Skip to content

Rethinking Latent Representations in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation

License

Notifications You must be signed in to change notification settings

BaiShuanghao/BC-IB

Repository files navigation

Rethinking Latent Representations in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation

python pytorch lightning hydra license

Project Page | Paper | arXiv

Shuanghao Bai, Wanqi Zhou, Pengxiang Ding, Wei Zhao, Donglin Wang, Badong Chen.

🌟 Highlights

Logo

Abstract: Behavior Cloning (BC) is a widely adopted visual imitation learning method in robot manipulation. Current BC approaches often enhance generalization by leveraging large datasets and incorporating additional visual and textual modalities to capture more diverse information. However, these methods overlook whether the learned representations contain redundant information and lack a solid theoretical foundation to guide the learning process. To address these limitations, we adopt an information-theoretic perspective and introduce mutual information to quantify and mitigate redundancy in latent representations. Building on this, we incorporate the Information Bottleneck (IB) principle into BC, which extends the idea of reducing redundancy by providing a structured framework for compressing irrelevant information while preserving task-relevant features. This work presents the first comprehensive study on redundancy in latent representations across various methods, backbones, and experimental settings, while extending the generalizability of the IB to BC. Extensive experiments and analyses on the CortexBench and LIBERO benchmarks demonstrate significant performance improvements with IB, underscoring the importance of reducing input data redundancy and highlighting its practical value for more practical applications.

Main Contributions
  1. We extend the IB to BC and provide a comprehensive study on the impact of latent representation redundancy in BC for robot manipulation.
  2. We empirically demonstrate that minimizing redundancy in latent representations helps existing BC algorithms significantly improve generalization performance on the Cortexbench and LIBERO benchmarks across various settings, indirectly highlighting the considerable redundancy present in current robot trajectory datasets.
  3. We provide a detailed theoretical analysis explaining why IB enhances the transferability of BC methods.

🛠️ Installation

Refer to INSTALL.md for installation instructions.

Check ERROR_CATCH.md for troubleshooting steps based on the errors I personally encountered during installation.

💾 Data Preparation

Refer to DOWNLOAD_DATASET.md for instructions on downloading datasets.

💻 Baseline Preparation

Refer to DOWNLOAD_MODEL.md for instructions on downloading pre-trained models of baselines.

📈 Usage

Refer to USAGE.md for instructions on training and evaluation.

📨 Contact

If you have any questions, please create an issue on this repository or contact us at baishuanghao@stu.xjtu.edu.cn.

📝 Citation

If you find our work useful, please consider citing:

@article{bai2025rethinking,
  title={Rethinking Latent Representations in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation},
  author={Bai, Shuanghao and Zhou, Wanqi and Ding, Pengxiang and Zhao, Wei and Wang, Donglin and Chen, Badong},
  journal={arXiv preprint arXiv:2502.02853},
  year={2025}
}

🙏 Acknowledgements

For LIBERO experiments, our code is primarily built upon the LIBERO source code. The baseline implementations are mainly adapted from the LIBERO source code, while the environment is set up using robosuite.

For CortexBench experiments, our code is primarily based on the eai-vc source code. The baselines reference ResNet, ViT, R3M, MVP, VC-1, Voltron, and MPI, while the environment setup is based on CortexBench, dmc2gym, MetaWorld, trifinger_simulation, gym, and MuJoCo.

About

Rethinking Latent Representations in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published