Reproduce “Jev with Foresight: Safe Language-Guided Autonomous Driving via a Language World Model” 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
Yansong Qu, Jiancong Chen, Samuel Labi
Jev-AD combines typed decisions with a language world model that forecasts nearby traffic for candidate actions, converts forecasts into concise outcomes, and lets the decision model choose. In simulated merges, collisions fell from 26.0% to 2.4%, while constant-velocity foresight reached 2.8%. These results concern simulation and traffic replay, not public-road safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.