Reproduce “EditWM: Event-Decomposed World Modeling with Incremental Correction for End-to-End Autonomous 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

Junjie Yang, Qingwei Zeng, Youyou Li, Zicheng Ding, Ziyi Shi, Shuqi Shen, Hongliang Lu, Hai Yang

EditWM improves future-feature prediction and planning in NAVSIM by freezing a Normal forecast and learning a gated, bounded correction from observed history, ego state, and candidate trajectories. On 12,146 navtest scenes with expert trajectories, MSE falls 5.35%, from 0.39352 to 0.37246. EPDMS rises from 90.51 to 91.05 in the same candidate-pool comparison, but these results do not establish universal candidate improvement or causation.

Summary by alphaXiv.

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