HILO-MPC Documentation

HILO-MPC is a toolbox for easy, flexible and fast development of machine-learning-supported optimal control and estimation problems.

This tool can leverage TensorFlow and PyTorch to create machine learning models, and the CasADi framework to efficiently build control and estimation problems.

_images/overview.svg

At the moment HILO-MPC supports:

  1. Control

    1. Nonlinear and linear model predictive control

    2. Model predictive control for path following

    3. Model predictive control for trajectory tracking

    4. PID controller and linear quadratic regulator

  2. Machine Learning

    1. Artificial neural networks

    2. Gaussian processes

  3. Estimation

    1. Moving horizon estimation

    2. Kalman filter (including nonlinear extensions)

    3. Particle filter

  4. Modeling

    1. Ordinary differential equations

    2. Differential algebraic equations

  5. Embedded

    1. \(\mu\text{AO-MPC}\) (code generation software for linear model predictive control)

On the pipeline we have:

  1. Control

    1. Mixed-integer linear model predictive control (work in progress)

    2. Tube-based model predictive control (work in progress)

    3. Multi-mode model predictive control

  2. Machine Learning

    1. Recurrent neural networks

    2. Physics-informed training of neural networks

    3. Reinforcement learning

  3. Estimation

    1. Multi-rate moving horizon estimation

  4. Embedded

    1. SAM (solver for Al’brekht’s method)

Quick start:

Examples:

Modules:

API:

Indices and tables