titleNVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle Simulation
authorsNVIDIA, Aarti Basant, Amlan Kar, Despoina Paschalidou, Fangyin Wei, Francesco Ferroni, Guillermo Garcia Cobo, Haithem Turki, Huan Ling, Jaewoo Seo, James Lucas, Jay Zhangjie Wu, Jialiang Wang, Jonathan Lorraine, Jun Gao, Kai He, Katarina Tothova, Kevin Xie, Michał Tyszkiewicz, Qi Wu, Riccardo de Lutio, Ruilong Li, Sanja Fidler, Seung Wook Kim, Tianchang Shen, Tianshi Cao, Tobias Pfaff, William Lew, Xindi Wu, Xuanchi Ren, Yifan Lu, Yuxuan Zhang, Zan Gojcic, Zian Wang
date2026-06-02
venuearXiv (preprint, not peer reviewed)
idarXiv:2606.03159
affiliationNVIDIA
upvotes21 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf code
primerWorld models
abstractAs autonomous vehicle capabilities advance, the safe evaluation of driving policies in long-tail scenarios remains a critical bottleneck. In closed-loop simulation, the driving policy model actively interacts with the environment, where its actions dynamically update the simulator state and directly influence the next set of generated sensor observations. While recent reconstruction-based neural simulators offer photorealism, they are fundamentally constrained by their initial captured data and struggle to generalize to highly dynamic or novel scenes. To overcome these limitations, we introduce OmniDreams, a foundation generative world model mid- and post-trained from the Cosmos diffusion model to autoregressively generate action-conditioned videos in real time. By leveraging the rich visual priors of Cosmos and mid- and post-training on 21k hours of driving scenarios, OmniDreams synthesizes complex, unobserved phenomena that are hard for traditional simulators to capture, such as extreme weather and unpredictable dynamic agent behaviors. Crucially, it autoregressively conditions its photorealistic sensor generation on past frames, the current simulator state, and immediate driving actions. Deployed in a closed-loop system with the Alpamayo 1 policy model and AlpaSim orchestrator, OmniDreams acts as a highly responsive, reactive environment, providing a scalable and comprehensive solution for training and evaluating next-generation autonomous driving policies. We additionally show preliminary results indicating that a world-action model (WAM) post-trained from OmniDreams achieves strong performance on the Physical AI Autonomous Vehicles NuRec dataset, surpassing the VLA-based Alpamayo 1.5 research policy model while using only 1/5 the total parameters. These results highlight the potential for a real-time world model like OmniDreams to also serve as a backbone for policy architectures.

cite

@misc{nvidia2026nvidia,
  title={NVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle Simulation},
  author={NVIDIA and Aarti Basant and Amlan Kar and Despoina Paschalidou and Fangyin Wei and Francesco Ferroni and Guillermo Garcia Cobo and Haithem Turki and Huan Ling and Jaewoo Seo and James Lucas and Jay Zhangjie Wu and Jialiang Wang and Jonathan Lorraine and Jun Gao and Kai He and Katarina Tothova and Kevin Xie and Michał Tyszkiewicz and Qi Wu and Riccardo de Lutio and Ruilong Li and Sanja Fidler and Seung Wook Kim and Tianchang Shen and Tianshi Cao and Tobias Pfaff and William Lew and Xindi Wu and Xuanchi Ren and Yifan Lu and Yuxuan Zhang and Zan Gojcic and Zian Wang},
  year={2026},
  eprint={2606.03159},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2606.03159}
}