maskjp / docker-ros-x11

A docker environment with ROS, Gazebo, X11 and Tensorflow

Docker container images with ROS, Gazebo, X11 windows and Tensorflow capability

This repository developed from nvidia/opengl and nvidia/cuda conatiners, combine these two together to create a ROS develope environment in docker

Current Image Build:

• henry2423/ros-x11-ubuntu:melodic : Ubuntu 16.04 with ROS melodic + Gazebo 9

  

• henry2423/ros-x11-ubuntu:kinetic : Ubuntu 16.04 with ROS Kinetic + Gazebo 8

  

Requirement

• Docker and Nvidia-docker(docker nvidia runtime) on the host: Check with NVIDIA/nvidia-docker

• X11 Server install:

  $ apt-get install xauth xorg openbox
  

Usage

• For the first time before runing up the docker, you should setup X11 socket and X11 auth files by runing command:

  XSOCK=/tmp/.X11-unix
  XAUTH=/tmp/.docker.xauth
  touch $XAUTH
  xauth nlist $DISPLAY | sed -e 's/^..../ffff/' | xauth -f $XAUTH nmerge -
  

• Run command with x11 configuration, to set up the display environment:

  nvidia-docker run -it \
    --volume /tmp/.X11-unix:/tmp/.X11-unix:rw \
    --volume /tmp/.docker.xauth:/tmp/.docker.xauth:rw \
    --env "XAUTHORITY=/tmp/.docker.xauth" \
    --env "DISPLAY" \
    --volume /etc/passwd:/etc/passwd \
    --volume /etc/group:/etc/group \
    --user $(id -u):$(id -g) \
    henry2423/ros-x11-ubuntu:kinetic \
    bash
  

• If you want to connect to tensorboard, run command with mapping to local port 6006:

  nvidia-docker run -it \
    -p 6006:6006 \
    --volume /tmp/.X11-unix:/tmp/.X11-unix:rw \
    --volume /tmp/.docker.xauth:/tmp/.docker.xauth:rw \
    --env "XAUTHORITY=/tmp/.docker.xauth" \
    --env "DISPLAY" \
    --volume /etc/passwd:/etc/passwd \
    --volume /etc/group:/etc/group \
    --volume /etc/shadow:/etc/shadow \
    --user $(id -u):$(id -g) \
    henry2423/ros-x11-ubuntu:kinetic \
    bash
  

• Build an image from scratch:

  docker build -t henry2423/ros-x11-ubuntu .
  

Detail Environment setting

1) Connecting jupyter notebook within container

• Run command with mapping to local port 8888 (jupyter protocol) and 8888 (host web access):

  nvidia-docker run -it \
    -p 8888:8888 \
    --volume /tmp/.X11-unix:/tmp/.X11-unix:rw \
    --volume /tmp/.docker.xauth:/tmp/.docker.xauth:rw \
    --env "XAUTHORITY=/tmp/.docker.xauth" \
    --env "DISPLAY" \
    --volume /etc/passwd:/etc/passwd \
    --volume /etc/group:/etc/group \
    --volume /etc/shadow:/etc/shadow \
    --user $(id -u):$(id -g) \
    henry2423/ros-x11-ubuntu:kinetic \
    bash
  

• Check your container assigned IP using $ifconfig within container, then you can start up jupyter notebook in container with following command:

  jupyter notebook --ip={YOUR CONTAINER IP} --port=8888 --allow-root
  

• After start up the jupyter kernel, you can access the notebook from host browser through HTTP service.

  http://localhost:8888/
  

Connect & Control

If the container runs up, you can connect to the container throught the following

• You can open the GUI program directly, such as rviz and gazebo
• Connect to Tensorboard if you do the tensorboard mapping above: http://localhost:6006
• Connect to Jupyter if you do the tensorboard mapping above: http://localhost:8888/
• The username and password in container is same as your host account