Action Chunking Transformer

Advanced robotics imitation learning project using transformer-based action chunking for fine-grained bimanual manipulation tasks.

Demonstration Video 1: Imitation learning demonstration

Demonstration Video 2: run time using trained model with 34 episodes

Resources

Dataset: realsense-black-green-background-lerobot
A comprehensive robotics video dataset for imitation learning, fully compatible with LeRobot and RLDS frameworks.
Trained Model: imitation-blue-realsense124Combo
Pre-trained model checkpoint with complete configuration files for robotics imitation learning applications.

Network Architecture

The Action Chunking Transformer architecture showing the hierarchical structure for processing sequential action data.

References

Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware

@article{zhao2023learning,
  title={Learning fine-grained bimanual manipulation with low-cost hardware},
  author={Zhao, Tony Z and Kumar, Vikash and Levine, Sergey and Finn, Chelsea},
  journal={arXiv preprint arXiv:2304.13705},
  year={2023}
}

Project Repository: https://github.com/huggingface/lerobot
Open-source implementation and framework for robotics learning and imitation.