Use the panda_autograsp package in python2.7

[rickstaa]

Since using the package in python3 has proven to be quite a challenge (see #72) I fixed the dependencies issue in python2.7. I was able to do this when using ROS Melodic but not ROS kinetic. As ROS kinetic doesn't play nicely with virtual environments for this version we need to add the python dependencies to the package.xml run and build dependencies. For now, I will, therefore, use ROS Melodic together with python2.7. I will, however, port my scripts to python3 when melodic has fully python3 support (See tracking issue). Below you will find the steps for installing and using the panda_Autograsp package with python2.7.

Installation steps

1. Install the requirements listed in the documentation.
2. Create python 2.7 conda environment conda create -n py2_grasp python=2.7.
3. Activate the new environment using conda activate py2_grasp.
4. Install rospkg" "catkin_pkg" "empy" "pyyaml through pip.
5. Install system ROS packages through rosdep install --from-paths src --ignore-src -r -y
6. Install opencv through conda conda install opencv=1.13.1 or use pip and install the required CUDA and Cudnn drivers yourself.
7. Build the panda_autograsp package using catkin build -j4 -DCMAKE_BUILD_TYPE=Release -DFranka_DIR:PATH="$HOME/libfranka/build" -Dfreenect2_DIR="$HOME/freenect2/lib/cmake/freenect2"
8. Run the pip install . command in the main workspace folder (Repository main folder).
9. Source the workspace devel/setup.bash.
10. Download the GQCNN models by running the .qgcnn/scripts/downloads/models/download_models.sh command.
11. Launch the node using the launch panda_autograsp panda_autograsp.launch command.

[issue-label-bot]

Issue Label Bot is not confident enough to auto-label this issue. See dashboard for more details.