Robust Adversarial Imitation Learning via Adaptively-Selected Demonstrations

This repository contains the PyTorch code for the paper "Robust Adversarial Imitation Learning via Adaptively-Selected Demonstrations" in IJCAI 2021. [Paper][Appendix]

Requirements

Experiments were run with Python 3.6 and these packages:

pytorch == 1.1.0
gym == 0.15.7
mujoco-py == 2.0.2.9

Train SAIL

SAIL with soft weight

 python sail_main.py --env_id 2/4/5/7 --il_method sail --weight --soft --seed 0

SAIL with binary weight

 python sail_main.py --env_id 2/4/5/7 --il_method sail --weight --seed 0

GAIL/AIRL/VAIL/InfoGAIL

 python sail_main.py --env_id 2/4/5/7 --il_method gail/airl/vail/infogail --seed 0

For other compared methods, the re-implementation of 2IWIL/IC-GAIL and D-REX can be found in core/irl.py and trex_main.py.

After saving the policy model during training, you can plot the optimization trajectories in Figure 2 under the instruction of https://github.com/tomgoldstein/loss-landscape.

Acknowledegement

We thank the authors of VILD. Our code structure is based on their source code and we also use expert data collected by VILD.

Contact

For any questions, please feel free to contact me. (Email: yunke.wang@whu.edu.cn)

Citation

@inproceedings{ijcai2021-0434,
  title     = {Robust Adversarial Imitation Learning via Adaptively-Selected Demonstrations},
  author    = {Wang, Yunke and Xu, Chang and Du, Bo},
  booktitle = {Proceedings of the Thirtieth International Joint Conference on
               Artificial Intelligence, {IJCAI-21}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Zhi-Hua Zhou},
  pages     = {3155--3161},
  year      = {2021},
  month     = {8},
  note      = {Main Track}
  doi       = {10.24963/ijcai.2021/434},
  url       = {https://doi.org/10.24963/ijcai.2021/434},
}

Reference

[1] Generative adversarial imitation learning. NeurIPS 2016.

[2] Learning robust rewards with adversarial inverse reinforcement learning. ICLR 2018.

[3] Variational discriminator bottleneck: Improving imitation learning, inverse rl, and gans by constraining information flow. ICLR 2017.

[4] InfoGAIL: Interpretable Imitation Learning from Visual Demonstrations. NeurIPS 2017

[5] Imitation learning from imperfect demonstration. ICML 2019.

[6] Extrapolating beyond suboptimal demonstrations via inversere inforcement learning from observations. ICML 2019.

[7] Better-than-demonstrator imitation learning via automatically-ranked demonstrations. CoRL 2020.

[8] Variational Imitation Learning with Diverse-quality Demonstrations. ICML 2020.

[9] Learning to Weight Imperfect Demonstrations. ICML 2021

[10] Unlabeled Imperfect Demonstrations in Adversarial Imitation Learning. AAAI 2023.

yunke-wang / SAIL