Zhao-Heng Yin 

Alex Zhao-Heng Yin

Ph.D. student at UC Berkeley

zhaohengyin@berkeley.edu


My Image
◤ Biography

I am a first-year Computer Science Ph.D. student at UC Berkeley. I work on machine learning and robotics at the Berkeley Artificial Intelligence Research Lab (BAIR). I am advised by Prof. Pieter Abbeel.

Previously, I obtained my M.Phil. degree in Electronic and Computer Engineering at HKUST in 2023, advised by Prof. Qifeng Chen. Before that, I obtained B.Sc. degree in Computer Science and Mathematics advised by Prof. Wu-Jun Li at Nanjing University in 2021. During M.Phil. research I was also fortunate to work with Prof. Yang Gao and Prof. Xiaolong Wang.

Here is my CV/Resume.

◤ Research

I am interested in understanding how an intelligent, embodied agent can acquire, store, and use knowledge of its surrounding world. Before my Ph.D. journey, I worked on reinforcement learning and imitation learning. I proposed imitation learning algorithms that can leverage cross domain expert data, learn more sample efficiently with MCTS planning, and generalize better with relational inductive bias. I also studied the application of robot learning algorithms on real robotics problem such as dexterous manipulation.

◤ Selected Publications

My recent representative works are marked with ★.

★ Rotating without Seeing: Towards In-hand Dexterity through Touch
Zhao-Heng Yin*, Binghao Huang*, Yuzhe Qin, Qifeng Chen, Xiaolong Wang
RSS 2023    PDF    Project
Dexterous robotic hand with touch-only control.
POINT
★ Spatial Generalization of Visual Imitation Learning with Position-Invariant Regularization
Zhao-Heng Yin, Yang Gao, Qifeng Chen
CoRL 2022 Workshop (Long Paper)   PDF   Code
"Aligning Robot Representation with Humans"
EI
★ Planning for Sample Efficient Imitation Learning
Zhao-Heng Yin, Weirui Ye, Qifeng Chen, Yang Gao
NeurIPS 2022   PDF   Code
We propose EfficientZero-Continuous and EfficientImitate for sample efficient robot learning.
CDIL
★ Cross Domain Robot Imitation with Invariant Representation
Zhao-Heng Yin, Lingfeng Sun, Hengbo Ma, Masayoshi Tomizuka, Wu-Jun Li
ICRA 2022   PDF   Code
DEXPC
DexPoint: Generalizable Point Cloud Reinforcement Learning for Sim-to-Real Dexterous Manipulation
Yuzhe Qin*, Binghao Huang*, Zhao-Heng Yin, Hao Su, Xiaolong Wang
CoRL 2022   PDF
RGAN
Diverse Critical Interaction Generation for Planning and Planner Evaluation
Zhao-Heng Yin*, Lingfeng Sun*, Liting Sun, Masayoshi Tomizuka, Wei Zhan
IROS 2021   PDF
TOMA
TOMA: Topological Map Abstraction for Reinforcement Learning
Zhao-Heng Yin, Wu-Jun Li
arXiv 2020   PDF
◤ Miscellaneous

Some fun projects I participated in the past.

PALM
Bio-inspired* Direct-Drive Palmar Manipulation with Visual Imitation Learning
with Ka Hei Mak and Jungwon Seo (2021)
* guess where the inspiration comes from :P

My Chinese Name (surname first): 殷 兆恒.