Reproduce “Adaptive Safety Filtering for Frozen ACC Policies via Conformal Residual Calibration” 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

Zhiruo Zhou, Rigaudiere Z. Li, Chen Xiwen, Yucheng Chen, Xiaojun Zhu, Houde Liu

Adaptive RACF filters actions from a frozen adaptive-cruise-control policy using calibrated one-step prediction residuals to set a safety margin across candidate vehicle models. In 2,400 simulated controller–trial units it achieved 94.3% episode safety versus 74.4% for nominal CBF-QP, while projecting less often. However, robust CBF-QP reached 99.0% safety; a controlled ablation supports a margin benefit in this ACC setting.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.