Hui Zhang
张 辉
I am currently a PhD student at PDZ, ETH Zurich, supervised by Prof. Mirko Meboldt, and an associated PhD student at ETH AI Center.
My research interests lie in physics-based hand-object interaction, dexterous robotic hand manipulation, and their applications in daily life and industry. I am specifically interested in building robust, generalizable, and highly agile low-level dexterous manipulation skills that are reusable and can serve as the foundation for various manipulation tasks.
Education & Experience
MSc Robotics, Systems & Control
Thesis at Robotic Systems Lab (Prof. Marco Hutter); Semester project at Computational Robotics Lab (Prof. Stelian Coros)
News
Selected Publications
* indicates equal contribution
UniCross: Unified Cross-Skill Dexterous Manipulation Synthesis
A unified framework for various dexterous skills and their long-horizon composition.
arXiv preprint, 2026
Assembling Two Parts in One Hand
Grasp, coordinate, and assemble two parts entirely within one hand.
Conference on Robot Learning (CoRL), 2026
HiFun: a Hierarchical Framework for Efficient Functional Dexterous Manipulation Learning
Efficient and effective real-world RL for functional dexterous manipulation.
Conference on Robot Learning (CoRL), 2026
FTP-1: A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation
Generalist tactile policy trained on data of 3000h & 21 sensors, transferable to new sensors.
Conference on Robot Learning (CoRL), 2026
RobustDexGrasp: Robust Dexterous Grasping of General Objects
Robust and generalizable dexterous grasping (95%+ success rate on 500+ objects.)
Conference on Robot Learning (CoRL), 2025, Spotlight Presentation
FunGrasp: Functional Grasping for Diverse Dexterous Hands
Enabling functional grasping across diverse dexterous hand morphologies from a single image.
Robotics and Automation Letters (RA-L), 2025
International Conference on Intelligent Robots and Systems (IROS), 2025
2nd Workshop on Dexterous Manipulation at CoRL, Best Poster Award, 2025
GraspXL: Generating Grasping Motions for Diverse Objects at Scale
A scalable approach to generate diverse grasping motions for a large variety of objects.
European Conference on Computer Vision (ECCV), 2024
ArtiGrasp: Physically Plausible Synthesis of Bi-Manual Dexterous Grasping and Articulation
Synthesizing physically plausible bi-manual dexterous grasping and articulation motions.
International Conference on 3D Vision (3DV), 2024, Spotlight Presentation
Dexterous Manipulation Workshop at RSS, 2024
Invited Talks
Build Skills for More Than Skills
---From Physics-based Synthesis to Reusable Interfaces
Institute for AI Industry Research (AIR), Tsinghua University
From HOI Synthesis to Dexterous Robot Manipulation
Wuhan University
Department of Computer Science, National University of Singapore (NUS)
Tsinghua Embodied AI Lab (TEA Lab)
Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University
Academic Service
Workshop Organizer
ICRA 2026 Workshop on Manipulation Robustness
Reviewer
ECCV, SIGGRAPH Asia
RA-L, 3DV, ICRA, CVPR
Autonomous Robots, RA-L, 3DV