nmpc-mav

Nonlinear model predictive control for leader-follower quadrotors with observability constraint. This simulation takes the range-based multi-robot model and Acados MPC library, to compute the optimal control inputs for better localization and control.

Installation of Acados

Install acados according to https://github.com/acados/acados. The steps shoud be:

$ git clone https://github.com/acados/acados.git
$ git submodule update --recursive --init
$ mkdir -p build
$ cd build
$ cmake ..
# cmake -DACADOS_WITH_QPOASES=ON ..
# add more optional arguments e.g. -DACADOS_WITH_OSQP=OFF/ON -DACADOS_INSTALL_DIR=<path_to_acados_installation_folder> above
$ make install

Create Python env

This procedure is according to https://docs.acados.org/python_interface/index.html, which should be

$ sudo apt install virtualenv
$ virtualenv envmpc --python=/usr/bin/python3.7 #sudo rm -rf envmpc
$ source envmpc/bin/activate
$ pip3 install -e /home/lss/acados/interfaces/acados_template
$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:"/home/lss/acados/lib"
$ export ACADOS_SOURCE_DIR="/home/lss/acados"
Hint: you can add these two lines to your .bashrc/.zshrc.

Run the nmpc-mav code

$ source envmpc/bin/activate
$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:"/home/lss/acados/lib"
$ export ACADOS_SOURCE_DIR="/home/lss/acados"
$ git clone https://github.com/shushuai3/nmpc-mav.git
$ cd nmpc-mav
$ python main.py