Mingdong Wu | 吴铭东

I have been a Ph.D. student since 2021 in the School of Computer Science at PKU, advised by Prof. Hao Dong. I received my bachelor degree in 2021, from Turing Class in PKU.

My research interest lies in robot learning, generative models, reinforcement learning and 3D perception problems. I aim to build intelligent embodied agents that automatically discover and perform tasks.

news

Sep 22, 2023	Three paper gets accepted to NeurIPS 2023 🎉
Sep 12, 2023	One paper gets accepted to BMVC 2023 (Oral) 🎉
Sep 11, 2023	One paper gets accepted to SIGGRAPH Asia 2023 🎉
Apr 18, 2023	One paper gets accepted to RAL 2023 🎉
Feb 28, 2023	One paper gets accepted to CVPR 2023 🎉
Sep 16, 2022	One paper gets accepted to NeurIPS 2022 🎉

Publications

(*) indicates equal contribution

2024

Arxiv

UniDexFPM: Universal Dexterous Functional Pre-grasp Manipulation Via Diffusion Policy

Tianhao Wu*, Yunchong Gan*, Mingdong Wu, Jingbo Cheng, Yaodong Yang, Yixin Zhu, and Hao Dong

Under Review, 2024

Bib PDF Website

@article{wu2024unidexfpm,
  title = {UniDexFPM: Universal Dexterous Functional Pre-grasp Manipulation Via Diffusion Policy},
  author = {Wu*, Tianhao and Gan*, Yunchong and Wu, Mingdong and Cheng, Jingbo and Yang, Yaodong and Zhu, Yixin and Dong, Hao},
  journal = {Under Review},
  year = {2024}
}

2023

Arxiv

Distilling Functional Rearrangement Priors from Large Models

Mingdong Wu*, Yiming Zeng*, Long Yang, Jiyao Zhang, Hao Ding, Hui Cheng, and Hao Dong

Under Review, 2023

Bib PDF Website

@article{zeng2023distilling,
  title = {Distilling Functional Rearrangement Priors from Large Models},
  author = {Wu*, Mingdong and Zeng*, Yiming and Yang, Long and Zhang, Jiyao and Ding, Hao and Cheng, Hui and Dong, Hao},
  journal = {Under Review},
  year = {2023}
}

NeurIPS

GenPose: Generative Category-level Object Pose Estimation via Diffusion Models

Jiyao Zhang*, Mingdong Wu*, and Hao Dong

Thirty-seventh Conference on Neural Information Processing Systems, 2023

Abs Bib PDF Website Code

机器之心

We explore a pure generative approach to tackle the multi-hypothesis issue in 6D Category-level Object Pose Estimation. The key idea is to generate pose candidates using a score-based diffusion model and filter out outliers using an energy-based diffusion model. By aggregating the remaining candidates, we can obtain a robust and high-quality output pose.

@article{zhang2023genpose,
  news = {机器之心},
  news_link = {https://mp.weixin.qq.com/s/RYV_aap9eYtwX_4_Ghr5Vw},
  sota_link = {https://paperswithcode.com/sota/6d-pose-estimation-using-rgbd-on-real275?p=genpose-generative-category-level-object-pose},
  sota_badge = {https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/genpose-generative-category-level-object-pose/6d-pose-estimation-using-rgbd-on-real275},
  star = {https://img.shields.io/github/stars/Jiyao06/GenPose?style=social&amp;label=Code+Stars},
  title = {GenPose: Generative Category-level Object Pose Estimation via Diffusion Models},
  author = {Zhang*, Jiyao and Wu*, Mingdong and Dong, Hao},
  journal = {Thirty-seventh Conference on Neural Information Processing Systems},
  year = {2023}
}

NeurIPS

Learning Score-based Grasping Primitive for Human-assisting Dexterous Grasping

Tianhao Wu*, Mingdong Wu*, Jiyao Zhang, Yunchong Gan, and Hao Dong

Thirty-seventh Conference on Neural Information Processing Systems, 2023

Bib PDF Website Code

新智元

@article{wu2023learning,
  news = {新智元},
  news_link = {https://mp.weixin.qq.com/s/hpzZWMizR8tPSGIvGVjPoA},
  star = {https://img.shields.io/github/stars/tianhaowuhz/human-assisting-dex-grasp?style=social&amp;label=Code+Stars},
  title = {Learning Score-based Grasping Primitive for Human-assisting Dexterous Grasping},
  author = {Wu*, Tianhao and Wu*, Mingdong and Zhang, Jiyao and Gan, Yunchong and Dong, Hao},
  journal = {Thirty-seventh Conference on Neural Information Processing Systems},
  year = {2023}
}

NeurIPS

Find What You Want: Learning Demand-conditioned Object Attribute Space for Demand-driven Navigation

