Reproduce “Beyond Retrieval Relevance: Scene-Grounded Risk Entailment for Vision-Language 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
Jiaxin Liu, Ruilin Yu, Liang Peng, Jingkai Wang, Chengxiang Zhao, Zhenxin Zhu, Bing Wang, Guang Chen and 3 more
The authors present a driving-risk system that checks whether rules are supported by current scene facts before a vision-language model selects an action. DRKG and SWRL reasoning derive events and object-bound risk relations for the planner. On a local nuReasoning split, the system raises NPS from 62.88 to 64.18 versus KnowVal top-16; perception quality and encoded risks limit coverage.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.