DiffCloth: Differentiable Cloth Simulation with Dry Frictional Contact

What is it about?

Cloth simulation has wide applications in computer animation, garment design, and robot-assisted dressing. This work presents a fast differentiable cloth simulator whose gradient information unlocks gradient-based optimization methods for solving inverse cloth-related problems. We demonstrate the efficacy of our simulator in system identification, trajectory optimization for assisted dressing, closed-loop control, inverse design, and real-to-sim transfer.

Featured Image

Photo by Andy Kelly on Unsplash

Photo by Andy Kelly on Unsplash

Why is it important?

Our work contributes the following: (1) We present a novel differentiable cloth simulator with dry frictional contact and an iterative solver for speeding up its gradient computation. (2) We evaluate the source of non-differentiability in the dry frictional contact model and discuss the usefulness of gradients in differentiable, contact-rich cloth simulation. (3) We show the efficacy of our simulator in various applications, including system identification, trajectory optimization for assisted dressing, closed-loop control, inverse design, and real-to-sim transfer.