Elasticity Versus Hyperelasticity Considerations in Quasistatic Modeling of a Soft Finger-Like Robotic Appendage for Real-Time Position and Force Estimation

Ali Shiva,Thrishantha Nanayakkara,Helmut Hauser,Ian D Walker,Kaspar Althoefer,Jan Fraś,Helge Würdemann,S.M Hadi Sadati,Ahmad Ataka,Yohan Noh

doi:10.1089/soro.2018.0060

Abstract

Various methods based on hyperelastic assumptions have been developed to address the mathematical complexities of modeling motion and deformation of continuum manipulators. In this study, we propose a quasistatic approach for 3D modeling and real-time simulation of a pneumatically actuated soft continuum robotic appendage to estimate the contact force and overall pose. Our model can incorporate external load at any arbitrary point on the body and deliver positional and force propagation information along the entire backbone. In line with the proposed model, the effectiveness of elasticity versus hyperelasticity assumptions (neo-Hookean and Gent) is investigated and compared. Experiments are carried out with and without external load, and simulations are validated across a range of Young's moduli. Results show best conformity with Hooke's model for limited strains with about 6% average normalized error of position; and a mean absolute error of less than 0.08 N for force applied at the tip and on the body, demonstrating high accuracy in estimating the position and the contact force.

Highlights

Qualities such as dexterity and high deformability in biological appendages like the octopus arm has sparked a research trend which aims to replicate these features using intrinsically soft materials in continuum robotic platforms; with the promise of safely performing delicate tasks 1 2, improving maneuverability in confined or unstructured environments 3, achieving higher dexterity for grasping 4 5 or for motion in dynamic biomimetic systems 6 7 such as submerged locomotion 8 9
OBJECTIVES & CONTRIBUTIONS Following on from our earlier works 32 42, for a pneumatic soft continuum robotic appendage comprising braided extensors, we propose a forward kinematics (FK), quasi-static, discrete variable curvature (VC) model for real-time contact force sensing
To the best of our knowledge, the detailed analysis on positional error was not witnessed in similar research works, yet comparing with some other approaches, our model shows to be the most accurate solution for a two-segment construction of this same manipulator, compared to a polynomial of order three where an average of 6% mean error is observed for the static model in 2D, and 8% in 3D motion using both the Ritz and Ritz-Galerkin method 41

Summary

Introduction

Qualities such as dexterity and high deformability in biological appendages like the octopus arm has sparked a research trend which aims to replicate these features using intrinsically soft materials in continuum robotic platforms; with the promise of safely performing delicate tasks 1 2, improving maneuverability in confined or unstructured environments 3, achieving higher dexterity for grasping 4 5 or for motion in dynamic biomimetic systems 6 7 such as submerged locomotion 8 9 These robots are appealing for investigating morphological computation 10 and embodied intelligence 8, providing a framework for bodily force sensing without the need for additional sensory hardware, in contrast to rigid-link robots 11. The “Priori” stage; to consider: Taking external loading into account vs. no external loads; and, Inertial (dynamic) vs. non-inertial (static/quasi-static, or kinematic) modelling

Objectives

Methods

Findings

Conclusion