This repository serves as a curated collection of research papers, workshops, and challenges focused on "World Models" for autonomous driving and robotics. It aims to track, benchmark, and provide a comprehensive overview of methods that enable AI agents to predict and simulate how the world evolves, thereby enhancing safety, reliability, and intelligence in autonomous systems.

How It Works

The collection highlights various approaches to building world models, including predictive models of physical phenomena, generative AI for autonomy, and techniques integrating perception, controllability, and future prediction. These models act as learned simulators or "what-if" engines for model-based reinforcement learning and planning, crucial for understanding complex driving scenarios and human behavior.

Quick Start & Requirements

This repository is a curated list of papers and does not have a direct installation or execution command. It serves as a reference for researchers and developers in the field.

Highlighted Details

• Comprehensive list of papers from major conferences (CVPR, ICLR, ICML, NeurIPS, ECCV, ICRA, IROS) and arXiv, spanning from 2018 to 2025.
• Includes links to official project pages, code repositories (JAX, TensorFlow, PyTorch), and technical blogs for many listed papers.
• Features information on related workshops and challenges, such as CVPR's "World Model" track and the "WorldModelBench" initiative.

Maintenance & Community

The repository is maintained by LMD0311 and welcomes contributions via pull requests or issues to expand its coverage. It encourages users to star the repository and provides a citation for its use in research.

Licensing & Compatibility

The repository itself is not software with a license. The licenses of the individual projects linked within the repository vary and should be checked on their respective pages.

Limitations & Caveats

This is a reference list and does not provide executable code or a unified framework. The rapid pace of research means the list may not be exhaustive and requires continuous updates.