/Autonomous-Vehicle-Trajectory-Control-Using-Non-Linear-MPC_in_C-

- Developed a non-linear Model Predictive Controller (MPC) for trajectory following on a kinematic car bicycle model. - Managed both lateral and longitudinal control for optimal vehicle navigation. - Implemented the solution in C++ within a Linux (Ubuntu) environment.

Primary LanguageC++

libmpc++

libmpc++ is a C++ library to solve linear and non-linear MPC. The library is written in modern C++20 and it is tested to work on Linux, macOS and Windows.

Requirements

  • gcc version (>= 10.3.0)

The libmpc++ website can be found at the following link: https://altairlab.gitlab.io/optcontrol/libmpc/

Starting from the version 0.5.0 the library is also available as a Python package (https://pypi.org/project/pympcxx/). The Python package is available on PyPI and can be installed using pip:

pip install pympcxx

Dependecies

The library depends on the following external libraries which must be installed on the machine before using libmpc++

If you are a developer, to setup the debug environment you also need to install:

Example

There is also a collection of examples that guide you through the most important features of libmpc++ in the examples folder

  • quadrotor_ex.cpp: example of a linear MPC for a quadrotor regulation problem
  • ugv_ex.cpp: example of a non-linear MPC for a UGV tracking problem with obstacles
  • vanderpol_ex.cpp: example of a non-linear MPC for a Van der Pol oscillator regulation problem
  • networked_oscillator_ex.cpp: example of a non-linear MPC for a set of coupled Van der Pol oscillator regulation problem

The examples are written in C++ and can be compiled using CMake:

mkdir build
cd build
cmake ..
make

Usage

The latest version of libmpc++ is available from GitHub https://github.com/nicolapiccinelli/libmpc/releases and does not require any installation process other than the one required by its dependecies.

System wide usage on an Ubuntu Linux

If you're a developer run the configure.sh script with superuser privileges to install the external dependencies

sudo ./configure.sh

or configure.sh --disable-test if you do not need the test-suite

sudo ./configure.sh --disable-test

Then, after installing all the dependencies configure and compile the library using CMake. From the folder containing the CMakeLists.txt file compile the library by doing

mkdir build
cd build
cmake ..
make

Finally, install the library with superuser privileges

sudo make install

This will allow you to install the MPC++ library on your system. Using the include

#include <mpc/LMPC.hpp>

in your package will be enough to add this library to your project.

CMakelists.txt example

This is an example of a CMakeLists.txt file to use libmpc++ in your project.

cmake_minimum_required(VERSION 3.0)
project(your_project_name)

# set the C++ standard to C++ 20
set(CMAKE_CXX_STANDARD 20)
# set the C++ compiler to use O3
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O3")

find_package(mpc++ CONFIG REQUIRED)

# Declare a C++ library
include_directories(${mpc++_INCLUDE_DIRS})
add_executable(${PROJECT_NAME} main.cpp)
target_link_libraries(${PROJECT_NAME} mpc++)

Development Container

Docker is helpful for creating a stable localized development environment. Install Docker with the convenience script:

curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh

You can build the development image with the command below. Note that the image will be tagged with the name specified after the '-t' option.

docker build -t dev_image .

###readme_abdalla

READ_ME_FILE Abdalla Mohamed ////// install the dependencies and set up the MPC main Library /////////////

extract MAIN_DIRECTORY cd lib cd libmpc

sudo ./configure.sh // run the configure.sh script with superuser privileges to install the external dependencies or configure.sh --disable-test if you do not need the test-suite

sudo ./configure.sh --disable-test Then, after installing all the dependencies configure and compile the library using CMake. From the folder containing the CMakeLists.txt file compile the library by doing

mkdir build cd build cmake .. make Finally, install the library with superuser privileges

sudo make install This will allow you to install the MPC++ library on your system. Using the include /////////////////////////////////////////////////////////////////////////////////////////////// // if it doesnt work git clone https://github.com/nicolapiccinelli/libmpc #include <mpc/LMPC.hpp>

└── main_directory └── lib └── libmpc └── examples

cd MPC_ISA cd MAIN_DIRECTORY cd lib cd libmpc cd examples mkdir build && cd build // make a build folder then change directory to build cmake .. make
cd bin ls ./MPC_CHANGING_LANES_VARYING_VELOCITY or ./MPC_SOLVER_LANE_CHANGE_CONSTANT_VELOCITY or ./MPC_solver_trajectory_const_vel ./ MPC_TRAJECTORY run any of the four testing scripts

  1. MPC_CHANGING_LANES_VARYING_VELOCITY -- > the lane changes the lane and varies its velocity doing both longitidunal and lateral control sucessfully after solving a csv file : MPC_CHANGING_LANES_VARYING_VELOCITY.csv" is generated in the bin folder ( same folder as the executable)

  2. MPC_SOLVER_LANE_CHANGE_CONSTANT_VELOCITY ---> changes the lane while keeping constant velocity a csv file "MPC_CHANGING_LANES_constant_VELOCITY.csv" containing all the model paramters is generated

  3. MPC_solver_trajectory_const_vel --> a constant velocity is initialized with a trajectory track of a polynorimal of n = 3 a csv file is generated "trajectory.csv" .

  4. ./ MPC_TRAJECTORY ----> a trajectory polynoimal control as well as a reference velocity is supplied to the model a csv file of all data is generated trajectory.csv

You can plot by the matlibplot scripts supplied within the zip folder .

:: reference library : https://github.com/nicolapiccinelli/libmpc

// old repo --> contains an obj oriented model and mpc solver implemented using singleon patterns and trying to solve using ipopt solver however the ipopt solver must have an HSL lisence ( wasted about 2 days time searching for those ) however it is fully implemented .