Ewerton R. Vieira, Aravind Sivaramakrishnan, Sumanth Tangirala, Edgar Granados, Konstantin Mischaikow, Kostas E. Bekris
2024 IEEE International Conference on Robotics and Automation (ICRA), 2024. Best Paper Award in Automation finalist.
MORALS combines autoencoding neural networks with Morse Graphs. It first projects the dynamics of the controlled system into a learned latent space. Then, it constructs a reduced form of Morse Graphs representing the bistability of the underlying dynamics, i.e., detecting when the controller results in a desired versus an undesired behavior.
pip install MORALS
examples/data/
. There should be a directory pendulum_lqr1k
and a labels file pendulum_lqr1k_success.txt
.python train.py --config pendulum_lqr.txt
.python get_MG_RoA.py --config pendulum_lqr.txt --name_out pendulum_lqr --RoA --sub 16
.MORALS/systems/
. At the very least, you must provide a name for your system so that its corresponding object can be returned by MORALS.systems.utils.get_system()
.examples/config/
. You can follow the format of the example config files provided.If you find this repository useful in your work, please consider citing:
@inproceedings{morals2024,
title={{\tt MORALS}: Analysis of High-Dimensional Robot Controllers via Topological Tools in a Latent Space},
author={Ewerton R. Vieira* and Aravind Sivaramakrishnan* and Sumanth Tangirala and Edgar Granados and Konstantin Mischaikow and Kostas E. Bekris},
booktitle={ICRA},
year={2024},
}