SafePILCO is a software tool for safe and data-efficient policy search with reinforcement learning. It extends the known PILCO algorithm, originally written in MATLAB, to support safe learning. We provide a Python implementation and leverage existing libraries that allow the codebase to remain short and modular, which is appropriate for wider use by the verification, reinforcement learning, and control communities.
@article{arxiv.2008.03273,
title = {SafePILCO: a software tool for safe and data-efficient policy synthesis},
author = {Kyriakos Polymenakos and Nikitas Rontsis and Alessandro Abate and Stephen Roberts},
journal= {arXiv preprint arXiv:2008.03273},
year = {2020}
}
Comments
Shorter Version published as a software tool demonstration at QEST 2020