The authors represent LiDAR sweeps as structured range images and compress them with a fully attention-based tokenizer.
One LiDAR model generates, forecasts and resimulates driving scenes in real time
LidarVerse uses a shared latent representation and conditioning interface to perform three simulation tasks across multiple LiDAR configurations.
Big Tech
Henry Che · Xuanchi Ren · Sanja Fidler · Shenlong Wang · Laura Leal-Taixé · Shengyu Huang
NVIDIA · University of Illinois, Urbana-Champaign · University of Toronto
Research Digest··3 min read
Che and colleagues present LidarVerse, a conditional world model that generates new LiDAR scenes, predicts future point clouds and resimulates recorded environments from new sensor trajectories.
Why this paper
From NVIDIA and 2 others
In one line
A single conditional LiDAR world model unifies generation, forecasting, and resimulation, real-time at 11 FPS on one H100 GPU.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§