Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/real-stanford/umpnet
[RA-L / ICRA 2022] UMPNet: Universal Manipulation Policy Network for Articulated Objects
https://github.com/real-stanford/umpnet
computer-vision robotics simulation
Last synced: 2 days ago
JSON representation
[RA-L / ICRA 2022] UMPNet: Universal Manipulation Policy Network for Articulated Objects
- Host: GitHub
- URL: https://github.com/real-stanford/umpnet
- Owner: real-stanford
- License: mit
- Created: 2022-02-05T04:22:27.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2022-02-16T07:11:56.000Z (almost 3 years ago)
- Last Synced: 2024-01-29T20:44:30.167Z (10 months ago)
- Topics: computer-vision, robotics, simulation
- Language: Python
- Homepage: https://ump-net.cs.columbia.edu/
- Size: 2.22 MB
- Stars: 44
- Watchers: 4
- Forks: 7
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# UMPNet: Universal Manipulation Policy Network for Articulated Objects
[Zhenjia Xu](http://www.zhenjiaxu.com/),
[Zhanpeng He](https://zhanpenghe.github.io/),
[Shuran Song](https://www.cs.columbia.edu/~shurans/)
Columbia University
Robotics and Automation Letters (RA-L) / ICRA 2022
### [Project Page](https://ump-net.cs.columbia.edu/) | [Video](https://youtu.be/KqlvcL9RqKM) | [arXiv](https://arxiv.org/abs/2109.05668)
## Overview
This repo contains the PyTorch implementation for paper "UMPNet: Universal Manipulation Policy Network for Articulated Objects".
## Content
- [Prerequisites](#prerequisites)
- [Data Preparation](#data-preparation)
- [Testing](#testing)
- [Training](#training)
## Prerequisites
We have prepared a conda YAML file which contains all the python dependencies.
```sh
conda env create -f environment.yml
```
## Data Preparation
Prepare object URDF and pretrained model.
- [mobility_dataset](https://ump-net.cs.columbia.edu/download/mobility_dataset.zip): URDF of 12 training and 10 testing object categories.
- [pretrained](https://ump-net.cs.columbia.edu/download/pretrained.zip): pretrained model.
Download, unzip, and organize as follows:
```
/umpnet
/mobility_dataset
/pretrained
...
```
## Testing
### Test with GUI
There are also two modes of testing: exploration and manipulation.
```sh
# Open-ended state exploration
python test_gui.py --mode exploration --category CATEGORY
# Goal conditioned manipulation
python test_gui.py --mode manipulation --category CATEGORY
```
Here `CATEGORY` can be chosen from:
- training categories]: Refrigerator, FoldingChair, Laptop, Stapler, TrashCan, Microwave, Toilet, Window, StorageFurniture, Switch, Kettle, Toy
- [Testing categories]: Box, Phone, Dishwasher, Safe, Oven, WashingMachine, Table, KitchenPot, Bucket, Door
### Quantitative Evaluation
There are also two modes of testing: exploration and manipulation.
```sh
# Open-ended state exploration
python test_quantitative.py --mode exploration
# Goal conditioned manipulation
python test_quantitative.py --mode manipulation
```
By default, it will run quantitative evaluation for each category. You can modify pool_list(L91) to run evaluation for a specific category.
## Training
Hyper-parameters mentioned in paper are provided in default arguments.
```sh
python train.py --exp EXP_NAME
```
Then a directory will be created at `exp/EXP_NAME`, in which checkpoints, visualization, and replay buffer will be stored.
## BibTeX
```
@article{xu2022umpnet,
title={UMPNet: Universal manipulation policy network for articulated objects},
author={Xu, Zhenjia and Zhanpeng, He and Song, Shuran},
journal={IEEE Robotics and Automation Letters},
year={2022},
publisher={IEEE}
}
```
## License
This repository is released under the MIT license. See [LICENSE](LICENSE) for additional details.
## Acknowledgement
- The code for [spherical sampling](spherical_sampling.py) is modified from [area-beamforming](https://github.com/marc1701/area-beamforming).
- The code for [UNet](unet_parts.py) is modified from [Pytorch-UNet](https://github.com/milesial/Pytorch-UNet).