# RoboSemanticBench基准揭示VLA模型语义接地与动作预测严重脱节

> 原文链接：https://robotics.hellobike.com/papers/2026-06-01-2606-02277-0250b615
> 频道：papers | 更新日期：2026-06-01 | 标签：auto, arxiv, cs.RO
> 核心摘要：新基准RSB通过数学与常识问答抓取任务诊断VLA模型，发现其在控制抓取成功率后，语义正确目标选择率接近或低于随机水平，暴露预训练语义理解未能有效转化为具身动作决策的关键瓶颈。

# RoboSemanticBench基准揭示VLA模型语义接地与动作预测严重脱节

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[RoboSemanticBench: Diagnosing Semantic Grounding in Action Prediction for VLA Models](https://arxiv.org/abs/2606.02277)（arXiv 2606.02277）

## 核心内容

针对VLA模型微调常退化为动作模仿、评测易受视觉捷径干扰的问题，研究提出RoboSemanticBench（RSB）具身基准。该基准要求机器人根据多选数学或常识问题，从候选方块中抓取对应正确答案的物体，涵盖算术、小学数学及常识事实理解，设置四选一与十选一两种难度套件，旨在定量诊断后训练VLA模型的语义接地能力。

实验结果显示，尽管代表性VLA模型能学会抓取候选方块，但在剔除抓取失败样本后，其选中语义正确目标的概率仍接近甚至低于随机基线。这表明模型虽继承了骨干网络的语义表征能力，却未能在动作预测阶段有效利用复杂指令语义进行物理目标选择，证实了当前VLA在语义理解到具身执行之间存在显著断层。

## 关键看点

- **评测机制**：RSB采用多选问答驱动抓取范式，覆盖算术、数学理解及常识事实，设4/10选项难度梯度
- **核心发现**：控制抓取成功率后，VLA模型语义正确目标选择率仍接近或低于随机水平
- **技术瓶颈**：揭示预训练语义能力未有效迁移至动作预测，存在语义接地与执行策略断层
- **基准价值**：提供排除视觉/指令-动作捷径的纯语义接地诊断工具，量化VLA具身推理缺陷

## 工程与产业观察

RSB揭示了当前VLA模型“懂语义但不会做”的工程痛点，提示算法研发需超越单纯的动作模仿学习。未来应强化语义符号到物理动作空间的显式对齐训练，或在推理链路中引入语义验证模块，避免端到端策略在复杂指令下退化为随机试探，这对提升具身智能在非结构化环境中的可靠性具有关键指导意义。

## 英文摘要 (Abstract)

Vision-language-action (VLA) models are built on the premise that semantic understanding from pretrained language or vision-language backbones should guide robot action prediction. Yet robot fine-tuning is optimized as imitation over task-specific action distributions, and many evaluations can be solved through visual or instruction-action shortcuts. We introduce RoboSemanticBench (RSB), an embodied benchmark for diagnosing semantic grounding in action prediction: whether post-trained VLA models can use complex instruction semantics to select and manipulate the correct physical target. In each episode, a robot receives a multiple-choice math or general-knowledge question, observes candidate answer blocks, and must grasp the block corresponding to the correct answer. RSB covers controlled arithmetic, grade-school mathematical understanding, and commonsense or factual understanding under four-choice and ten-choice suites. Across representative VLA models, we find that many policies learn to grasp candidate blocks but select the semantically correct block at near-random or below-random rates after controlling for grasp success, revealing a persistent gap between backbone-level semantic competence and action prediction.

作者：Bin Yu, Yao Zhang, Haishan Liu, Shijie Lian, Yuliang Wei, Xiaopeng Lin, Zhaolong Shen, Changti Wu, Ruina Hu, Bailing Wang, Cong Huang, Kai Chen
:::

## 原始参考来源 (Sources)

- [arXiv 2606.02277：RoboSemanticBench: Diagnosing Semantic Grounding in Action Prediction for VLA Models](https://arxiv.org/abs/2606.02277)