Reproduce “Trajectory Evaluation via Rollout-Free World Model 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
Junzhe Wang, Jinqing Zhang, Qingjie Liu
RFWorld scores candidate paths by querying a shared scene memory rather than generating a separate future for each path. During training, semantic maps at current and future times supervise memory; at inference, waypoints retrieve features. On the NAVSIM test, it reaches a PDMS of 90.5 and 37.5 ms at 1,024 candidates; the comparison does not isolate the mechanism’s effect.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.