Content uploaded by Salvatore Garofalo
Author content
All content in this area was uploaded by Salvatore Garofalo on Dec 10, 2024
Content may be subject to copyright.
Review Article
Journal of Intelligent Material Systems
and Structures
1–28
ÓThe Author(s) 2024
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1045389X241263878
journals.sagepub.com/home/jim
A critical review of transitioning from
conventional actuators to artificial
muscles in upper-limb rehabilitation
devices
Salvatore Garofalo, Chiara Morano, Michele Perrelli, Leonardo Pagnotta,
Giuseppe Carbone, Domenico Mundo and Luigi Bruno
Abstract
Brain injuries resulting from spinal cord injuries, strokes, or cerebral palsy are among the traumas most capable of com-
promising the motor activities of human limbs, hence the necessity for the development of exoskeletons dedicated to
the rehabilitation of these organs. This review examines the landscape of actuators essential for the design of cutting-
edge upper-limb rehabilitation exoskeletal structures. Beyond merely surveying the current types of actuators available,
the paper aims to provide guidelines for selecting actuators that fit optimally with the objectives of upper-limb rehabilita-
tion. The description starts with a brief discussion on the biomechanics of the upper limbs, focusing on the kinematics of
pivotal joints (wrist, elbow, shoulder). Subsequently, the existing actuators are systematically reviewed, offering detailed
insights into their primary features, operational principles, strengths, weaknesses, and noteworthy applications within
the realm of rehabilitation robotics. After the discussion about the actuators, the paper advances by furnishing valuable
guidelines for actuators’ selection tailored for upper limb rehabilitation. These guidelines discuss crucial factors, such as
the forces required and the natural Range Of Motions (ROMs) of upper limb joints. Finally, the manuscript serves as a
valuable resource for researchers, engineers, and practitioners involved in the development of innovative upper-limb
rehabilitation devices.
Keywords
Actuators, artificial muscles, upper limb rehabilitation, exoskeleton, exosuit
1. Introduction
Brain injuries can be associated with both non-
traumatic events such as spinal cord injury, stroke, or
cerebral palsy, and traumatic events due to wounds to
the body by external agents (de Miguel-Ferna
´ndez
et al., 2023). Non-traumatic brain injuries, specifically,
are among the most common disabling disorders asso-
ciated with impairment of the use of both upper and
lower limbs, which are also capable of generating
important physical and psychological impacts, even in
the performance of the simplest Activities of Daily
Living (ADLs; Cieza et al., 2020; Johnson et al., 2019).
Considering stroke as an example, it is known to date
that it is the second leading cause of death and one of
the leading causes of long-term disability in most coun-
tries of the world. Indeed, data updated to 2022 show
how, every year, there are .12.2 million new cases of
stroke worldwide, 6.5 million of which imply death
(Feigin et al., 2022). Additionally, among the brain
functions most affected by stroke is, unfortunately, the
ability to coordinate the body’s motor activities. This is
why, when a stroke affects any part of the brain that
deals with motor control, severe impairment of one or
more body parts can occur (Balasubramanian, 2015).
Although the clinical picture may differ from patient to
patient, these types of disorders usually generate typical
symptoms such as muscle weakness, paralysis, muscle
spasticity and consequent pain (Canning et al., 2004;
Theadom et al., 2014). Unfortunately, to date, some
types of neurological and neuromuscular disorders
Department of Mechanical, Energy, and Management Engineering,
University of Calabria, Cosenza, Rende, Italy
Corresponding author:
Luigi Bruno, Department of Mechanical, Energy, and Management
Engineering, University of Calabria, Via Bucci 44C, Cosenza, Rende
87036, Italy.
Email: luigi.bruno@unical.it
cannot be completely cured, and as a result, special reha-
bilitation devices are needed to improve the quality of
life. For example, it was estimated that globally, about
2.41 billion people manifested conditions that would
benefit from rehabilitation in 2019 (Cieza et al., 2020).
Nowadays, although many exoskeletons are aug-
mentative (i.e. they are meant to improve the physical
performance of healthy or partially impaired individu-
als through the reduction of muscle fatigue (Rosen and
Ferguson, 2020)), many of them are being developed
and used for purely rehabilitative and assistive purposes
(Ang et al., 2023). Specifically, in the rehabilitation con-
text, exoskeletons are used to provide important sup-
port to the therapist. In fact, from this point of view,
there are many advantages deriving from the use of
these devices. Ease of use is one of the most important
features of this type of device; in fact, the possibility of
wearing an exoskeleton autonomously makes them par-
ticularly convenient and engaging. Another appealing
aspect of an exoskeleton is its capability to enter into
symbiosis with users, responding to their movement
intentions. These potentials minimize assistance from
caregivers and speed up recovery and rehabilitation
time (Gassert and Dietz, 2018; Weber and Stein, 2018).
