Bo-Ruei (Ray) Huang

I'm applying for Ph.D. programs (EE/CS/Robotics) this cycle (2025 Fall)!

I'm a senior undergrad double major in electrical engineering and computer science at National Taiwan University, advised by Shao-Hua Sun as part of NTU Robot Learning Lab.

Currently, I'm visiting MIT CSAIL as a visiting student, working with Jiayuan Mao, Xiaolin Fang, Leslie Pack Kaelbling, and Joshua B. Tenenbaum. I spent the summer of 2023 at Caltech GPS working with Yuk L. Yung.

Email (b09901171 [AT] ntu.edu.tw) / CV / Google Scholar / Twitter / GitHub / LinkedIn

News

[Oct 2024] Our paper Diffusion Imitation from Observation is accepted by NeurIPS 2024. See you in Vancouver!
[Aug 2024] I'm fortunate to visit MIT CSAIL hosted by Jiayuan Mao in Joshua B. Tenenbaum's and Leslie Pack Kaelbling's group.
[Jun 2024] My personal website is now live. Stay tuned for more content!

Research

The goal of my research is to make robots learn to help humans in their daily lives, especially in long-horizon and open-world tasks. I am interested in reinforcement learning, imitation learning, planning, and representation learning.

Publications

	Keypoint Abstraction using Large Models for Object-Relative Imitation Learning Xiaolin Fang^, Bo-Ruei Huang^, Jiayuan Mao^, Jasmine Shone, Joshua B. Tenenbaum, Tomás Lozano-Pérez, Leslie Pack Kaelbling ( equal contribution) Workshop on Language and Robot Learning @ CoRL, 2024 (Best Paper) project page / arXiv / bibtex KALM uses VLMs to automatically generate task-relevant keypoints for better generalization in robot manipulation tasks.
	Diffusion Imitation from Observation Bo-Ruei Huang, Chun-Kai Yang, Chun-Mao Lai, Dai-Jie Wu, Shao-Hua Sun NeurIPS, 2024 project page / arXiv / bibtex DIFO is a learning from demonstration algorithm that integrates diffusion models to model state transitions and provide robust rewards to improve policy learning without action labels.
	Improving XCO2 Precision in OCO-2/3 Retrievals through Machine Learning-Enabled Extraction of Volcanic Aerosol Information from L1B Spectra Bo-Ruei Huang, Sihe Chen, Vijay Natraj, Zhao-Cheng Zeng, Yangcheng Luo, Yuk L. Yung AGU, 2023 poster / bibtex An improved machine learning algorithm refines aerosol data from OCO satellite spectra to enhance CO2 retrieval accuracy and advance climate impact models of volcanic aerosols.

Projects

Learning Robotic Tasks with Object-Centric Value-Implicit Pre-Training
Bo-Ruei Huang, Tung-Yu Wu, Po-Jung Chou, Yun-Rong Du
RL Course Final Project, 2023 (Best Project Award, Top 1/30)
report

Utilized Temporal Cycle-Consistency (TCC) to align robot arm and object features, making the Value-Implicit Pre-training (VIP) network object-centric, thereby enhancing its robustness across novel tasks and embodiments.

Template