The article: PLOS ONE

DQN, PPO, SAC for real/simulated cart-pole problem. The parameters are optimized using Optuna library.

Simulation based on real-setup parameters and model.

To generate the simulation of different parameters

python EJPH/generate_data.py

To plot the simulation results of different parameters

python EJPH/plots.py

rpi_control for environement interface of PC and raspberry pi written in python/c++

If you want c++ server/client with python client, you can use this. Also, tutorials and examples are available in this repository.

The video of installation and deep RL (DQN) control is here

LQR with Lyapunov-based swing-up is here