CockpitHAT：面向具身多智能体座舱的依赖图驱动分层归因方法

中文摘要 AI 辅助

大型语言模型（LLM）多智能体系统存在“正确性崩溃”问题，即高任务级准确率掩盖了严重的过程级故障，这在汽车座舱等安全关键型具身场景中尤为危险，因为语义正确的语句可能触发危险的物理操作。现有归因方法仅依赖文本轨迹，缺失依赖结构、多通道证据及安全感知评估。我们提出CockpitHAT，这是一个分层归因框架，它用交互有向无环图（DAG）的依赖距离阈值替代位置窗口，通过具身适配器整合多通道证据，并在置信度加权的分析师共识过程中对高风险故障应用安全提升机制。我们还发布了CockpitBench，这是一个包含212条带注释故障轨迹的基准，涵盖对话、车辆状态、环境和记忆通道，每条轨迹通过三位专家共识标注了ISO 26262的ASIL严重程度。在公开的Who&When基准上，CockpitHAT在手工划分集上的智能体级/步骤精确准确率为77.9%/37.8%，在算法生成划分集上为86.5%/46.0%，超越仅基于文本的当前最优方法ECHO达17.6/16.7个百分点；在CockpitBench上，其智能体级准确率达78.3%，步骤精确准确率达38.2%。这些结果表明，依赖感知、多通道、风险校准的归因是现实世界具身LLM多智能体系统可靠故障诊断的有效范式。

英文摘要

LLM multi-agent systems suffer from Correctness Collapse, where high task-level accuracy conceals severe process-level failures. This is especially hazardous in safety-critical embodied settings such as automotive cockpits, where lexically correct utterances may trigger dangerous physical operations. Existing attribution methods rely on text traces alone, missing dependency structure, multi-channel evidence, and safety-aware evaluation. We introduce CockpitHAT, a hierarchical attribution framework that replaces positional windows with dependency-distance thresholds from interaction DAGs, integrates multi-channel evidence via an embodied adapter, and applies a safety-uplift to high-risk failures during confidence-weighted analyst consensus. We further release CockpitBench, a benchmark of 212 annotated failure traces spanning dialogue, vehicle-state, environmental, and memory channels, each labeled with ISO 26262 ASIL severity via three-expert consensus. On the public Who&When benchmark, CockpitHAT achieves agent-level / step-exact accuracies of 77.9% / 37.8% on the Hand-Crafted split and 86.5% / 46.0% on the Algorithm-Generated split, surpassing the text-only SOTA ECHO by up to 17.6 / 16.7 points. On CockpitBench, it attains 78.3% agent-level and 38.2% step-exact accuracy. These results establish dependency-aware, multi-channel, risk-calibrated attribution as an effective paradigm for reliable failure diagnosis in real-world embodied LLM multi-agent systems.

发表机构

• ByteDance（字节跳动）

机构由 AI 辅助整理，请以论文原文为准。