Reproduce “PKRD-CoT: A Unified Chain-of-thought Prompting for Multi-Modal Large Language Models in Autonomous Driving” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xuewen Luo, Fan Ding, Yinsheng Song, Xiaofeng Zhang, Junnyong Loo
PKRD-CoT prompts pretrained multimodal language models to interpret driving scenes and choose actions without additional driving-specific training. It uses observation, target identification and prediction, structured JSON memory, then a decision based on current and stored information. On 100 real-scene samples, GPT-4.0 reached 94% decision accuracy, versus 88% with role-playing and 72% with zero-shot prompts; the comparison is between prompt packages.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.