Yiwen Hou  |  δΎ―ζ‡Ώζ–‡

Hi, there! I am a Ph.D student at School of Computing, National University of Singapore, advised by Prof. Lin Shao. Before that, I received my Master's degree and my Bachelor's degree from the School of Computer Science and Technology, University of Science and Technology of China, advised by Prof. Feng Wu.

My current research interests mainly lie in reinforcement learning and robotics. Specifically, I am interested in offline reinforcement learning. I am also interested in sim-to-real transfer, and applications in real-world robotic tasks.

profile photo

News

Research
$\mathcal{D(R,O)}$ Grasp: A Unified Representation of Robot and Object Interaction for Cross-Embodiment Dexterous Grasping
Zhenyu Wei*, Zhixuan Xu*, Jingxiang Guo, Yiwen Hou, Chongkai Gao, Zhehao Cai, Jiayu Luo, Lin Shao
In Submission to International Conference on Robotics and Automation (ICRA) 2025
CoRL 2024 MAPoDeL Workshop (Best Robotics Paper Award)
CoRL 2024 LFDM Workshop
Website  /  arXiv  /  Code
TL;DR: Introduce $\mathcal{D(R,O)}$, a novel interaction-centric representation for dexterous grasping tasks that goes beyond traditional robot-centric and object-centric approaches, enabling robust generalization across diverse robotic hands and objects.
Improving Offline Reinforcement Learning with Inaccurate Simulators
Yiwen Hou, Haoyuan Sun, Jinming Ma, Feng Wu,
International Conference on Robotics and Automation (ICRA) 2024
arXiv /  Video
TL;DR: Improve offline reinforcement learning by combining offline data with simulated data, using a GAN to align and reweight the simulated data, achieving better performance in real-world tasks.
Effective Offline Robot Learning with Structured Task Graph
Yiwen Hou*, Jinming Ma*, Haoyuan Sun, Feng Wu,
IEEE Robotics and Automation Letters (RA-L) 2024
IEEE /  Video
TL;DR: Extract the subtasks and build the structured task graph from offline datasets, then augment and relabel the datasets based on the task graph, enhancing long-horizon robot tasks.

This website is designed based on Jon Barron's wetsite. Last Update: Dec, 2024