About

Zhefei Gong

Hi! At present, I am a research assistant at MiLab, advised by Donglin Wang. Previously, I earned my BEng in Computer Science from Tongji University, and I also spent a wonderful semester as an exchange student at EPFL, under the guidance of Alexandre Alahi. Additionally, I was privileged to work as a research intern at IRCP Lab during my time at Tongji.

After an extensive period of exploration, my interest has gravitated towards solving real-world tasks. Currently, I am broadly interested in robot learning, especially in manipulation tasks. My aspiration is to develop general-purpose robots capable of autonomously performing complex, long-term daily tasks, thus transitioning them from laboratories to everyone’s homes. I am dedicated to realizing this vision over the next ten years.

Actively seeking research collaborations and PhD opportunities in the United States 🇺🇸 or Canada 🇨🇦 !

News

[Jul.9th, 2024] Joined MiLab as a research assistant working on Robot Learning. Wish me luck!
[Jul.1st, 2024] Graduated from Tongji University with four years of wonderful memories, and here's my undergrad thesis.

Research

CARP
CARP: Visuomotor Policy Learning via Coarse-to-Fine Autoregressive Prediction Zhefei Gong, Pengxiang Ding, Shangke Lyu, Siteng Huang, Minyang Sun, Wei Zhao, Zhaoxin Fan, Donglin Wang arXiv preprint arXiv:2412.06782 Links: arXiv | website
VLAS
VLAS: Vision-Language-Action Model with Speech Instructions for Customized Robot Manipulation Wei Zhao, Pengxiang Ding, Zhang Min, Zhefei Gong, Shuanghao Bai, Han Zhao, Donglin Wang Under Review Links: TBD

Projects

Manipulation-Consistency
On the Synergy of Structured Visual Representation Learning and Embodied Self-Supervised Learning Zhefei Gong, Alexandre Alahi Semester Project at EPFL Links: Report | Thesis(zh)
Vehicle-ReIdentification
Intelligent park unmanned parking method based on space-time vehicle re-identification Worked with professors and graduate students Completed during internship at IRCP Lab Links: Patent | Patent(zh)

Awards