This is an easy trial of the multimaster_fkie package using docker. You can try the package with one PC using docker containers.
You need Ubuntu 14.04LTS PC, and docker is installed. You don't need to install ROS(but who doesn't?).
$ git clone https://github.com/longjie/learn_ros_multimaster.git
$ docker pull longjie/learn_ros_multimaster
The image is large, it may take some time.
$ cd learn_ros_multimaster/script
On multiple terminals, you can run kobuki nodes (kobuki_softnode) like:
$ ./docker_kobuki.sh 1
$ ./docker_kobuki.sh 2
$ ./docker_kobuki.sh 3
$ ./docker_kobuki.sh 4
Note you have run individual four ROS systems. The each container has its own master(roscore) running. Yes, it is multi-master system.
The number after the script is to identify the node and topics. The mulitimaster_fkie syncs the node and topic name as is, so you need to use different namespace for each ROS node and topic. You cannot launch nodes of the same name.
Now you can run another container to run rviz to visualize the robots.
$ ./docker_rviz.sh
You can see four Kobukis are running around a circle with different speed. Again, this is multi-master system, and this rviz container has its own rosmaster(roscore).
If you have UVC camera, you can run another docker container to aquire image. Make sure you can see the device /dev/video0 and get image with camera applications(cheese, qv4l2, etc).
$ ./docker_camera.sh
You can see the image on the rviz's image plugin.
These are just memos.
$ cd learn_ros_multimaster
$ docker build -t longjie/learn_ros_multimaster .