The Simulated Industrial Manufacturing and Process Control Learning Environments (SMPL) supplements several process control environments to the Openai Gym family, which quenches the pain of performing Deep Reinforcement Learning algorithms on them. Furthermore, we provided d4rl-like wrappers for accompanied datasets, making Offline RL on those environments even smoother.
For the paper SMPL: Simulated Manufacturing and Process Control Learning Environments, you can cite with
@inproceedings{
zhang2022smpl,
title={{SMPL}: Simulated Industrial Manufacturing and Process Control Learning Environments},
author={Mohan Zhang and Xiaozhou Wang and Benjamin Decardi-Nelson and Song Bo and An Zhang and Jinfeng Liu and Sile Tao and Jiayi Cheng and Xiaohong Liu and Dengdeng Yu and Matthew Poon and Animesh Garg},
booktitle={Thirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
year={2022},
url={https://openreview.net/forum?id=TscdNx8udf5}
}
The documentation is available here! https://smpl-env.github.io/smpl-document/index.html
$ git [email protected]:smpl-env/smpl.git $ cd smpl $ pip install .
Note
You will need to build the PenSimPy environment with SMPL separately. Namely, build and install fastodeint following this instruction, then install PenSimPy.
For Linux users, you can just install fastodeint and PenSimPy by executing the following commands:
$ sudo apt-get install libomp-dev $ sudo apt-get install libboost-all-dev $ git clone --recursive [email protected]:smpl-env/fastodeint.git $ cd fastodeint $ pip install . $ cd .. $ git clone --recursive [email protected]:smpl-env/PenSimPy.git $ cd PenSimPy $ pip install .
If you also want to use the pre-built MPC and EMPC controllers, you would need to install mpctools by CasADi. For Linux users, you can execute the following commands:
$ git clone --recursive [email protected]:smpl-env/mpc-tools-casadi.git
$ cd mpc-tools-casadi
$ python mpctoolssetup.py install --user
See the jupyter notebook for example use cases.