Completed the setup for Franka Emika Panda Robot

[f4rh4ng]

Hi Matteo,

Thanks for the quick respond. I enjoyed a fair amount of time last year working on robo-gym as a part of my thesis project; However as I was not much familiar with how Github works, it took me some time to commit it here. Thank you for developing this toolkit. :slightly_smiling_face:

I adjusted the code according to your suggestion, sorry for messing up parts of your code. I probably was testing some stuffs back then and just forgot deleting the extra prints and comments. Also I was not sure if I would get a respond for my pull request as there was no new commits in some years, so I thought I first just send a request to see how it goes. :grin:

One more thing. I also have managed to integrate the Panda robot in robo-gym side and got some interesting results. But as I borrowed some codes from some other repositories like D4PG (https://github.com/schatty/oprl/tree/legacy_d4pg), I wasn't sure if making a pull request make sense. But I made a fork which you can find in following link:

https://github.com/f4rh4ng/robo-gym-panda

Please let me know if you would like to merge this too, and if yes which parts would make more sense for merging. :slightly_smiling_face:

Best Farhang

[matteolucchi]

Hello Farhang,

Sorry for the late reply.

I am glad to hear that our toolkit was useful too you :) No worries at all, it happens to everyone. Everything looks good to me so far and I think it is ready to merge. Unfortunately I do not have the rights to merge pull requests anymore since I parted ways with Joanneum Research, but I hope someone will take a look at it soon and approve it. Maybe @jr-b-reiterer ?

Your repository look very interesting, and I think it could be interesting to merge the environment part into the main repository, but before you put any more work into that I would wait and see if we get a reply from someone at JR.

I hope this gets merged soon, but in any case people will still be able to see your pull request and be able to see your addition.

All the best,

Matteo

[jr-b-reiterer]

Hello @f4rh4ng,

thanks a lot for your efforts! As the community surely has guessed in the meantime, since the departure of the original robo-gym team including @matteolucchi, resources to work on robo-gym have been scarce. Extra thanks to him for still taking the time to occasionally answer questions here.

We are currently setting up a project in which we are going to reactivate our own work on robo-gym that is supposed to start within Q2 of 2024. Not set in stone yet, but considered preparatory steps include a set of significant refactorings that are expected to make development and usage more sustainable in the long run. But this may lead to the creation of a set of successor repositories in order not to break setups based on existing versions that can't handle breaking changes from upstream.

Of course we will still have to manage our resources carefully, and as the training and testing with a real robot in this context is non-trivial, I cannot say yet at which stage the Panda will fit with our priorities. But I am sure we will be able to integrate your developments at some point not too far away, also regarding your work in your fork of the robo-gym repo. Possible that it's not going to be by accepting the merge request, but by a more selective process (acknowledging contributions where due). Your d4pg examples might be useful for others, too, but as you indicated yourself, license and/or attribution issues might need to be solved. As you surely have guessed from the pre-existence of Panda-related code in the robot servers repository, there were previous activities in that direction at our place that have not made it to the github repos yet. When we get to the stage in which we want to go through with integrating the Panda, we will also need to compare this with your suggested changes and find the best of both.

Thanks again and stay tuned!

BR, BR