2.1/ Introduction

In here we are going to explain the different way of obtaining the code base. And in particular the dependencies. We highly recommand a binary installation of the dependencies.

2.2/ Binary Installation

2.2.1/ Install using the official image on UBUNTU 16.04

First clone the official image git repository.

git clone https://git-amd.tuebingen.mpg.de/amd-clmc/ubuntu_installation_scripts

The repositories contains convenience installation scripts for ubuntu. The most useful script is "official/setup_ubuntu". This script is meant to be called after a fresh installation of ubuntu, and install all typical dependencies required for programming robots. These include ROS, dynamic graph and related "robot-pkg" software (e.g stack of task and pinocchio). Only 16.04 is fully supported. The script will also work on 14.04, but mostly ROS and dependencies related to SL will be installed. Usage:

cd ubuntu_installation_scripts/official

sudo ./setup_ubuntu install

This setup script is sometimes updated. You may run it again:

cd ubuntu_installation_scripts/official

sudo ./setup_ubuntu update

To see the list of software this script install (and how it install it), visit the related dockerfile (e.g. for 16.04 : https://git-amd.tuebingen.mpg.de/amd-clmc/ubuntu_installation_scripts/blob/master/official/ubuntu_16_04/docker/Dockerfile).

With this process a file in /opt/openrobots/setup.bash should have been created. In order to use the binary installation you MUST source this file:

source /opt/openrobots/setup.bash

one can add this line to its ".bashrc"

2.2.2/ Do the same as the official image but install uniquely the dynamic graph.

First get the robotpkg PPA. Assuming you are using Ubuntu 16.04, if not, please adjust the "xenial" to the output of lsb_release -c:

sudo sh -c "echo 'deb [arch=amd64] http://robotpkg.openrobots.org/wip/packages/debian/pub xenialrobotpkg\ndeb [arch=amd64] http://robotpkg.openrobots.org/packages/debian/pub xenial robotpkg' > /etc/apt/sources.list.d/robotpkg.list"
curl http://robotpkg.openrobots.org/packages/debian
/robotpkg.key | sudo apt-key add -
sudo apt-get update

Install the following packages from robotpkg using apt-get:

sudo apt-get install -y robotpkg-dynamic-graph-v3 `# The dynamic graph` \
                        robotpkg-py27-dynamic-graph-v3 `# Thedynamic graph python bindings` \
                        robotpkg-tsid `# AndreaDelprete Task Space Inverse Dynamics` \
                        robotpkg-pinocchio `# Eigenbased rigid body dynamics library` \
                        robotpkg-hpp-fcl `# collision detection for pinocchio` \
                        robotpkg-libccd `# notsure` \
                        robotpkg-octomap `# notsure` \
                        robotpkg-parametric-curves `# Splineand polynomes library` \
                        robotpkg-simple-humanoid-description `# Simplehumanoid robot_properties package` \
                        robotpkg-eigen-quadprog `# Quadprog package` \
                        robotpkg-sot-core-v3 `# DynamicGraph Utilities` \
                        robotpkg-sot-tools-v3 `# DynamicGraph Utilities` \
                        robotpkg-sot-dynamic-pinocchio-v3 `# DGwrapper around pinocchio` \
                        robotpkg-sot-torque-control `# AndreaDelprete dynamic graph entities` \
                        robotpkg-py27-eigenpy `# Pythonbindings` \
                        robotpkg-py27-pinocchio `# Pythonbindings` \
                        robotpkg-py27-parametric-curves `# Pythonbindings` \
                        robotpkg-py27-sot-core-v3 `# Pythonbindings` \
                        robotpkg-py27-sot-torque-control `# Pythonbindings` \
                        robotpkg-py27-sot-dynamic-pinocchio-v3 `# Pythonbindings` \
                        robotpkg-py27-qt4-gepetto-viewer-corba `# LAAS 3Drobot viewer network client/server` ;

Export the following environment variables e.g. by adding exports in ~/. bashrc file:

export PATH=/opt/openrobots/bin:$PATH
export PKG_CONFIG_PATH=/opt/openrobots/lib/pkgconfig:$PKG_CONFIG_PATH
export LD_LIBRARY_PATH=/opt/openrobots/lib:/opt/openrobots/lib/plugin:$LD_LIBRARY_PATH
export PYTHONPATH=/opt/openrobots/lib/python2.7/site-packages:$PYTHONPATH
export ROS_PACKAGE_PATH="/opt/openrobots/share:$ROS_PACKAGE_PATH"

2.2.3/ Troubleshooting

Python 3: When using python3 and you installed packages with prefix /opt/openrobots/, you have to remove the python2.7 path and add the python3.5 one.
Compilation problem: upon catkin_make the package "dynamic-graph" is not found.
1. make sure the environment variable concerning /opt/openrobots has been set. See the previous paragraph. In order to verify this one can run:
  env | grep openrobots
2. The folder /opt/openerobots does belong to roor and/or does not have the correct permissions. In order to fix this one can do:
  sudo chown root:root -R /opt/openerobots
  sudo chmod 755 -R /opt/openerobots

