# AGT-GSPO trains embodied VLMs through verifiable interaction to achieve execution error recovery

- Source: arXiv cs.RO
- Published: 2026-10-21T04:00:00.000Z
- RoboHot: https://robohot.news/en/items/hgjcwwyi76b93rykmx6dcn132
- Original: https://arxiv.org/abs/2610.27003

## Summary

The paper proposes the ERR-Env verifiable interaction environment, the ERR-Bench benchmark, and an online reinforcement learning method, Action-Grouped Turn-GSPO (AGT-GSPO), to train the 2B embodied VLM ERR-VLM to learn replanning and recovery after execution errors.