Imitation Learning from Observation with Automatic Discount Scheduling

[Website] [arXiv] [OpenReview]

Instructions

Environment Setup

• Install Mujoco based on the instructions given here.

• Install the following libraries:

  sudo apt update
  sudo apt install libosmesa6-dev libgl1-mesa-glx libglfw3
  

• Install other dependencies:

  conda env create -f conda_env.yml
  conda activate ads
  

Run the Code

• You can download the expert demonstrations used in our experiments from this link or generate new demonstrations through metaworld_generate_expert/generate_demo.py. Then place the expert_demos folder in ${root_dir}/IL.
• Run experiments by the following command:

  python train.py agent=ot suite=metaworld obs_type=pixels suite/metaworld_task=hammer num_demos=10 seed=1 suite.num_train_frames=2000000 adaptive_discount=true
  

  The hyperparameter adaptive_discount controls whether to use Automatic Discount Scheduling.

Citation

Please use the following bibtex for citations:

@inproceedings{liu2024imitation, title={Imitation Learning from Observation with Automatic Discount Scheduling},
author={Yuyang Liu and Weijun Dong and Yingdong Hu and Chuan Wen and Zhao-Heng Yin and Chongjie Zhang and Yang Gao},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024},
url={https://openreview.net/forum?id=pPJTQYOpNI}
}

Acknowlegment

This codebase is built upon the ROT codebase. The test environments are from Meta-World.