DriveWorld3D Evidence Constrained 3D World Modeling and Generative Simulation for Real-World Autonomous Driving

DriveWorld3D Evidence Constrained 3D World Modeling and Generative Simulation for Real-World Autonomous Driving

|Authors: Wei Wang (Robostreet research team Cambridge, MA, USA), Yiru Veronika Wang (Robostreet research team Cambridge, MA, USA), Yihan Wang (Robostreet research team Cambridge, MA, USA), Zihao Li (Robostreet research team Cambridge, MA, USA), Xiaohui Liang (Robostreet research team Cambridge, MA, USA)

|Venue: arXiv preprint | Year: 2026

|arXiv: 2610.22450

Abstract

Three dimensional modeling for autonomous driving must continuously estimate road geometry, obstacle occupancy, traffic participant behavior, and observational uncertainty under limited computational resources. Existing reconstruction methods can synthesize realistic novel views, but visual fidelity does not sufficiently characterize the geometric reliability required for driving. Generative world models can expand scenarios and predict future observations, yet may introduce content unsupported by measurements. We propose DriveWorld3D, a framework combining temporal video learning, measurement constraints from multiple sensors, and offline generative simulation. Its online component maintains a unified world state comprising road surfaces, volumetric occupancy, dynamic entities, and road relations, while separately representing physical occupancy, observability, and evidence provenance. Action conditioned future prediction and risk sensitive planning connect this representation to driving decisions. Its offline component reconstructs recorded scenes using neural radiance fields or 3D Gaussian Splatting, then expands long tail conditions using generative models while restricting the use of generated content as geometric ground truth. We specify data partitioning, sensor ablations, fault injection, paired closed loop testing, and real to simulation consistency assessment. The central hypothesis is that explicit evidence management can reduce dangerous false free space predictions and occlusion risk within a matched computational budget while retaining acceptable driving efficiency. All performance claims remain subject to experimental validation. Keywords: autonomous driving; 3D occupancy; sensor fusion; uncertainty; generative world models; NeRF; 3D Gaussian Splatting; closed loop simulation

Key Contributions

• Three dimensional modeling for autonomous driving must continuously estimate road geometry, obsta…
• Existing reconstruction methods can synthesize realistic novel views, but visual fidelity does no…
• Generative world models can expand scenarios and predict future observations, yet may introduce c…
• We propose DriveWorld3D, a framework combining temporal video learning, measurement constraints f…
• Its online component maintains a unified world state comprising road surfaces, volumetric occupan…

Topics

• vision
• reinforcement-learning
• planning

Code & Data

No code repository linked in paper metadata.

BibTeX

@article{Cambridge2026_261022450,
  title     = {DriveWorld3D Evidence Constrained 3D World Modeling and Generative Simulation for Real-World Autonomous Driving},
  author    = {Wei Wang (Robostreet research team Cambridge and MA and USA) and Yiru Veronika Wang (Robostreet research team Cambridge and MA and USA) and Yihan Wang (Robostreet research team Cambridge and MA and USA) and Zihao Li (Robostreet research team Cambridge and MA and USA) and Xiaohui Liang (Robostreet research team Cambridge and MA and USA)},
  year      = {2026},
  eprint    = {2610.22450},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  url       = {https://arxiv.org/abs/2610.22450}
}