This is an implementation of Learning-Based Model Predictive Control (LBMPC) that uses the LBmpcIPM solver. LBmpcIPM is a sparse primal-dual infeasible start interior point method implementation based on Mehrotra's predictor-corrector scheme. This open-source solver is written in C++ and freely available under the BSD licence. The solver is tailored towards QP-LBMPC, and is provided to enable the rapid implementation of LBMPC on other platforms. Another salient feature is that its solving times scales linearly in the prediction horizon.

Aside from the implementation, this repository also contains the source code of LBmpcIPM and a short documentation on how to use the solver. Further details on the implementation can be found here

Prior to compiling and running the simulations, make sure the following applications and libraries are available on the computer:

• LBmpcTP.h (template class file of LBmpcIPM)
• Eigen3
• CMake

cd lbmpc_ipm
mkdir build
cd build
cmake ..
export N_MPC_STEPS=15 # or whatever..
make

(in MATLAB):

>> cd lbmpc_ipm/matlab/example0
>> Init

(in MATLAB):

>> cd lbmpc_ipm/matlab/qr_example
>> Init

cd lbmpc_ipm
build/bin/example0 matlab/example0/ConstrParam.bin

cd lbmpc_ipm
build/bin/qr_example matlab/qr_example/quad.bin

Back to home