arXiv：cs.AI· Xiaojun Bi, Xiaoyuan Ma, Yiwen Sun, Tianren Huang, Chaoran Liu, Bokai Huang, Hao Yang, Baichuan Mo·· 16 小时前
CPW-Drive：以中国哲学智慧引导的检索增强大语言模型自动驾驶决策
Retrieval-Augmented Large Language Model Decision-Making for Autonomous Driving Guided by Chinese Philosophical Wisdom
AI 导读
研究者提出 CPW-Drive，一个由中国哲学价值引导的闭环 RAG 框架，用于自动驾驶决策，并以儒家思想经典文本作为知识源。该框架还包含物理感知空间相似度检索（PSSR），通过比较车辆布局与速度外推状态召回物理相关的历史案例。
正文
Abstract:Autonomous driving decision systems must balance safety, efficiency, and social norms in complex traffic interactions. Philosophical and ethical considerations have received limited attention in existing autonomous driving decision-making approaches based on numerical optimization, sequence prediction, and large language models (LLMs). We propose Chinese Philosophical Wisdom-Guided Driving (CPW-Drive), a closed-loop retrieval-augmented generation (RAG) framework that incorporates value guidance derived from Chinese philosophy into autonomous driving decision-making. Using Chinese Confucian thought as its knowledge source, CPW-Drive consolidates LLM-extracted keywords from relevant classical texts into driving-relevant value principles through manual screening and validation. It then contextualizes these principles through scenario-specific cases to form retrievable and reusable value guidance. We further propose Physics-aware Spatial Similarity Retrieval (PSSR), which compares vehicle layouts and velocity-extrapolated states to retrieve physically relevant historical cases. On Highway-env's multilane highway-driving task, CPW-Drive achieves success rates of 93.0%, 86.0%, and 72.0% across three traffic configurations. These results outperform the strongest baseline by 8.0, 22.5, and 25.0 percentage points, respectively. Across all configurations, CPW-Drive achieves the highest collision-free step count and maintains a low lane-change frequency. The results suggest that structured value guidance can improve simulated closed-loop safety and stability while introducing efficiency and latency trade-offs.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.03948 [cs.AI] |
| (or arXiv:2610.03948v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03948 arXiv-issued DOI via DataCite |
Submission history
From: Xiaoyuan Ma [view email]
[v1]
Fri, 2 Oct 2026 19:07:28 UTC (3,687 KB)
[v2]
Tue, 13 Oct 2026 04:52:45 UTC (3,687 KB)
[v3]
Wed, 14 Oct 2026 13:28:56 UTC (3,687 KB)
来源：arXiv：cs.AI · arxiv.org