2.3/ From source Installation

2.3.1/ For python2

first make sure you have treep installed.
pip2 install --user treep
or
pip3 install --user treep
Then clone the different packages.
- For just the dynamic graph:
  mkdir devel # or your favorite development folder
  cd devel
  git clone git@github.com:machines-in-motion/treep_machines_in_motion.git
  treep --clone DYNAMIC_GRAPH
  At this point you should see something like this:
- If you decided to clone every package it would look like this:
  mkdir devel # or your favorite development folder
  cd devel
  git clone git@github.com:machines-in-motion/treep_machines_in_motion.git
  treep --clone ALL_LAAS
  At this point you should see something like this:
Now let us assume you have something clone in your workspace/src folder.
- The next step is to generate the compilation script.
  treep --compilation-script
  The ouput:
  
  At this point treep generated a compilation script in bash that you need to execute anytime you do the compilation.
- If the compilation script is not an executable you should do a
  chmod +x compilation.sh
  The compilation script is based on the python script in the treep_dynamic_manager package. A quick glance should allow you to tune the installation folder and type of compilation (Debug/Release/...) A "dynamic_graph_setup.bash" is generated as well in the installation folder. If you source it you will tell your system to use the installed binary.
Launch the compilation:
./compilation.sh
This should build and install all the repositories.
- Troubleshooting with "sphinx"
  sudo -H pip2 install sphinx

2.3.2/ For using python3 alongside python2

Installing the eigenpy and pinocchio libraries into the same folder creates problem when loading the libraries. Instead of using the boost_python-py35 library, eigenpy also tries to load the boost_python- py27 library. To avoid this, it is recommended to

create a separate "devel" folder for the python3 installation like devel_py35
make sure to source the corresponding "dynamic_graph_setu p.bash". And to edit the
export PYTHONPATH=/opt/openrobots_py35/lib/python3.5/site-packages:$PYTHONPATH
Otherwise, eigenpy tries to load boost_python-py27 libraries. This boils down to using separate functions to source either python2 or python3:
source_devel_py27() {
# robotpkg installation.
export PATH=/opt/openrobots/bin:$PATH
export PKG_CONFIG_PATH=/opt/openrobots/lib/pkgconfig:$PKG_CONFIG_PATH
export LD_LIBRARY_PATH=/opt/openrobots/lib::/opt/openrobots/lib/plugin:$LD_LIBRARY_PATH
export PYTHONPATH=/opt/openrobots/lib/python2.7/site-packages:$PYTHONPATH
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/
}
source_devel_py35() {
export PATH=/opt/openrobots_py35/bin:$PATH
export PKG_CONFIG_PATH=/opt/openrobots_py35/lib/pkgconfig:$PKG_CONFIG_PATH
export LD_LIBRARY_PATH=/opt/openrobots_py35/lib:/opt/openrobots_py35/lib/plugin:$LD_LIBRARY_PATH
export PYTHONPATH=/opt/openrobots_py35/lib/python3.5/site-packages:$PYTHONPATH
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/
}
In order to build the pacakges for Python 3, use the cmake configuration as:
cmake .. -DCMAKE_BUILD_TYPE=Release -DPYTHON_EXECUTABLE=`which python3` -DCMAKE_INSTALL_PREFIX=/path/to/your/devel_py35/workspace/devel
For using the catkin project using python3, install the following packages
pip3 install rospkg catkin_pkg
and build the catkin workspace with python3 (make sure you have source the python3 environment variables using source_openrobots_py35)
catkin_make -DCMAKE_BUILD_TYPE=Release -DPYTHON_EXECUTABLE="`which python3`"

2.4/ Installation of dynamic graph manager

First make sure you have treep installed.
pip2 install --user treep
or
pip3 install --user treep
Clone the repository manager "treep_machines_in_motion". All actions on the repositories can be done via the "treep" executable.
cd
mkdir devel # devel could be also devel_dg or your favorite cat name
cd devel
git clone git@github.com:machines-in-motion/treep_machines_in_motion.git
treep --clone DYNAMIC_GRAPH_MANAGER
cd workspace
If you do not have the ssh key allowing you to download it, please contact Maximilien Naveau (maxim.nosp@m.ilie.nosp@m.n.nav.nosp@m.eau@.nosp@m.gmail.nosp@m..com) or Vincent Berenz (vbere.nosp@m.nz@t.nosp@m.ue.mp.nosp@m.g.de).
Source "ros". Pick your favorite one (On ubuntu16.04 kinetic is the one used):
source /opt/ros/kinetic/setup.bash
Finally run the build process by calling the catkin executable.
catkin_make install

Table of Contents