Working on Poppy via enhanced SSH on Windows

Technology
1

I am used to work on Poppy using Spyder in the Python(x,y) distribution on Windows.
But I now work on the ODROID U3 with Ubuntu. To do so, I connect to the ODROID via SSH protocol.
It is very easy to do so on a Linux system using the “ssh -Y”(-Y for X server forwarding for GUI) command line.
But if you use Windows like me, it is more difficult. I have found a configuration which is very easy to use, I share it here (if you have better I will enjoy :slight_smile: )
I use the MobaXterm application.

  1. Once installed, open the application
  2. Click on the first button on the toolbar: “Session”
  3. A windows open, click on the first “SSH” tab
  4. In “Remote host” type the IP adress of the ODROID system
  5. In “Specify username” type Poppy (or odroid if it is the first time you use your ODROID)
  6. In “Port” let it be to 22
  7. The only tab to configure is the first one “Advanced SSH settings”
  8. In this tab activate the “Follow SSH path (experimental)” which is very useful
  9. Click OK button

A terminal will open, type your login and password (you can save the password)
On the left side of the terminal you have a sftp connection to your ODROID to open the files you want with the application you want on Windows
And finally, every time you launch a GUI using Qt or Tkinter, the display will be redirected to windows !!

Here is an example of display.

scrrenshot.jpg1920×1031 336 KB

2

You can open terminal with jupyter also, it is very convenient. No more need to use ssh

3

I discovered and installed jupyter on Poppy for test… It is not trivial.
This tool is very interesting since:

  • it is based on IPython so there are notebooks and so on…
  • the interface is “kernel” oriented with several languages
  • No need of development tool

The issues:

  • It is not supporting git (except if you create a notebook running in bash launching the classical git commands)
  • It is oriented to scientific computation, not real-time robotics
  • I did not find a way to create a dynamic graphical user interface (to tune and visualise control algorithms
  • PyQt4 does not work anymore (it worked before jupyter installation :unamused: ) and I don’t know why…

Therefore, it is a good working environment for code edition but not for real-time robot code execution.

This is a first test, so if you have a better experience in Jupyter, I am interested if you do not agree.

4

You can use interactive widgets to do that. They are not so well documented (why?) but quite powerful. Yet, I agree with you this will not be “real time”.

1 Like
5

Wow, thanks for the link !