Qixin Liang

I received my PhD from HKU. I am interested in physics simulation, computational mechanics, and physical AI, with a focus on applying these techniques to real-world applications. I have some industry experience in data-driven multi-physics simulation and real2sim2real for robotics.

Email  /  Linkedin  /  Github

Research

Recent publications.

A technical report on coupling high-fidelity simulation and realistic rendering for robotics simulation.

We propose a graph-neural-network-based framework and a cloth motion capture pipeline for real-time prediction of real-world cloth deformation.

We propose a novel computational framework for modeling and simulating origami structures using bilinear solid-shell elements.

We present six thin plate/shell models, all featuring constant bending energy Hessians, which offer significant benefits for implicit simulations using thin plate/shell models. In the source code repository libThinPlateShells, we provide 13 different bending formulations for researchers and practitioners.

We present a computational model in which the membrane model effectively mitigates the sharp creasing artifact in cloth simulation, while incorporating an efficient bending model.

This thesis develops computational models for cloth and origami simulations.

A GPU-oriented implementation for incremental potential contact experiments using Warp.