Inverse Reinforcement Learning

[eirik-thorp]

Hi!

I am wondering if there is any plan to add any Inverse Reinforcement Learning algorithms in the near future?

From what I have experienced going through numerous code implementation, it seems that there is a dire need for some type of standard way of collecting and handling so-called expert trajectories. Something along the lines of the great Meta-RL implementations in torch.

If there is anyone who has implemented some Inverse-RL algorithms on top of garage, please let me know!

[ryanjulian]

I think @krzentner is most likely to have the answer to this one