Online Adaptation for Enhancing Imitation Learning Policies

This repo contains the minimal requirements and files to reproduce the experiments conducted in the related paper.

How to install

Simply clone this repo and run pip install -r requirements.txt

How to use

First, collect some trajectories using play.py. The script requires you to specify a save path for the recorded trajectories. To train an agent, use the related agent file. For example, to train a GAIL agent, use train_gail.py. To test an agent, run the related agent file. As an example, if you want to test BC, use test_bc.py.

Known issues and TODOs

• TODO: Re-organize files