Control Toolbox (C++)

This is my personal passion project C++ library for modern control and optimization algorithms. The goal of this repository is to learn control theory and optimization and the implementation of these in C++. I am building the algorithms from scratch myself to learn how they work. The rule for this project is to never use any other external libraries than the C++ standard library. The goal is that the developed algorithms are robust and efficient enough to be used in real-world aplications.

Key Features

• Model Predictive Control (MPC): Formulated for Linear TIme-invariant (LTI) MIMO systems. Supports Linear and Quadratic cost functions.
• Custom optimization algorithms
  • Simplex algorithm:
  • Active Set algortihm:
• PID control: Both algorithm support antiwindup
  • Positional PID-control algorithm
  • Velocity form PID-control algorithm

Validation and results

The C++ implementations in this library are verifies against MATLAB. Simplex- and Active Set -algortihms were tested against matlab and gave same results. The MPC-algortihm was also tested against matlab and gave almost identical results.

MPC-algortihm was tested on a 2 state integrator system and as seen below the results are almost identical to matlabs with the QP-cost

Building tests

This project uses CMake. From the repo root:

mkdir build && cd build
cmake ..
cmake --build .

This builds simplex_test, active_set_test, and mpc_test into the build/ folder.

Running tests

MPC solver for Linear time invariant MIMO systems

run the code:

./mpc_test

Active set algorithm

run the code:

./active_set_test

Simplex algorithm

run the code:

./simplex_test

Roadmap and future plans

I'm Currently working on implementing a system/plant class to streamline simulations and system definitions. Improvements and test for the PID class are also being worked on.

Next features

• State feedback controller
• State predictor
• Kitagawa algorithm
• Kalman predictor
• Kalman filter