This is an extension package for the NASA Astrobee project which allows MPC with Acados.

This assumes you have the astrobee software already installed and built.

1. Clone this repo in a suitable directory, e.g.

cd $ASTROBEE_WS/src/gnc
git clone https://github.com/DISCOWER/ctl_acados_mpc.git
cd ctl_acados_mpc

2. Now wee need to install acados. First run git submodule update --recursive --init

3. Make a build directory and install acados. You can choose which solvers you would like to use with optional flags

cd acados
mkdir -p build
cd build
cmake .. # with optional arguments e.g. -DACADOS_WITH_DAQP=OFF/ON -DACADOS_INSTALL_DIR=<path_to_acados_installation_folder>
make install

4. In case you want to rebuild the c-code for the MPC (e.g. with a different QP solver, different horizon): Install the python interface with pip: pip install -e ./interfaces/acados_template Note that some parameters can be changed online (costs, state tolerances, inertia and mass)

5. Add the compiled shared library paths to the following variables:

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:"<acados_root>/lib"
export ACADOS_SOURCE_DIR="<acados_root>"

6. Build this package by running catkin build --this --no-deps

1. Open 2 terminals, go to your $ASTROBEE_WS and run source devel/setup.bash on each

2. In the first terminal, launch the astrobee. E.g. for a simulation, do roslaunch astrobee astrobee_sim.launch. Check out the launch files for available settings

3. In the second terminal, you can send the astrobee on a mission! E.g. run: rosrun executive teleop_tool -move -relative -pos "0 -1 0"

5. Open /solver/generate_solver.py and tune the settings to your liking. Then run python generate_solver.py

6. Re-build the package. You can do this by making sure you are in the package root folder (e.g. $ASTROBEE_WS/gnc/ctl_acados_mpc) and running catkin build --this --no-deps

A special thanks to Pedro Roque for instights, bug-fixing help and code contribution in the CMakeLists.txt.