Support for end-to-end deep reinforcement learning

[utilizatori67]

Hello,
For my project I need to use end-to-end DRL with images as inputs, but my understanding is that Isaac Gym does not currently support this, right? Would I be better off using Isaac Sim for DRL at the moment? And could Isaac Sim be used with OpenAI Gym while I wait for Isaac Gym to get more features?

Thank you!

[DDPG7]

@utilizatori67

For my project I need to use end-to-end DRL with images as inputs, but my understanding is that Isaac Gym does not currently support this, right?

Are you saying that IsaacGym can’t input images? IsaacGym can capture images, so that should be possible. I also want a sample that has been confirmed to work with images as input. Without a sample, it’s difficult to achieve. It should be the same for IsaacSim.