Understanding World or Predicting Future? A Comprehensive Survey of World Models

arXiv 24.11 2024 62.5 survey

TLDR

A survey categorizing world models into understanding present and predicting future, with applications in games, driving, robotics, and social simulacra.

Reasoning

Strengths: systematic categorization, coverage of multiple real-world domains, and inclusion of code repositories. Weaknesses: survey format limits novel contributions, and depth may be sacrificed for breadth.

Read-first score

Read-first score 62.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 50.

Recency 8%

75.1

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

Topical relevance 42%

71.4

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%

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

Reproducibility 25%

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

Field roles

Candidate

Rank sensitivity

Stability: volatile; rank range: 215.

Keyword Scores

world model

world dynamics prediction

model-based reinforcement learning world model

world simulator

generative world model

interactive world model

video world model

Deep Analysis

Innovations

Systematic categorization of world models into two primary functions: constructing internal representations to understand world mechanisms and predicting future states to simulate and guide decision-making.
Comprehensive review covering applications in generative games, autonomous driving, robotics, and social simulacra, highlighting how each domain utilizes these two aspects.
Summary of representative papers along with their code repositories, providing a resource for further research.

Methodology

This paper conducts a comprehensive literature review, systematically categorizing world models based on two primary functions: understanding the present state of the world and predicting its future dynamics. It examines current progress in each category, explores applications across key domains, and outlines key challenges and future research directions.

Key Results

This is a survey paper; no experimental results are presented.

Resources

Paper