测试13个视觉语言模型对机器人操作失败的判断能力，发现性能普遍偏低且存在系统性偏差。
FailBench: How Reliable are VLMs at Judging Robot Task Success?
- 构建包含2197次操作的多源失败检测基准FailBench，覆盖真实与仿真场景。
- 最优模型仅达0.77平衡准确率，接触类任务性能跌至近随机水平（<0.60）。
- 局部裁剪输入可提升性能2.4个百分点，适合部署在资源受限的机器人系统中。
视觉语言模型（VLMs）被越来越多用于评估机器人操作结果，但现有基准缺乏跨领域泛化证据。本文提出FailBench，一个包含2,197次操作尝试的失败检测基准，涵盖14个公开来源（12个真实场景，2个仿真）。其中75%的失败自然发生，且六项真实数据来自非失败检测数据集。评估13个基于VLM的检测器，最佳模型仅达0.77平均平衡准确率。值得注意的是，专为失败检测微调的模型始终表现劣于通用VLM及其预训练基线。性能高度依赖可见视觉线索：当结果依赖物体运动时模型接近饱和，但在接触密集型装配任务中性能降至近随机水平（<0.60平衡准确率）。错误分析显示，模型在证据模糊时系统性倾向预测成功，该偏差即使增加推理努力仍持续存在。最后，我们证明输入级干预——空间定位并裁剪相关区域——可在不额外训练的情况下将最优模型性能提升2.4个百分点。
原文摘要 · Abstract (English)
Vision-Language Models (VLMs) are increasingly used to evaluate robot manipulation outcomes, but existing benchmarks offer limited evidence of cross-domain generalization. We introduce FailBench, a benchmark for robot failure detection comprising 2,197 manipulation attempts across 14 public sources (12 real-world, 2 simulated). In FailBench, 75% of failures occur naturally, and six real-world sources come from non-failure-detection datasets. Evaluating 13 VLM-based detectors, we find the best model achieves only 0.77 mean balanced accuracy. Notably, models fine-tuned for failure detection consistently underperform general-purpose VLMs and their own pretrained baselines. Performance depends heavily on required visual evidence: models approach saturation when outcomes depend on observable object motion, but degrade to near-chance (<0.60 balanced accuracy) on contact-intensive assembly tasks. Error analysis reveals a systematic bias toward predicting success under ambiguous evidence, which persists even with increased reasoning effort. Finally, we show that input-level intervention--spatially localizing and cropping outcome-relevant regions--improves the top detector by 2.4 percentage points without extra training.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。