| title | RoboHarness: A Simple Harness Outperforms VLA and World Action Models |
|---|---|
| authors | Bince Qu, Wei Chen, Bo Zhang |
| date | 2026-10-16 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.14018 |
| affiliation | Zhejiang University |
| upvotes | 19 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| primer | Vision-language-action models |
| abstract | The rise of large language models (LLMs) has opened a path toward generalist embodied AI, raising the question of how LLMs can act in the physical world. One straightforward approach is training the model to generate actions: collect robot-action data and video demonstration, refine LLM architecture with an action head, and train a vision-language-action (VLA) model at scale. However, we argue that coding agents have the potential to realize generalist embodied manipulation. All they need is a simple yet effective robot harness. We present RoboHarness: a robotic harness that gives LLM agents a visual-geometric control panel so that they can directly understand and invoke embodied tasks. With RoboHarness, Qwen3.8-Flash-Next outperforms state-of-the-art VLA models on BEHAVIOR Challenge 2025 tasks. The implementation is open source. |
cite
@misc{qu2026roboharness,
title={RoboHarness: A Simple Harness Outperforms VLA and World Action Models},
author={Bince Qu and Wei Chen and Bo Zhang},
year={2026},
eprint={2610.14018},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.14018}
}