Xincheng He 何欣成

I am an undergraduate student studying Mathematics at Zhiyuan College, Shanghai Jiao Tong University, and I expect to graduate in June 2027.

I began my research at Shanghai Jiao Tong University under the supervision of Prof. Wenqiang Xu and Prof. Cewu Lu, focusing on physical simulation, computer graphics, and robot learning. Currently, I work with Prof. Chenfanfu Jiang and Prof. Yin Yang at the University of California, Los Angeles (UCLA) on Skill2Real, an agentic sim-to-real framework.

CV|Github|Google Scholar|ORCID|Email: xinchenghe@sjtu.edu.cn

Research

My research focuses on building physical intelligence by combining scalable simulation, differentiable physics, and agentic robot learning. In my current Skill2Real project, I develop agent-callable robot interfaces and critic-guided hierarchical skill learning for zero-shot sim-to-real deployment. I am particularly interested in systems that use physical priors and real-world feedback to learn reusable manipulation skills.

* Equal contribution.

RigPI framework connecting real-world interaction data with differentiable simulation

RigPI: Dynamic Parameter Identification of Rigid Body via VLM-Seeded Differentiable Simulation

Xincheng He, Rongrong Zhang, Wei Jiang, Wenqiang Xu

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2026

arXiv|PDF

Illustration of tooth detection, plaque segmentation, and cleaning assessment

Evaluation of Toothbrush Quality Using AI-based Virtual Tooth Plaque Assessment

Junhao Gu*, Xincheng He*, Ning Luo, Weiwu Yan, Ziliang Guo, Yu Qiao

International Conference on Control, Automation and Diagnosis (ICCAD), 2025

IEEE Xplore|DOI

Experience