Skip to content

Unified framework for robot learning built on NVIDIA Isaac Sim - fork for research into Approximate Gradient Alternatives to trad. RL

License

Notifications You must be signed in to change notification settings

fl1ps1de5/IsaacLab_AGA2trl

 
 

Repository files navigation

Before I Begin, - A Warning!!!

The code that is present in this fork/repo, is not production quality code... at least not at the moment. Changes have been made to various files, and the commit messages have not been super thoughtful. Also, there's a lot of bloat - i.e. files that aren't needed for the exact point of my project, and files that I created for my project that I never ended up using.

That is true at least for the current moment in time (Nov 2024).

That being said, the actual functionality of this code (particularly es_trainer_complete.py & agda_run.py) is proven (see bottom of read me), and works as intended. However, I have not documented exactly how the files should/can be used, and the actual functionality of the scripts depends on a workflow that is not setup to be automatic (i.e. some manual operations must be done before the files are usable).

Anyways, looking to fix that at some point, but thought it was worth the warning...

What is this?

This is a fork of IsaacLab. For details on what that is, please check out the link.

Why does it exist?

I created this fork to allow me to use IsaacLab to work on my undergraduate thesis - "A Study Into Approximate Gradient Descent Alternatives to Deep Reinforcement Learning".

How did you use Isaac Lab in your thesis?

Well, because:

  1. I want to brag
  2. It's awesome
  3. I'm lazy

, instead of actually writing in depth about what is going on here, I'll just include an excerpt from my thesis.

image

About

Unified framework for robot learning built on NVIDIA Isaac Sim - fork for research into Approximate Gradient Alternatives to trad. RL

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 97.9%
  • Other 2.1%