NOTE: SOCKS is still under active (alpha) development 🚧👷🚧. Code is provided as-is, and some algorithms may not work as intended before the initial release. Please check back soon for the official release.
SOCKS is a suite of algorithms for stochastic optimal control using kernel methods.
It runs on top of OpenAI Gym, and comes with several classic controls Gym environments. In addition, it can integrate with many pre-existing Gym environments.
To install the toolbox, use pip install gym-socks. Alternatively, download the code
from the GitHub repo and install using pip install . from the code directory.
We support Python versions 3.7, 3.8, and 3.9 on Linux and macOS. We do not officially support Windows.
SOCKS comes with several examples in the GitHub repo. In order to run the examples,
first install the package and use python examples/<example> from the code directory.
For example, python python examples/control/tracking.py will run the optimal control
algorithm on the tracking benchmark using nonholonomic vehicle dynamics.
In order to cite the toolbox, use the following bibtex entry:
@inproceedings{thorpe2022hscc,
title={{SOCKS}: A Kernel-Based Stochastic Optimal Control and Reachability Toolbox},
authors={Thorpe, Adam J. and Oishi, Meeko M. K.},
year={2022},
booktitle={Proceedings of the 25th ACM International Conference on Hybrid Systems: Computation and Control (to appear)},
}