Technology where one robot helps another robot.

Utilization of computing resources

Schematic diagram of Transition and Scaleout pattern, Pattern for utilizing computing resource of ROS application on Kubernetes by RDBOX

Learning in this chapter

1. How to execute the applications using external computing resources such as cloud computing because those applications are difficult to run at SBC (Raspberry Pi) on TurtleBot3 that has few computer resources (CPU, memory, disk, GPU, etc.)
   • In this tutorial, continue to use the application of tutorial 1 "Deploy robot application with Kubernetes" to control multiple robots. Also deploy the remote application (turtlebot3_remote.launch etc.).
   • Up to the 7.3 Load a TurtleBot3 on Rviz of 7.Bringup of TurtleBot3 manual ROBOTIS e-Manual and 8.Basic Operation - 8.2.Teleop-8.2.1.keybord are the scope of this deployment. Please be sure to check in advance.
   • In this tutorial, "Transition and Scaleout pattern" is introduced instead of "Remote PC" in the manual of TurtleBot3 manual ROBOTIS e-Manual.
2. How to use "other rich computing resource", using the Kubernetes function such as "Deploy", "Rolling update", "Rollback", "Scale out / Scale in", practiced in Tutorial 1 "Deploying robot applications with Kubernetes".
3. How to apply the following functions of Kubernetes
   • Persistent volume
   • Container Lifecycle Hooks
4. Easy to use ROS debugging tools (such as rviz and rostopic commands)
   • Run rviz etc. from a GUI based PC on which ROS (kinetic Desktop full) is installed on the RDBOX network.

Application overview

In this chapter, you will use the Kubernetes deployment called Remote Application to practice the usage pattern of Transition and Scaleout pattern in ROS applications in Kubernetes. This section explains the remote application as the ROS application, and also the remote application as the Kubernetes deployment.

Remote application as ROS application

• As a remote application, you mainly run robot state publisher corresponding to roslaunch turtlebot3_bringup turtlebot3_remote.launch and loading URDF file.
  • In addition, as same as the tutorial 1 "Deploy robot applications with Kubernetes", as the control application, you will mainly execute the keyboard teleoperation by roslaunch turtlebot3_teleop turtlebot3_teleop_key.Launch. As a launch application, you mainly perform the launch process equivalent to roslaunch turtlebot3_bringup turtlebot3_robot.launch.
  • As a difference from Tutorial 1 "Deploy robot applications with Kubernetes", rqt_graph allows you to check the topic of joint_states on each namespace.
• Just as in Tutorial 1 "Deploy robot applications with Kubernetes", you adopt multi master which has one roscore for one TurtleBot3.
• And one remote application in TurtleBot3 (=roscore) in the multi master above. This remote application is set up using Kubernetes so that it runs on a different computer from TurtleBot3 (=roscore).
  • For each remote application, specify the corresponding roscore with the environment variable ROS_MASTER_URI. In addition, in the case of TurtleBot3, specify TURTLEBOT3_MODEL according to the model.
• Using the topic of joint_states, you can check and debug various information of the robot with rviz.

Remote application as Kubernetes application

We will explain the remote application as an example of "Schematic diagram of Transition and Scaleout pattern" at the beginning.

• The remote application is Kubernetes Deployment related to turtlebot3_remote.launch.
• The remote application is defined in the remote application's template file (yaml) called rdbox/tutorials/tutorial-2/tb3-remote.yaml.tmpl.
  • As the file name indicates, this yaml file is the template that can be replaced the part of it.
  • In this tutorial, the character string to be replaced is a character string enclosed by . They are "" and "".
• (1) Pod tb3-bringup-xxxxx running on TurtleBot3 uses Kubernetes persistent volume (PV) to refer to the remote application template file rdbox/tutorials/tutorial-2/tb3-remote.yaml.tmpl.
  • Various storage can be used for external storage used as Persistent Volume. In this tutorial, rdbox-master-00 is used as an NFS server.
  • Place the key file ("/etc/kubernetes/admin.conf" at rdbox-k8s-master (192.168.179.2)) on the persistent volume and use the kubectl command in the container on TurtleBot3 for the remote application template file to be deployed.
• According to the information of each TurtleBot3, change the remote application template file rdbox/tutorials/tutorial-2/tb3-remote.yaml.tmpl running on TurtleBot3 which can be referred to from tb3-bringup-xxxxx. And deploy it with kubectl apply command.
  • (2) Execute execute_kubectl-apply_on_each_master_for_tb3.bash using Kubernetes Container Lifecycle Hooks as a trigger.
  • (3) [Transition] By specifying ROS_MASTER_URI, you can run the ROS node of a specific TurtleBot3 on any computer. In this tutorial you will replace ROS_MASTER_URI with sed command and shell script.
    • You can also use various template engines such as JinJa 2. ( Is the replacement pattern string used in JinJa2.)
  • (4) [Expansion] Apply the matching template among the multiple template files placed in the permanent volume according to the information (node label) of each TurtleBot3.
    • Using the Kubernetes node label, since the applied yaml file is different, you can assign arbitrary processing to each TurtleBot3.

Prerequisites for this tutorial

• Tutorial 1 "Deploying Robot Applications with Kubernetes" has been done and you understand its processing.
• Prepare one 'working PC' connected to the wireless LAN access point of RDBOX. Please connect to wireless LAN according to each OS. Also, the working PC must be a GUI-based machine with ROS (kinetic Desktop full) installed. Also, you need to be able to use SSH on working PC.
  • SSID:g-rdbox-1234,a-rdbox-1234
  • Passphrase (common g-rdbox-1234,a-rdbox-1234) :gL7bQKFz

Please download the necessary files in advance.

• Please clone the yaml file describing the deployment to Kubernetes.

git clone https://github.com/rdbox-intec/rdbox.git

• The PC on which this operation is executed is as follows.
  • A PC accessible to the Kubernetes dashboard
    • PC with modern browser (Chrome or Firefox)
  • A PC on which kubectl can run
    • rdbox-k8s-master (192.168.179.2) etc.
• The yaml files used in this tutorial are as follows.

$ cd rdbox/tutorials/tutorial-2 $ ls docker_files

---

Up [ ToC ]
Next [ Setup persistent volumes ]

from Hacker News https://ift.tt/31t4mVD