Real2Sim

Abstract

This paper presents a method for optimizing visco-elastic material parameters of a finite element simulation to best approximate the dynamic motion of real-world soft objects. We compute the gradient with respect to the material parameters of a least-squares error objective function using either direct sensitivity analysis or an adjoint state method. We then optimize the material parameters such that the simulated motion matches real-world observations as closely as possible. In this way, we can directly build a useful simulation model that captures the visco-elastic behaviour of the specimen of interest. We demonstrate the effectiveness of our method on various examples such as numerical coarsening, custom-designed objective functions, and of course real-world flexible elastic objects made of foam or 3D printed lattice structures, including a demo application in soft robotics.

Add

“…Recall that for rigid bodies the mass matrix is configuration-dependent, see also our supplements for details. We note that in contrast to previous work [Hahn et al 2019;Zimmermann et al 2019b], we cannot treat time-stepping as an energy minimization problem due to this state dependent nature of the mass matrix. A recent alternative formulation for rigid bodies [Pan and Manocha 2018] could overcome this limitation.…”

“…In this setting, it is very unclear how to apply auto-differentiation, as it would need to act on the precise sequence of numerical operations performed during the solve. Our approach, on the other hand, analytically computes derivatives via sensitivity analysis, similar to recent work in graphics [Hahn et al 2019;Hoshyari et al 2019], which describes differentiable simulators for visco-elastic materials and constrained flexible multi-body dynamics. However, as a significant departure from prior work, frictional contact is at the core of our paper.…”

“…We begin by deriving the mathematical formulation underlying our analytically differentiable simulation model. While this derivation follows general concepts recently described in the literature [Hahn et al 2019;Zimmermann et al 2019b], we note that our quest for a unified simulation framework for multi-body systems composed of arbitrary arrangements of rigid and soft objects demands a somewhat different methodology, as detailed below.…”

“…To model the behaviour of real-world objects, the elastic forces described above must be complemented by internal damping forces. Most viscosity models, such as the ones described in [Hahn et al 2019], define the viscous stress (and consequently the damping force) based on the linear strain rate ( F + F T ). One major drawback of these models is that they are not invariant to rotational motion, and therefore damp out the angular velocity of a deformable object during free flight.…”

Designing actuation systems for animatronic figures via globally optimal discrete search

“…The animatronic designs generated with our method could also be used to drive the motion of such soft skins, although ensuring that the force transmission mechanisms are designed to be sufficiently strong remains an avenue for future work. We would also like to note that techniques developed to model [Geilinger et al 2020;Hahn et al 2019], design [Bern et al 2017;Ma et al 2017;Megaro et al 2017;Tang et al 2020] and control Hoshyari et al 2019] compliant mechanisms and soft robots could also be used in conjunction with the algorithmic methodology we present in order to create increasingly lifelike animatronic characters that are composed of a mix of rigid and deformable materials.…”