mpc_python

I keep here my (old) notebooks on Model Predictive Control for path-following problems. Includes a Pybullet simulation to demo the controller. This mainly uses CVXPY as a framework. This repo contains code from other projecs, check them out in the special thanks section.

Contents

Jupyter Notebooks

1. State space model derivation -> analytical and numerical derivaion of the model

2. MPC -> implementation and testing of various tweaks/improvements

3. Obstacle Avoidance -> Using halfplane constrains to avaoid track collisions -> Sill work in progress!

Results

Racing car model is from: https://github.com/erwincoumans/pybullet_robots.

Results:

The settings used for tuning the MPC controller are in the mpc_config class.

Usage

The results above can be reproduced both in Docker and Conda.

Docker

From this repository root directory:

docker build -t mpc-demo -f docker/Dockerfile .

• To run the pybullet demo:

xhost +local:docker
docker run -it --net=host --ipc=host --privileged \
    --env="DISPLAY" \
    --env="QT_X11_NO_MITSHM=1" \
    --volume="/tmp/.X11-unix:/tmp/.X11-unix:rw" \
    --volume="${XAUTHORITY}:/root/.Xauthority" \
    mpc-demo:latest \
    bash -c "python3 mpc_demo_pybullet.py"

• To run the simulation-less demo: change the last command to python3 mpc_demo_nosim.py.
  • this is a simpler demo that does not use pybullet, useful for debugging.

In both cases the script will promt the user to press enter before starting the demo, pybullet may take a few seconds.

Conda Environment

The environment used for this project can be repoduced via conda:

conda env create -f env.yml
conda activate simulation

The demos can be run with:

python3 mpc_demo_pybullet.py
python3 mpc_demo_nosim.py

This environment also includes jupyter lab to experiment with the jupyter notebooks.

References & Special Thanks ⭐ :

• Prof. Borrelli - mpc papers and material
• AtsushiSakai - pythonrobotics
• erwincoumans - pybullet
• alexliniger - mpcc and his paper
• arex18 - rocket-lander