DOME: Taming Diffusion Model into High-Fidelity Controllable Occupancy World Model

arXiv 24.10 2024 61.5 method, application

TLDR

DOME is a diffusion-based occupancy world model for autonomous driving, enabling high-fidelity long-duration prediction and fine-grained controllability via trajectory resampling.

Reasoning

The paper introduces a novel occupancy world model using a spatial-temporal diffusion transformer, addressing limitations of prior methods in detail loss and controllability. Strengths include strong empirical results on nuScenes and a clear focus on high-fidelity generation; weaknesses are not explicitly discussed in the abstract, but the approach is limited to occupancy representation and may not generalize to other modalities.

Read-first score

Read-first score 61.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 40.

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,dataset,evaluation,experiment

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Topical relevance 42%

57.1

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=dataset,github

Field roles

Methodology anchor

Rank sensitivity

Stability: volatile; rank range: 127.

Keyword Scores

world model

generative world model

world dynamics prediction

world simulator

interactive world model

video world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Spatial-temporal diffusion transformer for high-fidelity and long-duration occupancy prediction
Trajectory resampling method for fine-grained controllability in future occupancy generation

Methodology

DOME employs a spatial-temporal diffusion transformer to predict future occupancy frames from historical occupancy observations. It introduces a trajectory resampling method to enhance controllability. The model is trained and evaluated on the nuScenes dataset.

Key Results

DOME surpasses baselines by 10.5% in mIoU and 21.2% in IoU for occupancy reconstruction, and by 36.0% in mIoU and 24.6% in IoU for 4D occupancy forecasting on nuScenes.

Resources

Paper Project