1

Developed learning and optimization methods for complex fluid dynamics, using differentiable Navier-Stokes simulation to enable joint reasoning over geometry and control in high-dimensional flow systems.

aAphysics-based inverse modeling and reconstruction optimization framework to infer high quality garment geometry and material physical parameters from real 3D captures.

A topology optimization method with an anisotropic and differentiable constitutive model to design Stokes-flow fluidic devices.

A learning-based method that assembles parts together to form joints.

A computational framework to algorithmically design and fabricate quilts with decorative patterns.