Retrieval-Augmented Large Language Model Decision-Making for Autonomous Driving Guided by Chinese Philosophical Wisdom

2026-10-02

AI Summary

This study addresses the lack of ethical value guidance in autonomous driving decision-making, where balancing safety with social norms remains challenging. To this end, we propose CPW-Drive, a framework that pioneers the translation of Confucian philosophy into a structured value-guidance system. By integrating large language models with retrieval-augmented generation (RAG), and introducing a novel Physics-aware Spatial Similarity Retrieval (PSSR) algorithm to enhance the physical relevance of historical cases, the framework optimizes closed-loop driving decisions. In multi-lane highway simulations across three traffic configurations, CPW-Drive achieves success rates of 93.0%, 86.0%, and 72.0%, significantly outperforming baseline methods while attaining the highest collision-free steps and lower lane-changing frequencies.

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.

Problem

• Autonomous Driving
• Decision-Making
• Ethical Considerations
• Large Language Models
• Value Guidance

Innovation

• Retrieval-Augmented Generation
• Chinese Philosophical Wisdom
• Physics-aware Spatial Similarity Retrieval
• Autonomous Driving Decision-Making
• Large Language Models