This repository contains the code for an interactive imitation learning demo using EAGERx. Specifically, it allows to train a CLIPort model interactively using PARTNR.
The core repository is available here. Full documentation and tutorials (including package creation and contributing) are available here.
You can run the main code as follows:
Prerequisites: Install Poetry and ROS. Requirements for usage with real Franka robot:
sudo apt update && sudo apt install ffmpeg libportaudio2 libasound-dev
git clone git@github.com:eager-dev/eagerx_demo.git
cd eagerx_demo
poetry install
poetry run python scripts/main.py
Code within eagerx_demo.cliport modified from the original repo.
If you are using EAGERx for your scientific publications, please cite:
@article{vanderheijden2024eagerx,
author={van der Heijden, Bas and Luijkx, Jelle and Ferranti, Laura and Kober, Jens and Babuska, Robert},
journal={IEEE Robotics \& Automation Magazine},
title={Engine Agnostic Graph Environments for Robotics (EAGERx): A Graph-Based Framework for Sim2real Robot Learning},
year={2024},
volume={},
number={},
pages={2-15},
keywords={Robots;Engines;Robot sensing systems;Delays;Robot learning;Physics;Codes},
doi={10.1109/MRA.2024.3433172}
}
EAGERx is funded by the OpenDR Horizon 2020 project.