doddipatla (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-2507-06016.txt.
Description
Youmna Farag and four co-authors give embodied agents a staged plan for recovering when tasks go wrong. Their framework appeared in twenty twenty-five. Robots and virtual agents acting in the world fail often, and a single retry rarely fits every failure.
The system works in four stages, three during the task and one after it ends. Large language models reason about what failed, using the scene as context, and propose fixes with no task-specific training. Later stages engage only if earlier ones fall short. On a household-task benchmark, the method beat a no-recovery baseline by over eleven percent. It also topped the strongest prior model by nineteen percent. The authors stress staged escalation over one-shot repair.
Catalog record
| CREATOR | Youmna Farag, Svetlana Stoyanchev, Mohan Li, Simon Keizer, Rama Doddipatla |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-06016