DiffCloth: Differentiable Cloth Simulation with Dry Frictional Contact
This work addresses the need for efficient gradient computation in cloth simulation for applications in computer animation, garment design, and robotics, representing an incremental improvement over existing methods.
The authors tackled the problem of enabling gradient-based optimization in cloth simulation with dry frictional contact by developing a differentiable simulator, resulting in substantial speedups in applications like system identification and trajectory optimization.
Cloth simulation has wide applications in computer animation, garment design, and robot-assisted dressing. This work presents a differentiable cloth simulator whose additional gradient information facilitates cloth-related applications. Our differentiable simulator extends a state-of-the-art cloth simulator based on Projective Dynamics (PD) and with dry frictional contact. We draw inspiration from previous work to propose a fast and novel method for deriving gradients in PD-based cloth simulation with dry frictional contact. Furthermore, we conduct a comprehensive analysis and evaluation of the usefulness of gradients in contact-rich cloth simulation. Finally, we demonstrate the efficacy of our simulator in a number of downstream applications, including system identification, trajectory optimization for assisted dressing, closed-loop control, inverse design, and real-to-sim transfer. We observe a substantial speedup obtained from using our gradient information in solving most of these applications.