Hongcheng Wang, Andy Guan Hong Chen, Xiaoqi Li, Mingdong Wu, and Hao Dong

Thirty-seventh Conference on Neural Information Processing Systems, 2023

Bib PDF Website Code

机器之心

@article{wang2023find,
  news = {机器之心},
  news_link = {https://mp.weixin.qq.com/s/Sj2q02VkY6HMzHDot6X9_w},
  star = {https://img.shields.io/github/stars/whcpumpkin/Demand-driven-navigation?style=social&amp;label=Code+Stars},
  title = {Find What You Want: Learning Demand-conditioned Object Attribute Space for Demand-driven Navigation},
  author = {Wang, Hongcheng and Chen, Andy Guan Hong and Li, Xiaoqi and Wu, Mingdong and Dong, Hao},
  journal = {Thirty-seventh Conference on Neural Information Processing Systems},
  year = {2023}
}

SIGGRAPH Asia

Learning Gradient Fields for Scalable and Generalizable Irregular Packing

Tianyang Xue*, Mingdong Wu*, Lin Lu, Haoxuan Wang, Hao Dong, and Baoquan Chen

SIGGRAPH Asia, 2023

Bib PDF Website

@article{Xue2023learning,
  title = {Learning Gradient Fields for Scalable and Generalizable Irregular Packing},
  author = {Xue*, Tianyang and Wu*, Mingdong and Lu, Lin and Wang, Haoxuan and Dong, Hao and Chen, Baoquan},
  journal = {SIGGRAPH Asia},
  year = {2023}
}

BMVC
Oral

Score-PA: Score-based 3D Part Assembly

Junfeng Cheng, Mingdong Wu, Ruiyuan Zhang, Guanqi Zhan, Chao Wu, and Hao Dong

British Machine Vision Conference, 2023

Bib PDF Code

@article{cheng2023score,
  oral = {Oral},
  title = {Score-PA: Score-based 3D Part Assembly},
  author = {Cheng, Junfeng and Wu, Mingdong and Zhang, Ruiyuan and Zhan, Guanqi and Wu, Chao and Dong, Hao},
  journal = {British Machine Vision Conference},
  year = {2023}
}

RAL

Learning Semantic-Agnostic and Spatial-Aware Representation for Generalizable Visual-Audio Navigation

Hongcheng Wang, Yuxuan Wang, Fangwei Zhong, Mingdong Wu, Jianwei Zhang, Yizhou Wang, and Hao Dong

IEEE Robotics and Automation Letters, 2023

Bib PDF Website Code

@article{wang2023learning,
  title = {Learning Semantic-Agnostic and Spatial-Aware Representation for Generalizable Visual-Audio Navigation},
  author = {Wang, Hongcheng and Wang, Yuxuan and Zhong, Fangwei and Wu, Mingdong and Zhang, Jianwei and Wang, Yizhou and Dong, Hao},
  journal = {IEEE Robotics and Automation Letters},
  year = {2023},
  publisher = {IEEE}
}

CVPR

GFPose: Learning 3d human pose prior with gradient fields

Hai Ci, Mingdong Wu, Wentao Zhu, Xiaoxuan Ma, Hao Dong, Fangwei Zhong, and Yizhou Wang

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023

Abs Bib PDF Website Code

GFPose is a unified 3D human pose prior model that can be easily used for various applications, e.g., 3D human pose estimation, pose denoising and generation. Our key idea is to estimate the gradient field (a.k.a, score) of the perturbed human pose. We can leverage the gradient to adjust poses to be more plausible and feasible to a task specification.

@inproceedings{ci2023gfpose,
  sota_link = {https://paperswithcode.com/sota/multi-hypotheses-3d-human-pose-estimation-on?p=gfpose-learning-3d-human-pose-prior-with},
  sota_badge = {https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/gfpose-learning-3d-human-pose-prior-with/multi-hypotheses-3d-human-pose-estimation-on},
  star = {https://img.shields.io/github/stars/Embracing/GFPose?style=social&amp;label=Code+Stars},
  title = {GFPose: Learning 3d human pose prior with gradient fields},
  author = {Ci, Hai and Wu, Mingdong and Zhu, Wentao and Ma, Xiaoxuan and Dong, Hao and Zhong, Fangwei and Wang, Yizhou},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages = {4800--4810},
  year = {2023}
}

2022

NeurIPS
TarGF: Learning Target Gradient Field to Rearrange Objects without Explicit Goal Specification

Mingdong Wu*, Fangwei Zhong*, Yulong Xia, and Hao Dong

Advances in Neural Information Processing Systems, 2022

Abs Bib PDF Website Code

We study object rearrangement without explicit goal specification. The agent is given examples from a target distribution and aims at rearranging objects to increase the likelihood of the distribution. Our key idea is to learn a target gradient field that indicates the fastest direction to increase the likelihood from examples via score-matching.
@inproceedings{wu2022targf, title = {Tar{GF}: Learning Target Gradient Field to Rearrange Objects without Explicit Goal Specification}, author = {Wu*, Mingdong and Zhong*, Fangwei and Xia, Yulong and Dong, Hao}, booktitle = {Advances in Neural Information Processing Systems}, editor = {Oh, Alice H. and Agarwal, Alekh and Belgrave, Danielle and Cho, Kyunghyun}, year = {2022}, url = {https://openreview.net/forum?id=Euv1nXN98P3} }

”