control-playground

This is a repository for trying control algorithms, which is the content repository of the following article.

NonlinearMPCCasADi_.-180.0_0.0_dt-0.02-0.02_dead-0.0_f-0.0-0.0_n-False_e-False_q-False-72.mp4

I Implemented a double inverted pendulum simulator and tried the following algorithms.

Setup

Install rye according to the following.

curl -sSf https://rye.astral.sh/get | bash

Set up a virtual environment with rye.

source .rye/env
rye sync
source .venv/bin/activate

python control_double_inverted_pendulum.py

The default controller_model is LQR. You can change the controller_model by modifying the following line.

controller_mode = "LQR"

Also, if you set the following options to True, the selected controller will run the simulation only once from the given initial angle state.

run_once = True

If you set this to False, you can run control repeatedly from various initial values to find the ROA (Region of Attraction).

If matplotlib does not display properly in the virtual environment, run the following command.

sudo apt-get install python3-tk

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
controller		controller
dynamics		dynamics
estimator		estimator
media		media
simulator		simulator
utils		utils
visualization		visualization
README.md		README.md
control_double_inverted_pendulum.py		control_double_inverted_pendulum.py
double_inverted_pendulum.ipynb		double_inverted_pendulum.ipynb
pyproject.toml		pyproject.toml