arXiv：cs.AI· Bince Qu, Wei Chen, Bo Zhang·· 16 小时前
RoboHarness：简单框架让 Qwen3.8-Flash-Next 超越 VLA 与世界动作模型
RoboHarness: A Simple Harness Outperforms VLA and World Actions Models
AI 导读
RoboHarness 是一个为 LLM 智能体提供视觉-几何控制面板的机器人框架，使其能直接理解和调用具身任务。借助该框架，Qwen3.8-Flash-Next 在 BEHAVIOR Challenge 2025 任务上超越了当前最先进的 VLA 模型。该实现将开源发布。
正文
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 will be released as open source.
| Comments: | 18 pages, including references and appendices |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.14018 [cs.RO] |
| (or arXiv:2610.14018v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.14018 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Bince Qu [view email]
[v1]
Tue, 29 Sep 2026 11:36:07 UTC (6,258 KB)
来源：arXiv：cs.AI · arxiv.org