Similarly, in the assistive context, exoskeletons are
applied for the support of basic movements, such as
acts of drinking and eating. Consequently, their use can
be a useful and valuable tool suitable for the recovery
of motor independence, partial or complete, by patients
affected by neuromuscular disorders (Bardi et al., 2022;
Kapsalyamov et al., 2020).
Among the various disabilities implying neuromus-
cular disorders, those impacting the mobility of the
upper limbs greatly impact the quality of life, as they
adversely affect the ability to perform independently
simple daily activities (Faria-Fortini et al., 2011;
Janssen et al., 2014; Majidi Fard Vatan et al., 2021).
For these conditions, rehabilitation robotics can be
seen as an extremely useful tool. Conventional exoske-
letons for the upper limbs are often characterized by
rigid links, joints and actuators capable of applying
forces suitable for the movement of these body parts
and, at the same time, to respect human limb handling
ranges (Ochieze et al., 2023). However, they also have
weaknesses, such as excessive cost, heaviness, bulkiness
and rigidity, which could limit the natural mobility of
skeletal joints.
In the last decades, intending to overcome these lim-
itations, the so-called soft exoskeletons, also known as
exosuits, are gaining popularity. In fact, by combining
lightness, flexibility and portability, exosuits are being
used in a variety of applications. Due to their constitu-
ent materials (fabrics and soft materials in general),
these types of devices are naturally ergonomic, with sig-
nificant reduction in weight and size. In addition, the
restraint of the natural movement of human joints is
reduced, with significant improvement in comfort and
user confidence (Bardi et al., 2022; Xiloyannis et al.,
2022). Therefore, these features make exosuits particu-
larly attractive, from both the perspective of out-of-
hospital rehabilitation and the possibility of being worn
directly underneath the clothes, which greatly improves
the psychological impact on the users (Faria-Fortini
et al., 2011).
The important advantages discussed above stimu-
lated researchers and industry to tackle the several
challenges still not fully resolved. One of these derives
from the fact that the exosuits act as a kind of external
musculature rather than a true external skeleton;
therefore the user’s skeleton has to absorb part of the
loads introduced by the actuation system, which raises
the important problem of the optimal distribution of
the loads onto the human body (Xiloyannis et al.,
2022). In fact, quite often the anchor points of exo-
suits, due to their inherent softness, may encounter
slippage during actuation operations, affecting force
transmission and biasing the data collected by the sen-
sor system (Bardi et al., 2022).
A significant number of reviews on upper limb exos-
keletal devices can be found in the literature. Ochieze
et al. (2023) carried out a systematic review investiga-
tion on both rigid and soft exoskeletons, focusing par-
ticularly on their design; Majidi Fard Vatan et al.
(2021) analyzed rehabilitative exoskeletal devices with a
particular focus on the shoulder joint; Kapsalyamov
et al. (2020) published a review article to explore the
different solutions available for wearable assistive
robotics that can be used by elderly people during
ADLs. The review article by Huamanchahua et al.
(2021), conversely, evaluates recent developments in
exoskeletons aimed at the rehabilitation of elbow flex-
ion/extension movements.
Likewise, a copious number of reviews are available
regarding actuators. In the work of Greco et al. (2022),
a systematic review of the main types of actuators is
carried out, starting with conventional ones and ending
with the most recent artificial muscles. Similarly,
Grellmann et al. (2022) provided a wide overview of
the most promising technologies for the manufacture
of artificial muscles, with particular emphasis on possi-
ble textile applications. Pan et al. (2022b) in their work
highlighted the state-of-the-art regarding soft actuators,
turning their attention toward their application for
rehabilitation and assistive robotic devices. Something
similar was also done by Chen et al. (2021), whose
review deals with the peculiar aspects and some recent
applications of wearable actuators.
