English

Conditional Multi-Stage Failure Recovery for Embodied Agents

Computation and Language 2025-07-09 v1

Abstract

Embodied agents performing complex tasks are susceptible to execution failures, motivating the need for effective failure recovery mechanisms. In this work, we introduce a conditional multistage failure recovery framework that employs zero-shot chain prompting. The framework is structured into four error-handling stages, with three operating during task execution and one functioning as a post-execution reflection phase. Our approach utilises the reasoning capabilities of LLMs to analyse execution challenges within their environmental context and devise strategic solutions. We evaluate our method on the TfD benchmark of the TEACH dataset and achieve state-of-the-art performance, outperforming a baseline without error recovery by 11.5% and surpassing the strongest existing model by 19%.

Keywords

Cite

@article{arxiv.2507.06016,
  title  = {Conditional Multi-Stage Failure Recovery for Embodied Agents},
  author = {Youmna Farag and Svetlana Stoyanchev and Mohan Li and Simon Keizer and Rama Doddipatla},
  journal= {arXiv preprint arXiv:2507.06016},
  year   = {2025}
}

Comments

Accepted at REALM 2025

R2 v1 2026-07-01T03:51:43.112Z