SafePath：面向安全的基于 LLM 自主导航的保形预测
SafePath: Conformal Prediction for Safe LLM-Based Autonomous Navigation
AI总结：
本文提出 SafePath 框架，结合保形预测与候选路径筛选、风险选择及人工委托，为基于 LLM 的自动驾驶路径规划提供概率安全保证，并显著降低不确定性与碰撞率。
摘要：
大语言模型（LLM）通过对复杂交通场景进行推理以生成路径规划，在自动驾驶领域展现出日益增长的应用前景。然而，它们的过度自信倾向和幻觉引发了严重的安全担忧。我们提出 SafePath，这是一个模块化框架，利用保形预测为基于 LLM 的路径规划增强形式化安全保证。SafePath 分为三个阶段运行。第一阶段，我们使用 LLM 生成一组多样化的候选路径，根据智能体行为和环境线索探索可能的轨迹。第二阶段，SafePath 通过整合保形预测的多项选择问答形式，滤除高风险轨迹，同时保证以用户定义的概率至少包含一个安全选项。最后阶段，当不确定性较低时，我们的方法选择预期碰撞风险最低的路径；当不确定性较高时，则将控制权委托给人类。我们从理论上证明 SafePath 以用户定义的概率保证安全轨迹，并展示如何调节其人工委托率，以平衡自主性与安全性。在 nuScenes 和 Highway-env 上的大量实验表明，SafePath 将规划不确定性降低了 77%，碰撞率最高降低 70%，证明了其在使 LLM 驱动的路径规划更安全方面的有效性。
英文摘要：
Large Language Models (LLMs) show growing promise in autonomous driving by reasoning over complex traffic scenarios to generate path plans. However, their tendencies toward overconfidence, and hallucinations raise critical safety concerns. We introduce SafePath, a modular framework that augments LLM-based path planning with formal safety guarantees using conformal prediction. SafePath operates in three stages. In the first stage, we use an LLM that generates a set of diverse candidate paths, exploring possible trajectories based on agent behaviors and environmental cues. In the second stage, SafePath filters out high-risk trajectories while guaranteeing that at least one safe option is included with a user-defined probability, through a multiple-choice question-answering formulation that integrates conformal prediction. In the final stage, our approach selects the path with the lowest expected collision risk when uncertainty is low or delegates control to a human when uncertainty is high. We theoretically prove that SafePath guarantees a safe trajectory with a user-defined probability, and we show how its human delegation rate can be tuned to balance autonomy and safety. Extensive experiments on nuScenes and Highway-env show that SafePath reduces planning uncertainty by 77\% and collision rates by up to 70\%, demonstrating effectiveness in making LLM-driven path planning more safer.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
