IEEE Robotics & Automation Magazine - December 2020 - 9

STUDENT'S CORNER

Thoughts on Modeling
By Marwa ElDiwiny

W

ere there any directions
you thought would work
out very well, but empirical
results proved otherwise?
What is an area or direction of research
you think is very promising even though
the community seems to disagree or
does not currently give much attention
to? The answer is " modeling, " but is this
statement still true for soft robots? Do
we understand how soft robots behave
and the most important aspects that
should be considered in their design?
First of all, there is no certain de--
finition of soft robots-they could
have a completely soft or a hybrid
design. It is crystal clear that the field
has limitations; for example, there
i s a tradeoff between mechanical
performance and bandwidth, which
leads to other questions: Do we have a
deep understanding of how materials
behave? What is the missing piece
between Mother Nature and bioinspired
soft robots?
Let me tell you that modeling is chal-
lenging. However, having a physical,
descriptive model that captures the phys-
ics is essential for figuring out the most
significant parameters for designing soft
robots. I give the simple example of a
smart soft actuator (an ionic electroactive
polymer) that simultaneously works as an
actuator and a sensor for stimulating a
passive material (a catheter); see Figure 1.
It turns out that thickness plays a signifi-
cant role in the force produced and the
actuator response time. What are the
modeling scales? The scales are micro,

Digital Object Identifier 10.1109/MRA.2020.3029686
Date of current version: 9 December 2020

meso, and macro. Then the question
becomes: Which modeling scale do we
go for, taking into account that we are
looking for a design recipe and real-time
application? To answer this, the commu-
nity seems to agree on the macro scale (or
continuum scale).
I asked Prof. Carmel Maijdi from
Carnegie Mellon University for his
thoughts about modeling, " For enhanc-
ing the modeling and simulation of soft
robotics, you can depend on datadriven models, " he said, emphasizing
that there is still a need for determinis-
tic modeling. He thinks there are inter-
esting data-driven techniques and
learning methods in a variety of con-
texts. There are diverse tools-includ-
ing the Large-scale Atomic/Molecular
Massively Parallel Simulator and the
finite element method (FEM)-that are
effective for modeling on different

scales. There have been many demon-
strations of how the predictions gener-
ated using such
tools agree with
experimental We have to take
measurements.
a deeper look at
H o w e v e r,
computational designing descriptive
tools are quite
models, taking
slow. But Prof.
Maijdi thinks into account the
that in soft robot-
morphology of the
ics-when we
want to use com- dynamic shape.
putational tech-
niques and then
loop them into controls, path planning,
genetic algorithms, and thousands or
hundreds of thousands of simulations to
come up with a new design-there is a
(continued on page 77)

Displacement

Catheter

Potential B
Actuator B

Connection
Actuator A

Potential A

Figure 1. The design of the ionic electroactive polymer-actuating passive catheter.

DECEMBER 2020

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

9
IEEE Robotics & Automation Magazine - December 2020

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2020
