Table of Contents
Fetching ...

Actuators À La Mode: Modal Actuations for Soft Body Locomotion

Otman Benchekroun, Kaixiang Xie, Hsueh-Ti Derek Liu, Eitan Grinspun, Sheldon Andrews, Victor Zordan

TL;DR

This work addresses locomotion for highly deformable soft-body characters by introducing a modal actuation subspace derived from the natural vibration modes of the geometry and coupling it to a reduced-order soft-body simulation. The method reduces the control and simulation complexity via a Skinning Eigenmode-based spatial reduction and a clustered rotation approach, enabling fast, mesh-resolution–independent optimization of locomotion with CMA-ES. Key contributions include the construction of a low-dimensional actuation space, a reduced-energy formulation with angular-momentum-preserving actuation, a local-global solver in reduced space, and extensive demonstrations across diverse high-resolution geometries, showing substantial speedups and flexible motion style control. The approach broadens the range of animatable deformable characters beyond rigid skeletons, with practical implications for real-time design workflows and potential integration with learning-based controllers.

Abstract

Traditional character animation specializes in characters with a rigidly articulated skeleton and a bipedal/quadripedal morphology. This assumption simplifies many aspects for designing physically based animations, like locomotion, but comes with the price of excluding characters of arbitrary deformable geometries. To remedy this, our framework makes use of a spatio-temporal actuation subspace built off of the natural vibration modes of the character geometry. The resulting actuation is coupled to a reduced fast soft body simulation, allowing us to formulate a locomotion optimization problem that is tractable for a wide variety of high resolution deformable characters.

Actuators À La Mode: Modal Actuations for Soft Body Locomotion

TL;DR

This work addresses locomotion for highly deformable soft-body characters by introducing a modal actuation subspace derived from the natural vibration modes of the geometry and coupling it to a reduced-order soft-body simulation. The method reduces the control and simulation complexity via a Skinning Eigenmode-based spatial reduction and a clustered rotation approach, enabling fast, mesh-resolution–independent optimization of locomotion with CMA-ES. Key contributions include the construction of a low-dimensional actuation space, a reduced-energy formulation with angular-momentum-preserving actuation, a local-global solver in reduced space, and extensive demonstrations across diverse high-resolution geometries, showing substantial speedups and flexible motion style control. The approach broadens the range of animatable deformable characters beyond rigid skeletons, with practical implications for real-time design workflows and potential integration with learning-based controllers.

Abstract

Traditional character animation specializes in characters with a rigidly articulated skeleton and a bipedal/quadripedal morphology. This assumption simplifies many aspects for designing physically based animations, like locomotion, but comes with the price of excluding characters of arbitrary deformable geometries. To remedy this, our framework makes use of a spatio-temporal actuation subspace built off of the natural vibration modes of the character geometry. The resulting actuation is coupled to a reduced fast soft body simulation, allowing us to formulate a locomotion optimization problem that is tractable for a wide variety of high resolution deformable characters.
Paper Structure (36 sections, 63 equations, 14 figures, 1 table)

This paper contains 36 sections, 63 equations, 14 figures, 1 table.

Figures (14)

  • Figure 1: Traditional animation techniques an underlying skeleton structure, which is actuated via joint torques. The resulting actuation exhibits kinky, piecewise rigid deformation. Our actuation properly models the octopus tentacle as a deformable body, allowing for a naturally smooth deformation.
  • Figure 2: Our plasticity based actuation energy conserves angular momentum, while taking on the expected target shape. The force-based actuation from liang2023learningreducedordersoftrobotcontroller does not conserve angular momentum, creating supernatural rotational motion upon actuation .
  • Figure 3: The first 5 non-rigid elastic vibration modes of a seal, corresponding to reasonable low-energy motions one could expect to see from a seal.
  • Figure 4: Decreasing the number of actuation clusters results in a more globally actuated shape, accelerating the speed at which the simulation matches the target actuation signal.
  • Figure 5: Varying the mesh resolution for the Arc de Triomphe mesh leads to a very similar locomotion style, with negligible extra optimization time.
  • ...and 9 more figures