AI

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

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 06/10/2026, 04:18:24 UTC.

Briefings

This research introduces CPW-Drive, a Retrieval-Augmented Large Language Model framework integrating Chinese Confucian philosophy into autonomous driving. The system uses LLMs to extract keywords from classical texts, which are then filtered by manual screening and contextualized into driving value principles. A new Physics-aware Spatial Similarity Retrieval (PSSR) module compares vehicle layouts and velocity states to retrieve relevant historical cases. On the Highway-env dataset, CPW-Drive achieved 93.0%, 86.0%, and 72.0% success rates across three traffic configurations, outperforming the strongest baseline by 8.0, 22.5, and 25.0 percentage points respectively. The method prioritizes safety and collision avoidance while managing efficiency and latency trade-offs.

Key points

• CPW-Drive is a closed-loop RAG framework that uses Chinese Confucian thought to generate value guidance for autonomous driving via LLM keyword extraction and manual validation.
• Physics-aware Spatial Similarity Retrieval (PSSR) retrieves relevant historical driving cases by comparing vehicle layouts and velocity-extrapolated states to the current scenario.

Context and limitations

Results are limited to the Highway-env multilane highway-driving task; performance metrics and trade-offs are specific to this simulated environment.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 06/10/2026, 04:55:07 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeif6myiftp4gpwmgiao2fx3n2sp44ydnacuf6tjrg6bhhojmeqlmeq