Software that integrates various imitation learning methods and benchmark task environments to provide baselines for robot manipulation
RoboManipBaselines_VideoVer100.mp4
This quick start allows you to collect data in the MuJoCo simulation and train and rollout the ACT policy.
See the installation documentation.
Spatial attention recurrent neural network
Action Chunking with Transformers
Diffusion Policy
Multi-Task Action Chunking Transformer
See the dataset list.
See the learned parameters for policies learned from these datasets.
See teleop.
See the environment catalog for a full list of environments.
See envs for installation procedures for each environment.
See utils.
If you would like to contribute to this repository, please check out the contribution guide.
Files that originate from this repository are subject to the BSD 2-Clause License. If a file explicitly states a different license, or if there are different license files in a directory, those licenses will take precedence. For files in third-party directories, please follow the respective licenses.
You can cite this work with:
@software{RoboManipBaselines_GitHub2024,
author = {Murooka, Masaki and Motoda, Tomohiro and Nakajo, Ryoichi},
title = {{RoboManipBaselines}},
url = {https://github.com/isri-aist/RoboManipBaselines},
version = {1.0.0},
year = {2024}
month = dec,
}