MPCLib

MPCLib is a C++ library that solves Model Predictive Control problems. It can also be used to solve strictly convex quadratic minimization problems by using the embedded QuadProg++ library.

The library defines a very simple interface that can be used to enable control and output constraints, slack variables, minimization weights, etc. In turn, the MPC interface uses the QuadraticProblem and the Constraint classes to easily define constraints and add them to the minimization problem.

Please be aware that this software is still in its early life. It works fine but it is not complete. The software still doesn't provide useful errors or exceptions when something is misconfigured. The project will be updated over time. Feel free to contribute.

Building and running the sample application

The project is cmake based. To build it simply type

mkdir build
cd build
cmake ..
make

To run some tests and generate some plots simply run test-mpc.sh in the main folder. The script runs 6 different tests producing csv files as output. If you have R installed, the script will also plot the results. For more formal information on the problem the library solves and on the examples please refer to the PDF in the doc folder.

Included software

The library includes other open source software (please see the LICENSE file):

• QuadProg++: this is the library that solves the quadratic optimization problem;
• TCLAP: this is a command line arguments parsing library that is used in the sample program;

Contribution

Copyright (C) 2017 Michele Segata