---
library_name: transformers
pipeline_tag: image-text-to-text
---

# EMBGuard

EMBGuard is the first MLLM-based safety guardrail for embodied agents designed to decouple physical risk reasoning from agent policy. By evaluating a (visual observation, action) pair, EMBGuard identifies hazardous configurations and provides natural language explanations of potential risks.

- **Paper:** [EMBGuard: Constructing Hazard-Aware Guardrails for Safe Planning in Embodied Agents](https://huggingface.co/papers/2605.30924)
- **Repository:** [https://github.com/dongwxxkchoi/EMBGuard](https://github.com/dongwxxkchoi/EMBGuard)

## Model Description

MLLM-powered embodied agents deployed in real-world environments encounter physical hazards. EMBGuard addresses the lack of explicit mechanisms for identifying these hazards by reasoning about action-conditioned risks. Despite its compact size (available in 2B and 4B variants), EMBGuard achieves performance competitive with proprietary MLLMs while significantly reducing false-positive rates that can hinder real-time deployment.

The model is based on the Qwen3-VL architecture and has been fine-tuned to identify hazardous configurations and provide natural language explanations of potential risks.

## Datasets

The model was developed using the following datasets:
- **EMBHazard:** A training dataset of 15.1K action-conditioned pairs.
- **EMBGuardTest:** A benchmark of 329 manually curated real-world scenarios spanning seven physical risk categories.

## Citation

If you use EMBGuard in your research, please cite the following paper:

```bibtex
@article{choi2025embguard,
  title={EMBGuard: Constructing Hazard-Aware Guardrails for Safe Planning in Embodied Agents},
  author={Choi, Dongwook and others},
  journal={arXiv preprint arXiv:2605.30924},
  year={2025}
}
```