VLA-Retain: Benchmarking Cross-Dimension Effects of Vision Tuning in VLAs

📄 arXiv: 2610.27698v1 📥 PDF

作者: Borui Xu, Han Chen, Bingsheng He, Ming Wu, Jason Zeng, Michael Heinrich

分类: cs.CV, cs.AI

发布日期: 2026-10-20

💡 一句话要点

提出VLA-Retain以评估视觉调优在VLA中的跨维度影响

🎯 匹配领域: 支柱一：机器人控制 (Robot Control) 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 视觉-语言-动作 模型微调 鲁棒性评估 跨维度保留 机器人控制 智能制造 调优策略

📋 核心要点

  1. 现有VLA模型在微调后对目标条件以外的鲁棒性评估不足，导致适应性效果不明确。
  2. 本文提出VLA-Retain基准，通过比较不同模型在多维扰动下的表现，评估视觉调优的跨维度影响。
  3. 研究结果显示，目标调优的增益与保留率无关，且不同模型的调优策略表现出显著差异。

📝 摘要（中文）

视觉-语言-动作（VLA）模型将大规模预训练知识转移到机器人控制中，但其操作性能对观察条件的变化敏感。近期研究通过在目标领域数据上微调视觉模块来解决这一问题。然而，如何评估这种适应性对目标条件以外的鲁棒性影响仍不明确。为此，本文提出了VLA-Retain基准，比较了五个开放的VLA模型在不同扰动维度下的表现。研究发现，目标调优并不局限于局部，且其目标增益无法预测保留率，表明仅依赖目标领域的成功是不够的，跨维度保留是VLA适应的重要标准。

🔬 方法详解

问题定义：本文旨在解决现有VLA模型在微调后对目标条件以外的鲁棒性评估不足的问题。现有的鲁棒性基准仅评估未微调的模型，无法反映微调对适应性的真实影响。

核心思路：提出VLA-Retain基准，通过对五个开放VLA模型进行比较，分析在不同扰动维度下的表现，以评估视觉调优的跨维度效果。

技术框架：VLA-Retain基准在LIBERO-Plus数据集上进行评估，涵盖七个扰动维度。模型在Camera-view或Sensor-noise数据上进行调优，并与其基础检查点进行比较。

关键创新：最重要的创新在于提出了跨维度保留的概念，强调目标领域成功不足以保证模型在其他条件下的鲁棒性，揭示了调优策略的模型依赖性。

关键设计：在调优过程中，采用了DELTA风格的特征保留策略，针对不同模型的表现进行了细致的参数设置和损失函数设计，以确保评估的全面性和准确性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示，在Camera-view调优下，X-VLA在目标任务上获得69.4个百分点的增益，但在其他条件下平均损失4.6个百分点。OpenVLA-OFT在目标任务上获得39.3个百分点的增益，其他条件下仅增加1.4个百分点，表明调优策略对模型表现的影响显著。

🎯 应用场景

该研究的潜在应用领域包括机器人控制、智能制造和自动化系统等。通过优化VLA模型的适应性，能够提升机器人在复杂环境中的操作能力，具有重要的实际价值和广泛的应用前景。

📄 摘要（原文）

Vision-language-action (VLA) models transfer knowledge from large-scale pretraining into robot control, but their manipulation performance remains sensitive to changes in observation conditions, which recent work addresses by fine-tuning the visual module on target-domain data. However, it remains unclear how such adaptation affects robustness beyond the target condition, as existing robustness benchmarks only evaluate models without fine-tuning and adaptation studies only report the target gain. Measuring this cross-dimension effect requires separating target gains from off-target changes, isolating adaptation effects from architectural differences, ruling out generic forgetting, and comparing tuning choices. Therefore, we introduce VLA-Retain, a benchmark that compares five open VLA models, tuned on Camera-view or Sensor-noise data, with their own base checkpoints on LIBERO-Plus dataset spanning seven perturbation dimensions. Our study reveals four findings. First, targeted vision tuning is not local, and its target gain does not predict retention. Mean off-target gain is negative in 8 of 10 model--domain configurations. Under Camera-view tuning, X-VLA gains $69.4$ percentage points on the target but loses $4.6$ points on average elsewhere, whereas OpenVLA-OFT gains $39.3$ and $1.4$ points, respectively. Second, these regressions are not explained only by performance loss on the same tasks without perturbations. Third, broader training coverage raises overall gain over target-specific training but does not eliminate dimension-level regressions. Fourth, tuning strategies are model-dependent, as DELTA-style feature preservation raises X-VLA's mean off-target gain to $+1.2$ points but lowers OpenVLA-OFT's to $-23.1$. Overall, these findings show that target-domain success alone is insufficient and highlight cross-dimensional retention as a key criterion for targeted VLA adaptation.