Retrofitting Existing 3D Objects with Surface-Conforming Capacitive Sensing

Andjela Ilic, Junpeng Gao, Zhipeng Li, Yijing Jiang, Rachel Schuchert, Manuel Meier, Philipp Herholz, Christian Holz

ACM SIGGRAPH 2026

A computational fabrication pipeline that retrofits existing 3D objects with mutual-capacitance touch sensing conforming to their curved surfaces. I contributed the geometry modeling part of the project, including surface curve sampling and the 3D-to-2D unfolding that turns the sensor layout into fabrication-ready stencils.

Sim-to-Real of Soft Robots with Learned Residual Physics

Junpeng Gao, Mike Y. Michelis, Andrew Spielberg, Robert K. Katzschmann

IEEE Robotics and Automation Letters (RA-L) 2024 Best Paper Award 2024

Our paper on sim-to-real transfer of soft robots using learned residual physics was accepted by RA-L and received the Best Paper Award as one of only five papers for the year 2024 from among more than 1,500 papers published in RA-L during 2024.

[Project Page] [IEEE Xplore] [arXiv]

M2N: Mesh Movement Networks for PDE Solvers

Wenbin Song, Mingrui Zhang, Joseph G Wallwork, Junpeng Gao, Zheng Tian, Fanglei Sun, Matthew Piggott, Junqing Chen, Zuoqiang Shi, Xiang Chen

Advances in Neural Information Processing Systems (NeurIPS) 2022

We present Mesh Movement Networks (M2N), a novel approach for learning mesh movement in PDE solvers. Our method combines neural networks with traditional mesh adaptation techniques to improve computational efficiency and accuracy.

[NeurIPS]