Hanzhi CHEN - Mobile Robotics Lab, TU Munich & ETH Zurich

Hanzhi CHEN

PhD Student

Mobile Robotics Lab, TU Munich & ETH Zurich

ABOUT ME

I am a PhD student in the Mobile Robotics Lab at TU Munich and ETH Zurich, working with Prof. Stefan Leutenegger and Dr. Helen Oleynikova. I aim to bridge robot learning and human experience, using human videos to help robots learn, adapt, and improve autonomously.

I received my M.Sc. degree in Robotics, Cognition, Intelligence with high distinction from Technical University of Munich. During my master's studies, I worked at CAMP under the supervision of Dr. Fabian Manhardt, Prof. Benjamin Busam, and Prof. Nassir Navab, where I worked on self-supervised depth and pose learning, as well as neural scene representations.

In my free time, I lift and boulder.

SELECTED PUBLICATIONS

(^* and ^† indicate equal contribution and equal advising, respectively.)

	Actron3D: Learning Actionable Neural Functions from Videos for Transferable Robotic Manipulation Anran Zhang^, Hanzhi Chen^, Yannick Burkhardt, Yao Zhong, Johannes Betz, Helen Oleynikova, Stefan Leutenegger ICRA 2026 project page / paper / video / code Actron3D enables robots to acquire transferable 6-DoF manipulation skills from just a few monocular, uncalibrated, RGB-only human videos.
	FrontierNet: Learning Visual Cues to Explore Boyang Sun, Hanzhi Chen, Stefan Leutenegger, Cesar Cadena, Marc Pollefeys, Hermann Blum RA-L 2025 project page / paper / video / code We propose a color-only frontier-based system for efficient exploration, with FrontierNet as a core module to predict frontiers and information gain.
	VidBot: Learning Generalizable 3D Actions from In-the-Wild 2D Human Videos for Zero-Shot Robotic Manipulation Hanzhi Chen, Boyang Sun, Anran Zhang, Marc Pollefeys, Stefan Leutenegger CVPR 2025, Oral at EgoVis Workshop project page / paper / video / code We present a framework to learn 3D affordance from in-the-wild human videos, enabling zero-shot robotic manipulation in real-world environments.
	FuncGrasp: Learning Object-Centric Neural Grasp Functions from Single Annotated Example Object Hanzhi Chen, Binbin Xu, Stefan Leutenegger ICRA 2024 project page / paper / video We propose a framework to infer continuous grasp functions of unseen objects using only one annotated example.
	Anthropomorphic Grasping with Neural Object Shape Completion Hanzhi Chen^, Diego Hidalgo Carvajal^, Gemma Bettelani, Jaesug Jung, Melissa Zavaglia, Laura Busse, Abdeldjallil Naceri, Sami Haddadin^†; Stefan Leutenegger^†, RA-L 2023 project page / paper / video We design an anthropomorphic grasping system capable of manipulating previously unseen objects using single-view visual input.
	TexPose: Neural Texture Learning for Self-Supervised 6D Object Pose Estimation Hanzhi Chen, Fabian Manhardt, Nassir Navab, Benjamin Busam CVPR 2023 project page / paper / video / code We re-formulate self-supervised 6D object pose estimation as two sub-optimization problems on texture learning and pose learning.
	Attention meets Geometry: Geometry Guided Spatial-Temporal Attention for Consistent Self-Supervised Monocular Depth Estimation Hanzhi Chen^, Daoyi Gao^, Patrick Ruhkamp, Nassir Navab, Benjamin Busam 3DV 2021 project page / paper / video / code We propose a self-supervised monocular depth estimator providing temporally coherent reconstruction.