|
I am a second year Master student at Tsinghua University, SIGS, supervised by Prof. Zhi Wang on 3D Vision and Embodied AI. My research interests include Robot Data Generation, Real-to-Sim-to-Real, Generalizable Robot Learning, 3D Visual Reconstruction, and Interactive AI Generation. Email / Google Scholar / Github / |
|
|
My research focuses on computer graphics and computer vision, particularly on 3D reconstruction and simulation techniques that facilitate effective robot learning. |
|
|
|
Tsinghua University, SIGS
M.S. in Data Science and Information Technology 2024.09 – 2027.06 (Expected) |
|
Tencent Robotics X
World Model Research Intern 2026.02 – Present |
|
|
|
Chenghao Gu*, Haolan Kang*, Junchao Lin*, Jinghe Wang, Duo Wu, Shuzhao Xie, Fanding Huang, Junchen Ge, Ziyang Gong, Letian Li, Hongying Zheng, Changwei Lv, Zhi Wang IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026 project page / arXiv / pdf IGen scalably generates realistic visual observations and executable robot actions from open-world images, enabling policies trained purely on synthesized data to match real-world data performance for robotic manipulation. |
|
Chenghao Gu, Zhenzhe Li, Zhengqi Zhang, Yunpeng Bai, Shuzhao Xie, Zhi Wang arXiv preprint, 2024 arXiv / pdf DragScene is an effective drag-style 3D scene editing framework, enabling controllable and view-consistent edits on real-world 3D scenes from a single reference view by combining 2D latent optimization with point-based 3D clues. |
|
Xiaojie Li, Chenghao Gu, Shuzhao Xie, Yunpeng Bai, Weixiang Zhang, Zhi Wang arXiv preprint, 2024 arXiv / pdf / code VisCtrl is a tuning-free method that injects the appearance and structure of a user-specified subject into a target image via iterative self-attention control, enabling consistent personalized editing of images, videos, and 3D scenes with only a single reference image. |
|
Template from Jon Barron. Last updated in July 2025. |