Modélisation de robots souples fait à partir de matériaux mesostructurés

Abstract

Soft robotics is a recent field of robotics. It differs from classical rigid robotics by exploiting the compliance of the robot to create motion by deformation, rather than relying on articulations. Thanks to their inherent compliant composition, soft robots are particularly adapted for motion for deploying through unknown environment, grasping fragile objects or notably for human interactions. Soft systems are essentially fabricated with a homogeneous compliant material, like silicone. However, other, more relevant material choices could be made to design theserobots. Recent works started to investigate the use of multimaterials, metamaterials or even smart materials inside soft structures to build systems with programmed compliance. In this work, we make the choice of using a stochastic foam to program soft robots structures with anisotropy using fiber direction on a given geometry. Using this foam, we show that we can create new controllable kinematics without changing the geometry of soft robots. The anisotropic soft robots are modeled using a Finite Element Method (FEM) with homogenised mechanical properties. We present a complete workflow: from model simulation and design to fabrication and interactive control with inverse simulation. Finally, we present a new generic mechanical calibration method and use it to reduce the sim2real discrepancy.