LIBERO-Plus：面向视觉-语言-动作模型的渐进式鲁棒性基准
LIBERO-Plus: A Progressive Robustness Benchmark for Visual-Language-Action Models
对做 VLA 或机器人操作策略的视觉研究者，这篇提供了七维度可控扰动的自动化评测协议，可直接用来暴露策略在视角、光照、纹理、传感器噪声变化下的脆弱性，并作为训练时数据增广与鲁棒性改进的对照基准。
原文依据 we propose LIBERO-Plus, a comprehensive, automatic, and fine-grained evaluation framework with controlled perturbations across seven dimensions
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视觉-语言-动作（VLA）模型在机器人操作基准上报告了超过 95% 的成功率，但这些结果可能掩盖了鲁棒性方面的根本性弱点。当前基于仿真的鲁棒性评估存在扰动覆盖范围窄、依赖人工设计、分析粒度粗等问题，无法揭示模型在何时以及如何失败。为填补这一空白，我们提出 LIBERO-Plus，一个全面、自动且细粒度的评估框架，在七个维度上施加受控扰动：物体布局、相机视角、机器人初始状态、语言指令、光照条件、背景纹理和传感器噪声。我们对十个最优模型的系统分析揭示了表面能力之下的持续脆弱性，在适度扰动下性能从 95% 降至 30% 以下。我们的发现挑战了「高基准分数等同于真实能力」这一假设，并凸显了在真实变化下评估可靠性的评测实践的必要性。
英文原文摘要 1061 字符
Visual-Language-Action (VLA) models report impressive success rates exceeding 95% on robotic manipulation benchmarks, yet these results may mask fundamental weaknesses in robustness. Current simulation-based robustness evaluations suffer from narrow perturbation coverage, manual design constraints, and coarse-grained analysis that fails to reveal when and how models fail. To address this gap, we propose LIBERO-Plus, a comprehensive, automatic, and fine-grained evaluation framework with controlled perturbations across seven dimensions: object layouts, camera viewpoints, robot initial states, language instructions, lighting conditions, background textures, and sensor noise. Our systematic analysis of ten state-of-the-art models reveals consistent brittleness beneath apparent competence, with performance dropping from 95% to below 30% under modest perturbations. Our findings challenge the assumption that high benchmark scores equate to true competency and highlight the need for evaluation practices that assess reliability under realistic variation.