#POD2C Partially-Observed Decoupled Data-based Control implementation on MuJoCo 200 Win64

---

The algorithm has two steps:

1. Find the optimal control sequence in the deterministic system using arma-ilqr method.
2. Wrap a linear local feedback policy around the openloop nominal trajectory using arma-lqg method.

Requirements

• Matlab 2018 or later

• MuJoCo 200 Windows x64 version

  Authors: Ran Wang, Raman Goyal.

  The algorithm is the implementation of the work in the paper "Information State based Reinforcement Learning for the Control of Partially Observed Nonlinear Systems" Arxiv: https://arxiv.org/pdf/2107.08086