Reproduce “Breaking survivorship bias in driving world 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
Zihao Sheng, Zilin Huang, Yuhao Luo, Zhengyang Wan, Boyue Wang, Soyoung Ahn, David A. Noyce, Sikai Chen
The NWM is a world model that predicts futures for ego trajectories. It learns from routine and crash footage, giving extra weight to regions containing other road users. In 100 accident scenes, it depicts collisions in 86% of predictions started 0.5 seconds before impact, versus 9% for Vista and 12% for GEM; tests do not establish closed-loop safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.