Rofunc: The Full Process Python Package for Robot Learning from Demonstration and Robot Manipulation

Repository address: https://github.com/Skylark0924/Rofunc

Rofunc package focuses on the robotic Imitation Learning (IL) and Learning from Demonstration (LfD) fields and provides valuable and convenient python functions for robotics, including demonstration collection, data pre-processing, LfD algorithms, planning, and control methods. We also provide an Isaac Gym-based robot simulator for evaluation. This package aims to advance the field by building a full-process toolkit and validation platform that simplifies and standardizes the process of demonstration data collection, processing, learning, and its deployment on robots.

Installation

Install from PyPI (stable version)

The installation is very easy,

pip install rofunc

and as you'll find later, it's easy to use as well!

import rofunc as rf

Thus, have fun in the robotics world!

Note Several requirements need to be installed before using the package. Please refer to the installation guide for more details.

Install from Source (nightly version)

git clone https://github.com/Skylark0924/Rofunc.git
cd Rofunc

# Create a conda environment
# Python 3.8 is strongly recommended
conda create -n rofunc python=3.8
conda activate rofunc

# Install the requirements and rofunc
pip install -r requirements.txt
pip install .

Note If you want to use functions related to ZED camera, you need to install ZED SDK manually. (We have tried to package it as a .whl file to add it to requirements.txt, unfortunately, the ZED SDK is not very friendly and doesn't support direct installation.)

Documentation

Note Currently, we provide a simple document; please refer to here. A comprehensive one with both English and Chinese versions is built via the readthedoc. We provide a simple but interesting example: learning to play Taichi by learning from human demonstration.

To give you a quick overview of the pipeline of rofunc, we provide an interesting example of learning to play Taichi from human demonstration. You can find it in the Quick start section of the documentation.

The available functions and plans can be found as follows.

Note ✅: Achieved 🔃: Reformatting ⛔: TODO

Data		Learning		P&C		Tools		Simulator
xsens.record	✅	DMP	⛔	LQT	✅	Config	✅	Franka	✅
xsens.export	✅	GMR	✅	LQTBi	✅	robolab.fk	✅	CURI	✅
xsens.visual	✅	TPGMM	✅	LQTFb	✅	robolab.ik	✅	CURIMini	🔃
opti.record	✅	TPGMMBi	✅	LQTCP	✅	robolab.fd	⛔	CURISoftHand	🔃
opti.export	✅	TPGMMBiCoordLQR	✅	LQTCPDMP	✅	robolab.id	⛔	Walker	✅
opti.visual	✅	TPGMR	✅	LQR	✅	robolab.tran	✅	Gluon	🔃
zed.record	✅	TPGMRBi	✅	PoGLQRBi	✅	visualab.dist	✅	Baxter	🔃
zed.export	✅	BCO	🔃	iLQR	🔃	visualab.ellip	✅	Sawyer	🔃
zed.visual	✅	STrans	⛔	iLQRBi	🔃	visualab.traj	✅
emg.record	✅	PPO(SKRL)	✅	iLQRFb	🔃
emg.export	✅	SAC(SKRL)	✅	iLQRCP	🔃
emg.visual	✅	TD3(SKRL)	✅	iLQRDyna	🔃
mmodal.record	⛔	PPO(SB3)	⛔	iLQRObs	🔃
mmodal.export	✅	SAC(SB3)	⛔	MPC	⛔
		TD3(SB3)	⛔	CIO	⛔
		PPO(RLlib)	✅
		SAC(RLlib)	✅
		TD3(RLlib)	✅
		PPO(ElegRL)	✅
		SAC(ElegRL)	✅
		TD3(ElegRL)	✅
		PPO(RofuncRL)	🔃
		SAC(RofuncRL)	⛔
		TD3(RofuncRL)	⛔
		CQL(RofuncRL)	⛔

Star History

Cite

If you use rofunc in a scientific publication, we would appreciate citations to the following paper:

@misc{Rofunc2022,
      author = {Liu, Junjia and Li, Zhihao and Li, Chenzui and Chen, Fei},
      title = {Rofunc: The full process python package for robot learning from demonstration},
      year = {2022},
      publisher = {GitHub},
      journal = {GitHub repository},
      howpublished = {\url{https://github.com/Skylark0924/Rofunc}},
}

The Team

Rofunc is developed and maintained by the CLOVER Lab (Collaborative and Versatile Robots Laboratory), CUHK.

Acknowledge

We would like to acknowledge the following projects:

Learning from Demonstration

1. pbdlib
2. Ray RLlib
3. ElegantRL
4. SKRL

Planning and Control

1. Robotics codes from scratch (RCFS)
2. pddlstream