This repository is a curated, continuously updated list of research papers and resources on Model-Based Reinforcement Learning (MBRL). It serves as a comprehensive reference for researchers and practitioners in the field, tracking the latest advancements and providing a structured overview of the landscape.

How It Works

The repository categorizes MBRL papers by conference (ICLR, NeurIPS, ICML) and year, starting from 2021, with a dedicated section for "Classic Model-Based RL Papers." It also includes a taxonomy that broadly divides MBRL algorithms into "Learn the Model" and "Given the Model" approaches, with examples and links to relevant papers. Each paper entry typically includes the title, authors, a link, key insights, and experimental environments.

Quick Start & Requirements

This is a resource repository, not a software library. No installation or execution is required.

Highlighted Details

• Extensive collection of papers from major AI conferences (ICLR, NeurIPS, ICML) from 2021 onwards, plus foundational MBRL work.
• A taxonomy categorizing MBRL algorithms into "Learn the Model" and "Given the Model" paradigms.
• Detailed paper entries with links, authors, key insights, and experimental environments.
• Regular updates to include the latest research, such as ICLR 2025 and NeurIPS 2024 papers.

Maintenance & Community

The repository is maintained by opendilab and is open for contributions. Further details on contributing can be found via a link within the README.

Licensing & Compatibility

Awesome Model-Based Reinforcement Learning is released under the Apache 2.0 license. This license is permissive and allows for commercial use and integration into closed-source projects.

Limitations & Caveats

As a curated list, the repository itself does not implement any algorithms. The "Key" insights for each paper are brief summaries and may not capture the full nuance of the research. The "ExpEnv" field lists environments, but actual implementation details or code availability for each paper are not guaranteed by this repository.