This docker image does not support graphical interfaces with OpenGL,
because support for this depends on the locally used
The problem is that it looks like the same graphic drivers as the
host is required in order to enable hardware acceleartion.
This image installs graspit in the /usr directory of the image.
Build this image
sudo docker build -t jenniferbuehler/graspit .
Log onto bash of the image
The GRASPIT environment variable is set to
/graspit_home on the local image. You can mount
a folder on your computer to it so that you can use yor own robot files.
You will need to connect your display to the docker image,
because even though the graspit code runs in headless mode,
it internally still requires X for Qt.
mkdir <your-graspit-home> cp -rf <your-graspit-models-dir> <your-graspit-home> xhost + sudo docker run -ti --rm -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix -v <your-graspit-home>:/graspit_home jenniferbuehler/grasp_planning_graspit
You can find the graspit sources on /usr.
Note that no soft links are supported within
<your-graspit-home>, you will have to copy
entire files and directories into it.
This image alone is not very useful by itself at this stage, due to the limited support of graphical interfaces. But you can base other images on this, which access the GraspIt! code in headless mode.
For example, check out the docker image jenniferbuehler/grasp_planning_graspit.