发表机构
Shanghai University; Xi’an Jiaotong-Liverpool University（上海大学; 西交利物浦大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对自动驾驶中异常检测问题，提出LARAD方法，通过空间逻辑推理，引入SLVS管道生成训练样本，并用轻量级注意力分支增强分割网络，显著提升对逻辑异常的鲁棒性，保持单模型架构高效率并达新的技术水平。
AI 中文摘要
准确的开放世界障碍物检测对自动驾驶至关重要。当前的异常分割方法存在一个基本盲点：过度依赖纹理新颖性来识别分布外（OoD）物体，而忽略上下文空间逻辑。此外，减轻由此产生的误报通常需要级联大量视觉模型，导致不可接受的推理延迟。为解决这些问题，我们提出布局感知道路异常检测（LARAD），将范式从外观匹配转变为空间逻辑推理。首先，我们引入空间逻辑违反合成（SLVS）管道，生成纹理一致但空间无效的训练样本，迫使模型学习上下文违反情况。其次，我们用轻量级的OoD引导注意力分支增强标准的封闭集分割网络。大量实验表明，LARAD显著增强了对逻辑异常的鲁棒性，并建立了新的技术水平，同时保持单模型架构的高效率。
英文摘要
Accurate open-world obstacle detection is critical for autonomous driving. Current anomaly segmentation methods suffer from a fundamental blind spot: they over-rely on texture novelty to identify out-of-distribution (OoD) objects while ignoring contextual spatial logic. Furthermore, mitigating the resulting false positives often requires cascading massive vision models, introducing unacceptable inference latency. To address these issues, we propose Layout-Aware Road Anomaly Detection (LARAD), shifting the paradigm from appearance matching to spatial-logic reasoning. First, we introduce the Spatial-Logic Violation Synthesis (SLVS) pipeline, which generates training samples that are texture-consistent yet spatially invalid, forcing the model to learn contextual violations. Second, we augment a standard closed-set segmentation network with a lightweight, OoD-guided attention branch. Extensive experiments demonstrate that LARAD significantly enhances robustness against logical anomalies and establishes a new state-of-the-art, all while retaining the high efficiency of a single-model architecture.
