Reproduce “REARL: A Closed-loop Autonomous Driving Simulation Enhancement Framework with Real Traffic Data and Large Language Models” 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
Xiaojun Bi, Jun Jiang, Yiwen Sun, Quanyi Ou, Yizhi Ma, Ke Cheng, Mingjie Bi, Yexin Li and 1 more
The authors present REARL, a system that monitors traffic during autonomous-driving simulations and adjusts background-vehicle decisions when selected statistics diverge from real-traffic examples. It uses clustered HighD scenarios as references and an LLM to guide interventions when speed-distribution or mean-spacing discrepancies cross thresholds. In a controlled highway evaluation, REARL reduced both errors versus CRITICAL, but the evidence covers only selected measures and settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.