titlePanDWM: Panoramic 4D Driving World Models
authorsJiajun Sun, Zeyu Zhang, Zeyu Ren, Guangfeng Jiang, Yang Zhao, Hao Tang
date2026-11-15
venuearXiv (preprint, not peer reviewed)
idarXiv:2611.19857
affiliationPeking University
upvotes1 (Hugging Face Daily Papers, 2026-11-19)
linksabs pdf hf code
cite
primerWorld models
relatedSpatialScout: Can VLM Agents Actively Explore and Map 3D Scenes?
DataFlow-Harness: A Grounded Code-Agent Platform for Constructing Editable LLM Data Pipelines
OmniEdu: Open Foundation Models for Learning and Teaching
DataFlex-RL: An Evaluation Platform for RLVR Data Policies
K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs
abstractDriving world models usually generate videos from six pinhole cameras with separate optical centres and small overlaps, and must learn to keep these views consistent. We present PanDWM, which stitches them into one 360^circ panoramic band. Camera boundaries then fall inside one image, so keeping the views consistent means keeping the image continuous, and each pixel still records its camera ray. Conditioned on layouts, a causal student generates the band window by window in four steps; it is distilled from a causal teacher and trained on its own rollouts. A feed-forward decoder lifts each frame along these rays into 3D Gaussians, and static Gaussians from neighbouring frames complete them into 4D scenes. On nuScenes, scored in one stitched 360^circ view, PanDWM has lower FID and FVD than the multi-camera baselines at 8-46times their speed, and keeps its 50-step teacher's FID while lowering FVD by 22-25%.

bibtex

@misc{sun2026pandwm,
  title={PanDWM: Panoramic 4D Driving World Models},
  author={Jiajun Sun and Zeyu Zhang and Zeyu Ren and Guangfeng Jiang and Yang Zhao and Hao Tang},
  year={2026},
  eprint={2611.19857},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2611.19857}
}