This repository serves as the gathering place for algorithms and new ideas in emerging theories in robust control, reinforcement learning, and optimization.
Most of the code is written by Lekan Molu in python and occasionally in matlab.
All matlab files are in the folder matlab while python files are in the root folder.
- The code in the folder
tac_codecontains the matlab code and figures used in the paper:
-
`Mixed $H_2/H_\infty$ LQ Games for Robust Policy Optimization Under Unknown Dynamics. Transactions in Automatic Control (2022/2023) by Leilei Cui and Lekan Molu`.
-
All the codes in the root folder contain the routines and subroutines for the paper,
Mixed $\mathcal{H}_2/\mathcal{H}_\infty$-Policy Learning Synthesis. The International Federation of Automatic Control (IFAC) World Congress (2022/2023) by Lekan Molu and Hosein Hasanbeig.
-
Folders structure
-
dynsys: This folder defines the classes for the dynamical systems we have considered so far. -
hinfinity: Contains cool implementations of$H_\infty$ control algorithms. -
identify: Contains codes for interpretable neural network policies and NARMAX models used in our identification experiments. -
notebooks: Contains Jupyter notebooks used for testing and papers experimentation. -
linearsys: Contains common linear system utility routines and sub-routines. -
utils: Contains general purpose matlab-like functions for controllers and policy designs.
-
-
To run experiments for the TAC paper, open this notebook
-
To run the experiments for the IFAC World Congress paper, open this notebook
- Integrate this with the Reachability codebase.