Tianshu Wu

I am a Ph.D. student at Peking University, advised by Prof. Hao Dong. Before this, I earned my Bachelor's degree from the Harbin Institute of Technology.

My current research is centered on humanoid robot learning, with a specific focus on loco-manipulation and humanoid-object interaction. I am dedicated to developing general, generalizable, and scalable learning methods, aiming to empower humanoid robots to execute a wide variety of real-world tasks with human-like capability.

Email  /  GitHub  /  Scholar

profile photo

News

Publications
SUGARv2 real-world humanoid loco-manipulation demonstrations SUGARv2: Learning Diverse Visual Humanoid Loco-Manipulation Skills from Synthesized Interaction References
Tianshu Wu, Tianxiang Gui, Yue Chen, Tingxu Liu, Ziye Peng, Guangjie Yang, Xiangqi Kong, Ye Shi, Jingya Wang, Hao Dong
arXiv, 2026
paper / project

We present SUGARv2, a unified framework that synthesizes diverse interaction references and learns visual humanoid loco-manipulation skills through a shared training pipeline, enabling sim-to-real transfer across five tasks.

🍬SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework
Tianshu Wu*, Xiangqi Kong*, Yue Chen*, Qize Yu, Hang Ye, Jia Li, Yizhou Wang, Hao Dong
Advances in Neural Information Processing Systems (NeurIPS), 2026
paper / code / project

We present SUGAR, a data-driven framework that converts diverse human videos into deployable humanoid loco-manipulation skills in a scalable and generalizable manner.

RoboDojo: A Unified Sim-and-Real Benchmark for Comprehensive Evaluation of Generalist Robot Manipulation Policies
Tianxing Chen*§, Yue Chen*§†, Zixuan Li*, et al. [show all authors] [hide full author list]

Tianxing Chen*§, Yue Chen*§†, Zixuan Li*, Junyuan Tang*, Kailun Su*, Haoran Lu*, Weijie Wan*, Baijun Chen*, Songling Liu*, Haowen Yan, Honghao Su, Zhiyang Dou, Kaixuan Wang, Dandan Zhang, Yunze Liu, Yan Qin, Qiwei Liang, Qiwei Wu, Zijian Lin, Wenwei Lin, Yuran Wang, Minghua He, Tianshu Wu, Ruihai Wu, Jingquan Zhou, Kai-Chong Lei, Haibao Yu, Yuanfeng Ji, Weiyang Jin, Guanyu Lin, Xiaofan Li, Qi Xiong, Renjing Xu, Zhongyu Li, Wenhao Chai, Enze Xie, Ziwei Wang, Yao Mu, Hao Dong, Wojciech Matusik, Mingyu Ding†, Wenbo Ding†, Ping Luo†, Masayoshi Tomizuka†

Arxiv, 2026
paper / code / project

We present RoboDojo, a unified sim-and-real benchmark with 42 simulation tasks and 18 real-world tasks for comprehensive evaluation of generalist robot manipulation policies.

Foundation Feature-Driven Online End-Effector Pose Estimation: A Marker-Free and Learning-Free Approach
Tianshu Wu*, Jiyao Zhang*, Shiqian Liang*, Zhengxiao Han, Hao Dong
International Conference on Robotics and Automation (ICRA), 2025
paper / code / project

FEEPE uses foundation features for accurate, real-time end-effector pose estimation without markers or task-specific training.

OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints
Mingjie Pan*, Jiyao Zhang*, Tianshu Wu, Yinghao Zhao, Wenlong Gao, Hao Dong
Conference on Computer Vision and Pattern Recognition (CVPR), 2025, Highlight
paper / project

OmniManip combines VLM-guided planning with object-centric 3D constraints and 6D pose tracking for generalizable robotic manipulation without task-specific training.

CheckManual: A New Challenge and Benchmark for Manual-based Appliance Manipulation
Yuxing Long, Jiyao Zhang, Mingjie Pan, Tianshu Wu, Taewhan Kim, Hao Dong
Conference on Computer Vision and Pattern Recognition (CVPR), 2025, Highlight
paper / code / project

The first benchmark for manual-based appliance manipulation.

Website template comes from Jon Barron .