IEEE Robotics & Automation Magazine - December 2020 - 10

FROM THE GUEST EDITORS

Design Optimization of Soft Robots
By Surya G. Nurzaman, Liyu Wang, Fumiya Iida, Jeffrey Lipton, Dario Floreano, and Daniela Rus

he past few years have seen
a significant increase in
research activities that focus
on the use of active or passive
compliance in robotic systems. The
emerging field, generally known as soft
robotics, promises robotic systems that
are safer, more resilient to da--mage,
richly sensorized, and more adaptable
than conventional rigid-bodied robots,
emphasizing the importance of soft and
deformable structures. Several unique
aspects can characterize soft robotic
systems, such as their elastic and
deformable bodies, the large number of
degrees of freedom, the viable use of
unconventional material, the involvement
of intrinsic passive mechanical dy--namics,
and the display of proprioception
and exteroception.
The last few years have also seen the
emergence of a modern view of intelligence, known as embodied intelligence
or morphological computation, that
leverages the properties of soft robots.
Embodied intelligence emphasizes the
importance of the codesign, or coevolution, of the brain (controller) and
morphology (shape, size, and materials) of a machine to best fit the environment in which it operates. From the
perspective of this view of intelligence,
optimizing the design of the soft
robot is, therefore, of great importance.
Such design optimization can be led by
machines and/or humans, or it can
take inspiration from the coevolutionary processes of biological systems.
Digital Object Identifier 10.1109/MRA.2020.3028658
Date of current version: 9 December 2020

IEEE ROBOTICS & AUTOMATION MAGAZINE

The goal of this special issue is to
provide an overview of the state of the
art in the design optimization of soft
robots and identify common perspectives and challenges, shared scientific
goals, and high-impact applications.
The accepted articles were carefully
selected for novelty, thoroughness,
and clarity.
In the first article, Liu et al. present
an integrated mobile benthic platform
and a soft manipulator that can be used
for seafloor exploration, marine samples collection, and other underwater
tasks. The benthic platform consists of a
six-legged, crab-inspired robot, and the
manipulator consists of a four-fingered
soft gripper that can perform delicate
grasping. Both systems are designed
and optimized separately to serve as
general-purpose underwater robots
with locomotion and manipulation
capabilities. The benthic platform is
designed to have a payload for instruments, sensors, or a robot arm, and the
soft arm is designed to be mounted on
underwater vehicles. Based on the integrated platform, the authors experimentally employ spatial manipulation with
inverse kinematics specifically for collecting tasks in a natural underwater
environment. They show that the soft
manipulator's workspace can be significantly extended by adding a benthic
legged robot as a mobile base. Furthermore, they find that the designed system can approach objects precisely
and effectively and perform dexterous
grasping tasks, including retrieving
objects from deep apertures in overhang environments.

DECEMBER 2020

In the second article, Chen and Wang
provide a comprehensive review of the
state of the art in the design of soft
robots. They refer to " design optimization " as the broad innovations in any
design aspect that lead to better performance by soft robots, including
algorithms for solving a formulated
mathematical problem related to design.
The survey considers the design optimization of soft robots as a process to
improve material/structure and properties/performance relations. The article
not only covers nearly 100 references in
modeling, simulation, and computational optimization but, more importantly,
also connects them in a much more
detailed manner than previous surveys
have. For example, the article categorizes
geometry, material, and actuation as the
three areas where design variables may
be identified for modeling soft robots,
with the acknowledgment that the
boundaries between those variables may
change. It also describes simulation
methods based on spring-mass models,
finite elements, kinematic geometric
transformation, and a nonsmooth Newton approach. Finally, the authors
provide a high-level comparison of optimization methods, including gradient-based algorithms, evolutionary
computation, and bio-inspired design. In
the end, the survey identifies three main
challenges in the design of soft robots.
1) Present models are usually simple,
restrictive, and customized for a particular class of tasks, providing only
limited insights into design problems.
2) An effective, efficient, and robust
simulation tool that allows rapid

IEEE Robotics & Automation Magazine - December 2020

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