Carview!

CARVIEW

MOTORHOMES

Select Language

HTTP/2 200 server: GitHub.com content-type: text/html; charset=utf-8 last-modified: Fri, 19 Dec 2025 06:10:36 GMT access-control-allow-origin: * strict-transport-security: max-age=31556952 etag: W/"6944ec5c-3f96" expires: Tue, 30 Dec 2025 06:45:42 GMT cache-control: max-age=600 content-encoding: gzip x-proxy-cache: MISS x-github-request-id: 13DD:318CF6:9AA420:ADCD5F:695372BE accept-ranges: bytes age: 0 date: Tue, 30 Dec 2025 06:35:42 GMT via: 1.1 varnish x-served-by: cache-bom-vanm7210081-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1767076543.526433,VS0,VE226 vary: Accept-Encoding x-fastly-request-id: 1649ea3ec04373415d3d87dd385a7f03a66e2307 content-length: 4314 Zihao He | 何子浩

Zihao He 何子浩

I am a senior undergraduate student at Shanghai Jiao Tong University, majoring in Electrical and Computer Engineering. I am currently a visiting research intern at UCSD, advised by Prof. Hao Su. Previously, I was a member of SJTU Machine Vision and Intelligence Group (MVIG), advised by Dr. Hao-Shu Fang, and Prof. Cewu Lu.

My research interests mainly lie in dexterous manipulation, tactile sensing, and cross-embodiment learning with the ultimate goal of building intelligent robots with human-level manipulation dexterity.

Email / Google Scholar / X / Github

News

[Aug. 2025] AirExo-2 is accepted by CoRL 2025.
[Jun. 2025] FoAR is accepted by IROS 2025. See you in Hangzhou!
[Apr. 2025] FoAR is accepted by RA-L.
[Mar. 2025] AirExo-2 is released! Check our website for more details.
[Nov. 2024] FoAR is released! Check our website for more details.

Research

Representative papers are highlighted. * denotes equal contribution. † denotes corresponding author(s).

	Scaling Cross-Embodiment World Models for Dexterous Manipulation Zihao He, Bo Ai, Tongzhou Mu, Yulin Liu, Weikang Wan, Jiawei Fu, Yilun Du, Henrik I. Christensen†, Hao Su† arXiv, 2025 paper Develop a particle-based cross-embodiment world model for dexterous manipulation. The core idea was to represent robot and human hands, as well as objects, as sets of 3D particles and to express actions as particle displacement field, so all embodiments live in a shared particle space. In this space, we use a particle-based graph neural-network (DPI-Net) as the world model. This graph network incorporates strong inductive biases. The learned world model then transferred zero-shot to diverse real robot hands with different degrees of freedom (e.g., a 6-DoF Ability Hand and a 12-DoF XHand), successfully reshaping deformable plasticine into target letters.
	Learning Dexterous Manipulation with Quantized Hand State Ying Feng, Hongjie Fang, Yinong He, Jingjing Chen, Chenxi Wang, Zihao He, Ruonan Liu, Cewu Lu† arXiv*, 2025 paper / code comming soon / project page Propose DQ-RISE, which quantizes hand states to simplify hand motion prediction while preserving essential patterns, and applies a continuous relaxation that allows arm actions to diffuse jointly with these compact hand states. This design enables the policy to learn arm-hand coordination from data while preventing hand actions from overwhelming the action space. Experiments show that DQ-RISE achieves more balanced and efficient learning, paving the way toward structured and generalizable dexterous manipulation.
	AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons Hongjie Fang, Chenxi Wang, Yiming Wang, Jingjing Chen, Shangning Xia, Jun Lv, Zihao He, Xiyan Yi, Yunhan Guo, Xinyu Zhan, Lixin Yang, Weiming Wang, Cewu Lu†, Hao-Shu Fang† CoRL, 2025 (oral) paper / data collection code / policy code / project page Develop AirExo-2, an updated low-cost exoskeleton system for large-scale in-the-wild demonstration collection. By transforming the collected in-the-wild demonstrations into pseudo-robot demonstrations, our system addresses key challenges in utilizing in-the-wild demonstrations for downstream imitation learning in the real world. Propose RISE-2, a generalizable imitation policy that integrates 2D and 3D perceptions, outperforming previous imitation learning policies in both in-domain and out-of-domain tasks, even with limited demonstrations. By leveraging in-the-wild demonstrations collected and transformed by the AirExo-2 system, without the need for additional robot demonstrations, RISE-2 achieves comparable or superior performance to policies trained with teleoperated data, highlighting the potential of AirExo-2 for scalable and generalizable imitation learning.
	FoAR: Force-Aware Reactive Policy for Contact-Rich Robotic Manipulation Zihao He, Hongjie Fang, Jingjing Chen, Hao-Shu Fang†, Cewu Lu† RA-L, 2025 IROS, 2025 paper / code / project page / X Propose FoAR, a force-aware reactive policy that combines high-frequency force/torque sensing with visual inputs to enhance the performance in contact-rich manipulation. Built upon the RISE policy, FoAR incorporates a multimodal feature fusion mechanism guided by a future contact predictor, enabling dynamic adjustment of force/torque data usage between non-contact and contact phases. Its reactive control strategy also allows FoAR to accomplish contact-rich tasks accurately through simple position control.

Selected Awards and Honors

2025: The Sam and Daisy Wu Scholarship (10 winners in JI)
2025: Student Development Scholarship - Sports (5 winners in JI)
2024: John Wu & Jane Sun Excellence Scholarship (10 winners in JI)
2024: Student Development Scholarship - Sports (5 winners in JI)
2024: Fan Hsu-chi Scholarship (15 winners in SJTU)
2023: National Scholarship (Top 0.2% nationwide)
2023: John Wu & Jane Sun Excellence Scholarship (10 winners in JI)
2023: A-level Merit Scholarship (Top 1% SJTU)
2023: Merit Student (Top 5% SJTU)
2022: Silver Medal Winner of University Physics Competition
2021: First Prize in Chinese Physics Olympiad (CPhO), Zhejiang Province

The website is built upon this template.

Original Source | Taken Source