IEEE Robotics & Automation Magazine - September 2023 - 164

YO U N G PRO F E SS I ONA L S
Advancing Wearable Robotics for Shaping
the Human Musculoskeletal System
By Massimo Sartori
Movement is critical for human wellbeing.
Physical inactivity is the fourth
leading cause of global mortality. Therefore,
developing robotic technologies
that can preserve our ability to move as
we age or restore it following an injury
is a key necessity. Over the past decade,
advances in wearable robotics have led to
exoskeletons and exosuits that can deliver
assistive torques to lower limb joints,
which have resulted in improvements in
locomotion efficiency and reduced neuromechanical
effort. Recent advances in
electronics, actuation, and form factor
are enabling robots to become slimmer,
softer, and lighter than ever, opening the
way to technologies that can be worn as
a " second skin. "
Over the next decade, exoskeletons
and exosuits have the potential to become
" chronic " wearable devices, supporting
users across various domains, including
industry, recreation, and rehabilitation.
With prolonged use, however,
fundamentally new questions will arise:
How will human-robot physical interaction
evolve over large time scales, such
as weeks, months, or even years? Can
exosuits be used to induce structural
changes in musculoskeletal tissue properties
in the long term, such as changes
in muscle strength, tendon stiffness, and
skeletal bone density? How should these
robots be controlled to prevent for tissue
maladaptation, or to restore physiological
properties in damaged tissues?
Digital Object Identifier 10.1109/MRA.2023.3293338
Date of current version: 11 September 2023
164
To address these questions, we must
bridge critical knowledge gaps concerning
the short-term response and
long-term adaptation of the human musculoskeletal
system to robot-induced
stimuli, such as mechanical stress and
strain. When we connect our body to
wearable robots, we expose biological
tissues to mechanical loads. Biological
tissues continually remodel in response
to the cyclic mechanical loading experienced.
Skeletal muscles, tendons, and
bones develop or heal based on optimal
mechanical strains or loads. However,
excessive or insufficient stimuli can
lead to tissue damage or atrophy. Current
robots interact with the human body
with no feedback on how musculoskeletal
tissues adapt their material properties
over prolonged use.
Filling this gap is vital for the emergen -
ce of radically new technologies capable
of effectively controlling mechanical
stimuli applied to the musculoskeletal
system by utilizing estimates of adaptations
in bones, tendons, and muscles.
To achieve this, we need to be able to
establish an interface with the musculoskeletal
system to understand how
skeletal tissues respond to interactive
robots in vivo. Subsequently, we need to
create predictive numerical formulations
to estimate how skeletal tissues remodel
over time in response to robot-induced
mechanical stimuli. Finally, we need to
integrate predictive formulation within
closed-loop schemes to control for tissue
function and adaptation with sufficient
precision to induce targeted positive
changes in the future.
IEEE ROBOTICS & AUTOMATION MAGAZINE SEPTEMBER 2023
In this context, we can employ noninvasive
or minimally invasive wearable
sensors, including high-density electromyogram
electrodes and ultrasound
transducers. These sensors enable measuring
muscle bioelectrical and kinematic
activity with high spatiotemporal resolution.
New portable and wearable sensors
are needed for measuring muscle
contraction in tasks like locomotion or
rehabilitation. Stretchable electronics,
including printed tattoo-like options,
and textile electrodes in smart clothing
offer promising solutions for stable
electrode-to-skin contact and prolonged
use. Thin-film transducers in wearable
garments can record ultrasonography
data without restricting movement.
By integrating these measurements
into multiscale mechanistic numerical
models of the musculoskeletal system,
we can accurately estimate the mechanics
of biological joints, tendons, and muscles
during a large repertoire of movements as
well as in motor tasks involving human-
robot interaction. These multiscale models
could be complemented with statistical
modeling and machine learning to predict
tissue remodeling as a function of
the estimated tissue mechanics across
different timescales. When incorporated
into real-time model predictive
control schemes, this framework could
enable controlled tissue-robot interaction.
This three-pronged approach
aims to merge multimodal, noninvasive
acquisition of biological signals, multiscale
neuromuscular modeling, and
nonlinear optimal robotic control theory
into an integrative framework.
1070-9932/23©2023IEEE
https://orcid.org/0000-0003-0930-6535
IEEE Robotics & Automation Magazine - September 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2023
