DriveDreamer4D: World Models Are Effective Data Machines for 4D Driving Scene Representation

  • Guosheng Zhao1,2
  • Chaojun Ni1,3
  • Xiaofeng Wang1,2
  • Zheng Zhu1✉
  • Guan Huang1
  • Xinze Chen1
  • Boyuan Wang1,2
  • Youyi Zhang4
  • Wenjun Mei3
  • Xingang Wang2✉
  • 1 GigaAI
  • 2 Institute of Automation, Chinese Academy of Sciences
  • 3 Peking University
  • 4 Technical University of Munich

The pipeline of DriveDreamer4D

Abstract

Closed-loop simulation is essential for advancing end-to-end autonomous driving systems. Contemporary sensor simulation methods, such as NeRF and 3DGS, rely predominantly on conditions closely aligned with training data distributions, which are largely confined to forward-driving scenarios. Consequently, these methods face limitations when rendering complex maneuvers (e.g., lane change, acceleration, deceleration). Recent advancements in autonomous-driving world models have demonstrated the potential to generate diverse driving videos. However, these approaches remain constrained to 2D video generation, inherently lacking the spatiotemporal coherence required to capture intricacies of dynamic driving environments. In this paper, we introduce DriveDreamer4D, which enhances 4D driving scene representation leveraging world model priors. Specifically, we utilize the world model as a data machine to synthesize novel trajectory videos based on real-world driving data. Notably, we explicitly leverage structured conditions to control the spatial-temporal consistency of foreground and background elements, thus the generated data adheres closely to traffic constraints. To our knowledge, DriveDreamer4D is the first to utilize video generation models for improving 4D reconstruction in driving scenarios. Experimental results reveal that DriveDreamer4D significantly enhances generation quality under novel trajectory views, achieving a relative improvement in FID by 24.5%, 39.0%, and 10.5% compared to PVG, S3Gaussian, and Deformable-GS. Moreover, DriveDreamer4D markedly enhances the spatiotemporal coherence of driving agents, which is verified by a comprehensive user study and the relative increases of 20.3%, 42.0%, and 13.7% in the NTA-IoU metric.

Comparisons

Video
Single Frame
The left shows S3Gaussian, while the right shows DriveDreamer4D-S3Gaussian.
The left shows S3Gaussian, while the right shows DriveDreamer4D-S3Gaussian.

Rendering results in lane change novel trajectory

Turn icon
Comparisons of novel trajectory renderings during lane change scenarios. The left column shows PVG, S3Gaussian, and Deformable-GS, while the right column shows DriveDreamer4D-PVG, DriveDreamer4D-S3Gaussian, and DriveDreamer4D-Deformable-GS.

Rendering results in speed change Turn icon

Comparisons of novel trajectory renderings during speed change scenarios. The left column shows PVG, S3Gaussian, and Deformable-GS, while the right column shows DriveDreamer4D-PVG, DriveDreamer4D-S3Gaussian, and DriveDreamer4D-Deformable-GS.

Citation

Acknowledgements

The website template was borrowed from Michaël Gharbi and Jon Barron.