⌁ Awesome World Model Hub Papers · Datasets · Projects
中文 GitHub
← Back to papers

Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous Driving

arXiv 24.8 2024 64.1 method, application

TLDR

Drive-OccWorld adapts a vision-centric 4D occupancy forecasting world model to end-to-end planning for autonomous driving.

Reasoning

The paper presents a clear contribution: a world model that forecasts 4D occupancy and flow, integrates action conditions, and uses occupancy-based cost for planning. Strengths include comprehensive experiments on multiple real-world datasets and a novel memory module. Weaknesses: limited discussion of limitations or failure cases, and the abstract does not detail comparison to baselines or ablation studies.

Read-first score

Read-first score 64.1, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 51.

Recency 8%
75.1

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

Topical relevance 42%
72.9

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

Methodology quality 25%
60

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

Reproducibility 25%
50

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

Field roles

Candidate

Rank sensitivity

Stability: volatile; rank range: 194.

Keyword Scores

world model
10
world dynamics prediction
9
generative world model
8
interactive world model
7
world simulator
6
model-based reinforcement learning world model
6
video world model
5

Deep Analysis

Innovations

  • Adapting a vision-centric 4D forecasting world model to end-to-end planning for autonomous driving
  • Semantic and motion-conditional normalization in the memory module to accumulate historical BEV embeddings
  • Injecting flexible action conditions (velocity, steering angle, trajectory, commands) into the world model for controllable generation
  • Integrating generative capabilities of the 4D world model with end-to-end planning using an occupancy-based cost function for optimal trajectory selection

Methodology

Drive-OccWorld proposes a vision-centric 4D forecasting world model for end-to-end planning. It employs a semantic and motion-conditional normalization in the memory module to accumulate historical BEV embeddings, which are then fed into a world decoder for future occupancy and flow forecasting with geometry and spatiotemporal modeling. Flexible action conditions are injected to enable controllable generation, and an occupancy-based cost function is used to select optimal trajectories from the forecasted states.

Key Results

Comprehensive experiments on nuScenes, nuScenes-Occupancy, and Lyft-Level5 datasets demonstrate that the method generates plausible and controllable 4D occupancy, advancing driving world generation and end-to-end planning.

Resources

Paper

Tags