Implementation of "Push Environment" for Franka Robotic Arm

[Yannyehao]

Hi,

I am a very new person in this field, having been working with ORBIT for three months now. Even the initial installation was a significant challenge for me. This has been a very challenging yet rewarding experience. As a milestone in my learning journey, I modified the reward class in the existing Isaac-Lift-Cube-Franka-v0 environment and trained an agent similar to the one described here: Fetch Push. The agent can push a small cube to a randomly generated target on a two-dimensional plane.

Screencast.from.2024-04-24.12_11_10.PM.webm

Honestly, given my limited knowledge, I am not even sure if this can be considered a proper environment. However, if there is even the slightest possibility that it could be useful, I would be very happy to submit a pull request with my code.

I understand that for many experienced researchers, this might seem like a minor change, but I am very proud of myself.

To other beginners who are also new to ORBIT and robotic learning, I want to say, if you encounter difficulties, please don't give up. The truth is, this is not easy at all, and you are doing great.