Deploying Sphinx on a remote server

Hey there,

After trying to run Sphinx on Docker and realizing is somewhat tricky to get it up and running, I know I definitely need a full-blown server for enabling video/image output from the drone’s camera.

From what I have scarcely found, I believe that Gazebo instances can be ran in AWS RoboMaker and in TheConstruct’s ROSDE. However, do you know any other cloud services with dedicated GPU where Sphinx can be ran?

notice that it needs to run firmwared, which in some cases is not available due to systemd / apparmor being blocked by sysadmins.

Sphinx is kind of a specialized fork of Gazebo 7 + dedicated plugins that use firmwared to run a simulated drone firmware. I would be very surprised to see off the shelf Gazebo integrated cloud services to support such customization. That would really be awesome though. You would basically need to provide a custom VM image with Sphinx pre-installed + customized start/stop shell scripts to launch sphinx and firmwared + additional advanced port forwarding options.

I think that running Sphinx on a dedicated AWS EC2 GPU (or any other general purpose cloud GPU instance) should be possible. However, to my knowledge no one tried this before so you would be the first to try.



Hey @ndessart,

Thanks for the quick answer! I will definitely try it and get back to you guys with some feedback about it :smiley:.



@espetro We’re also interested in running Sphinx on a remote server, and are looking at the feasibility. Did the dedicated AWS EC2 GPU approach work for you?

Hi espetro,

I am trying to build a docker with olympe and sphinx on docker so it is easier to kick-off but facing some issues:

  1. Sphinx command prompt, I can not get rid of the yes / no question:

apt-get -y install parrot-sphinx

Did you make this work?

Thanks in advanced