MATMPC
This tool aims at providing an easy-to-use nonlinear MPC implementation. The optimal control problem (OCP) that should be solved is transcribed by multiple shooting and the resulting nonlinear program (NLP) is solved by Sequential Quadratic Programming (SQP) method.
The tool supports fixed step (explicit/implicit) Runge-Kutta (RK) integrator for multiple shooting. The derivatives that are needed to perform optimization are obtained by CasADi (https://github.com/casadi/casadi/wiki), the state-of-the-art automatic/algorithmic differentiation toolbox. The Quadratic Programming (QP) problems can be solved by both dense and sparse solvers. By now, MATMPC supports interfaces with the following external solvers: qpOASES (https://projects.coin-or.org/qpOASES/wiki/QpoasesInstallation), Ipopt (https://projects.coin-or.org/Ipopt), hpipm (https://github.com/giaf/hpipm), osqp (https://osqp.org/) and qpalm (https://github.com/Benny44/QPALM).
The most unique feature of MATMPC is that it does not require to install any external libraries. The users need not to understand how to make, compile and link any library. Except for external QP solvers, all algorithmic routines are written directly using MATLAB C API and can be compiled into independent MEX functions using compilers that belong to GCC class (e.g. GCC, MinGW and Clang). MATMPC employs MATLAB built-in linear algebra library provided by Intel MKL. Therefore, MATMPC is able to provide compatible runtime performance as other libraries written directly in C/C++.
The mathematics and parameter tuning of MATMPC is explained at matmpc-docs.readthedocs.io/en/latest/
Please install Matlab supported MinGW compiler at https://www.mathworks.com/matlabcentral/fileexchange/52848-matlab-support-for-mingw-w64-c-c-compiler
Install gcc by running "sudo apt-get install gcc"
Note: MATLAB is incompatible with the latest releases of gcc, It is recommended to download the gcc version that is compatible with your MATLAB version. Go check Compilers - MATLAB & Simulink (mathworks.com).
Install Xcode from app store
-
Download and install CasADi-3.3.0. MATMPC currently is compatible with CasADi 3.3.0.
-
Write your own model using following the styles given by examples, e.g. Inverted Pendulum.
-
In the model file you created, set your own sampling time and multiple shooting time.
-
Run Model_Generation.m
-
In InitData.m, set your own initialization data, e.g. data path, initial states, references and etc.
-
In Simulation.m, choose your integrator, set the prediction horizon and the solver options. You may also need to modify the reference type for constant or time-varying reference tracking problems.
-
In Draw.m, write your own plot functions to display your results.
-
Run Simulation.m and see the results!
please read /doc/HPIPM-tutorial for the detailed installation process
引用格式
Yutao Chen (2024). MATMPC (https://github.com/chenyutao36/MATMPC/releases/tag/v1.0.0), GitHub. 检索来源 .
MATLAB 版本兼容性
平台兼容性
Windows macOS Linux标签
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!examples
mex_core
nmpc
solver/linux/osqp
solver/linux/qpoases
solver/mac/qpalm
solver/mac/qpoases
solver/win64/osqp
solver/win64/qpoases
版本 | 已发布 | 发行说明 | |
---|---|---|---|
1.0.0 |