|
Tianhai Liang | 梁天海 I am Tianhai Liang, a Master student in Artificial Intelligence & Robotics at Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University under the supervision of Prof. Huazhe Xu. I obtained a Bachelor's degree in Automation from Harbin Institute of Technology, Shenzhen, and a minor Bachelor's degree in Computer Science as an outstanding graduate. Both my graduation projects in the Automation major and the Computer Science minor received Best Bachelor Thesis Awards, making me the first person in HIT's history to receive dual Best Bachelor Thesis Awards. During my undergraduate study, I am fortunate to be mentored by Prof. Haoyao Chen and Prof. Yongyong Chen. My research interests mainly include reinforcement learning🖥️, Computer haptics🖐️ and robotics🤖. I’m particularly interested in the integration of reinforcement learning algorithms with actual robotic hardware🦾. Apart from this, I am also particularly interested in medicine🩺 and electronic circuits🔌. |
|
Publications |
|
MoE-DP: An MoE-Enhanced Diffusion Policy for Robust Long-Horizon Robotic Manipulation with Skill Decomposition and Failure Recovery
Baiye Cheng*, Tianhai Liang*, Suning Huang, Maanping Shao, Feihong Zhang, Botian Xu, Zhengrong Xue, Huazhe Xu, arXiv, 2025 project page / arXiv We present MoE-DP, a diffusion-policy framework with a Mixture-of-Experts layer that decomposes visuomotor skills into interpretable subtasks and greatly boosts robustness. It improves disturbed success rates by 36% across long-horizon tasks and transfers effectively to real robots. The expert structure also enables subtask reordering without retraining. |
|
MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning
Suning Huang*, Zheyu Zhang*, Tianhai Liang, Yihan Xu, Zhehao Kou, Chenhao Lu, Guowei Xu, Zhengrong Xue, Huazhe Xu, International Conference on Machine Learning (ICML), 2025 project page / arXiv A leading model-free visual RL algorithm which achieves state-of-the-art performances in learning efficiency and performance on various tasks. We can even directly train it on the real robot! |
|
Catch It! Learning to Catch in Flight with Mobile Dexterous Hands
Yuanhang Zhang*, Tianhai Liang*, Zhenyang Chen, Yanjie Ze, Huazhe Xu, IEEE International Conference on Robotics and Automation (ICRA), 2025 CoRL 2024 Workshop LFDM(Best Paper Nominations), 2024 project page / arXiv We build a mobile manipulator with a dexterous hand, and leverage reinforcement learning to train a whole-body control policy for the robot to catch diverse objects randomly thrown by humans. |
|
Data Completion-guided Unified Graph Learning for Incomplete Multi-View Clustering
Tianhai Liang*, Qiangqiang Shen*, Shuqin Wang, Yongyong Chen, Guokai Zhang, Junxin Chen, ACM Transactions on Knowledge Discovery from Data (TKDD), 2024 paper We propose one novel IMVC method named Data Completion-guided Unified Graph Learning (DCUGL), which could complete the data of missing views and fuse multiple learned view-specific similarity matrices into one unified graph. |
|
RiEMann: Near Real-Time SE(3)-Equivariant Robot Manipulation without Point Cloud Segmentation
Chongkai Gao, Zhengrong Xue, Shuying Deng, Tianhai Liang, Siqi Yang, Lin Shao, Huazhe Xu Conference on Robot Learning (CoRL), 2024 project page / arXiv We present RiEMann, an end-to-end near Real-time SE(3)-Equivariant Robot Manipulation imitation learning framework from scene point cloud input. Compared to previous methods that rely on descriptor field matching, RiEMann directly predicts the target poses of objects for manipulation without any object segmentation. |
|
ClothesNet: An Information-Rich 3D Garment Model Repository with Simulated Clothes Environment
Bingyang Zhou, Haoyu Zhou*, Tianhai Liang*, Qiaojun Yu, Siheng Zhao, Yuwei Zeng, Jun Lv, Siyuan Luo, Qiancai Wang, Xinyuan Yu, Haonan Chen, Cewu Lu, Lin Shao IEEE International Conference on Computer Vision (ICCV), 2023 project page / arXiv We present ClothesNet: a large-scale dataset of 3D clothes objects with information-rich annotations. Our dataset consists of around 4400 models covering 11 categories annotated with clothes features, boundary lines, and key points. |
Experience |
 |
Tsinghua University, China
2024.07 - Present Master Student in Computer Science Advisor: Prof. Huazhe Xu |
 |
Shanghai Qi Zhi Institute, China
2024.06 - Present Intern |
 |
Harbin Institute of Technology, Shenzhen, China
2020.09 - 2024.06 B.Eng. in Automation with a Minor in Computer Science |
Selected Awards and Honors |
|
|
Academic Performance |
|
|
Personal Interests |
|
|
|
The source code is stolen from Jon Barron. What a concise tempelate! |
