Deep learning with domain randomization module

Dear Colleagues,

I am trying to replicate the codes available in the module ‘DEEP LEARNING WITH DOMAIN RANDOMIZATION’ on my computer platform using Ubuntu 18 and ROS Melodic, as the aim is to implement it on NVIDIA Jetson as the final piece of hardware. I have two fundamental questions for that

1- Is the developed ROS module compatible with this new platform (Ubuntu 18 and ROS Melodic) or not? If not what sort of changes are required to adapt it with this platform (Ubuntu 18 and ROS Melodic)?

2- If we want to prepare the environment exactly similar to what you are using in the remote computer, what type of setting and installation (including ROS dependencies) are necessary to run everything smoothly on my computer?

We have priority lane subscription and we need the results as an urgent matter due to time limit, so I would be very grateful if I could get the answers to resolve the issues as soon as possible.

Many thanks

Hello @allahyar.montazeri,

1- The Domain Randomization Course has not been tested on Ubuntu 18 + ROS Melodic, so I cannot guarantee that everything will work smoothly there. However, it should be possible to make it work in that setup. The problems that you might find during the process can go from ROS packages that do not exist on ROS Melodic, to incompatibilities with other software used on the course such as Keras or OpenCV.

2- The setup that is used in this Course is Ubuntu 16 + ROS Melodic + Gazebo 7. So, if you use the same setup everything should work in the same way.

If you choose to work on the same setup of the course, you shouldn’t have any problem replicating the contents od the course there.

If you want to try to adapt everything to the ROS Melodic setup, I would suggest you create a ROSject on ROSDS ( Here you can create a new ROSject choosing the Ubuntu 18 + ROS Melodic setup, and test everything there. This way, you will be able to share the ROSject with us so that when can help you in case you find any problems.


Checkout the guideline to migrate to Gazebo 9+. Some functions are deprecated. For instance, gazebo::math is now ignition::math.

This topic was automatically closed after 21 hours. New replies are no longer allowed.