Simple MPPI Implementation with Python

Model Predictive Path-Integral (MPPI) Control [G. Williams et al., 2018] is a promising sampling-based optimal control algorithm.
This repository is for understanding the basic idea of the algorithm.

Dependency

• uv

  • A simple and fast Python package manager. Refer to the official documentation for one-command installation.
  • Only numpy, matplotlib, notebook are needed to run all scripts in this repository.

• ffmpeg

  • mp4 movie writer

  • installation details

    • For Ubuntu Users
      • sudo apt-get update
      • sudo apt-get -y install ffmpeg
    • For Windows Users
      • Install scoop
      • scoop install ffmpeg
    • For macOS Users
      • Install homebrew
      • brew install ffmpeg
    • Check the official website if necessary
      • https://ffmpeg.org/

Setup

[Option 1] Native environment

git clone https://github.com/MizuhoAOKI/python_simple_mppi.git
cd python_simple_mppi
uv sync

[Option 2] Docker environment

CLICK HERE TO EXPAND

1. Install docker.

2. Clone the project repository.

   cd <path to your workspace>
   git clone https://github.com/MizuhoAOKI/python_simple_mppi.git
   

3. Run setup to build the docker image, launch the docker container and get into the bash inside. Building the image might take a few minutes.

   cd <path to your workspace>/python_simple_mppi
   docker compose run --rm dev
   

   This command will automatically build the Docker image if needed. Once the container starts, any changes made in the local repository on the host will be reflected inside the container, and vice versa.

Usage

Path Tracking

Simple Path Tracking

• Run simulation

  cd python_simple_mppi
  uv run scripts/mppi_pathtracking.py

• Run jupyter notebook if you would like to check mathematical explanations on the algorithm.

  cd python_simple_mppi
  uv run jupyter notebook --notebook-dir=. notebooks/mppi_pathtracking.ipynb

Path Tracking with Obstacle Avoidance

• Run simulation

  cd python_simple_mppi
  uv run scripts/mppi_pathtracking_obav.py

• Run jupyter notebook if you would like to check mathematical explanations on the algorithm.

  cd python_simple_mppi
  uv run jupyter notebook --notebook-dir=. notebooks/mppi_pathtracking_obav.ipynb

Pendulum

• Run simulation to swing up a pendulum.

  cd python_simple_mppi
  uv run scripts/mppi_pendulum.py

• Run jupyter notebook if you would like to check mathematical explanations on the algorithm.

  cd python_simple_mppi
  uv run jupyter notebook --notebook-dir=. notebooks/mppi_pendulum.ipynb

CartPole

• Run simulation of cartpole

  cd python_simple_mppi
  uv run scripts/mppi_cartpole.py

• Run jupyter notebook if you would like to check mathematical explanations on the algorithm.

  cd python_simple_mppi
  uv run jupyter notebook --notebook-dir=. notebooks/mppi_cartpole.ipynb

References

1. G. Williams et al. "Information-Theoretic Model Predictive Control: Theory and Applications to Autonomous Driving"
   • URL : https://ieeexplore.ieee.org/document/8558663
   • PDF : https://arxiv.org/pdf/1707.02342.pdf