To the best of our knowledge, no actuator review is
yet available to serve as a guide for actuator selection
for a possible upper limb exoskeleton. Consequently,
the purpose of this paper is not only to review the main
types of actuators available today for the design of
rehabilitative upper limbs exoskeletal structures, but
also to provide a guide to select actuators that are most
2Journal of Intelligent Material Systems and Structures 00(0)
conformable to fulfill this objective. Firstly, the biome-
chanics of the upper limbs are briefly discussed, with
particular emphasis on the kinematics of the three most
important joints (wrist, elbow and shoulder). After
that, a systematic review of existing actuators is given,
where detailed information is provided on their main
features, working principles, strengths and weaknesses,
and some typical applications in the field of rehabilita-
tion robotics. Finally, general guidelines are presented
on how to select the most appropriate actuator for the
rehabilitation of upper limb joints, according to the
forces required and natural Range Of Motions
(ROMs).
2. Literature analysis and methodology
The research strategy is schematically described in the
flow chart shown in Figure 1. A series of keywords
were used to define the bibliographic framework. After
several attempts carried out by properly combining dif-
ferent keywords with Boolean AND/OR operators, the
final research string in the first block of Figure 1 was
identified. The most common scientific databases
(Scopus, Web of Science, IEEE Xplore, and PubMed)
yielded a total of 470 documents. After removing all
duplicate documents, and those whose DOI and
authors were missing, a final list of 260 unique docu-
ments was obtained.
Afterward, a statistical analysis was performed on
the results of Scopus and Web of Science, widely recog-
nized as the most influential databases. The analysis
was facilitated through the utilization of Bibliometrix,
an application developed by Aria and Cuccurullo
(2017). Figure 2(a) emphasizes the main keywords in
the context of rehabilitation using exoskeletons, while
Figure 2(b) illustrates the annual scientific production
in the relevant field between 2001 and 2023. It is note-
worthy that the number of related publications showed
a sharp increase starting from 2017 and is still continu-
ing, without any apparent influence from the COVID-
19 pandemic experienced in recent years. The difference
in terms of publications between 2022 and 2023 is obvi-
ous since 2023 has not yet concluded. It is therefore
reasonable to expect a further increase in publications
in the near future.
A part of existing reviews was additionally used like
an auxiliary search database, to broaden the view on
further potentially relevant literature. At times, it was
necessary to apply filters to the obtained results, priori-
tizing the most highly cited and recent references. Once
the state-of-the-art was outlined, proper tables were
established for each type of actuator identified. The lit-
erature analysis continued with the study of the biome-
chanics of the human upper limb, with particular
emphasis on the shoulder, elbow and wrist joints. This
analysis holds significant importance as an erroneous
assessment of the required forces could imply the exclu-
sion of otherwise feasible actuator(s). The literature
review was concluded by examining existing exoskele-
ton designs and assessing how the identified actuators
are applied in practice.
3. Anatomy and biomechanics of the
human upper limb
The main function of the upper limb consists in
enabling different types of tasks, such as reaching
points in the space or manipulating objects. To do this,
the upper limb makes use of a true linkage system,
which uses the hand as an end effector and the wrist,
elbow and shoulder as positioning joints (Pang et al.,
2020; Zhao et al., 2019).
The mathematical modeling of the upper limb move-
ments requires the definition of a reference system.
Despite the three-dimensional nature of these move-
ments, the specific system shown in Figure 3 utilizes
Figure 1. Graph of the selection process of the bibliographic
references.
Garofalo et al. 3
anatomical planes to adopt a 2D mathematical model.
These planes include the sagittal plane (observing the
body from the side), the transverse plane (observing the
body from top to bottom), and the frontal plane (obser-
ving the body from the front). Movements are then
defined from a standard anatomical position which, as
highlighted in Figure 3, involves the feet being together
and the head, eyes and palms facing forward (Forner-
Cordero et al., 2008; Freivalds, 2011).
The above-mentioned anatomical planes allow the
definition of all movements within the musculoskeletal
system of the human body. Specifically, focusing on
upper limb movements, the Degrees Of Freedom
(DOFs) associated with the upper limb and its principal
joints (wrist, elbow and shoulder) can be described.
3.1. Shoulder joint
Due to its great variety of movements, the mechanics
of the shoulder is very complex, and it can be described
as a complex of several joints. To comprehend the true
number of DOFs associated with the scapular girdle, it
is crucial to acknowledge the continuous motion of its
center of rotation. This is why the shoulder is, in some
cases, modeled with five or even six DOFs joint.
However, the main connection of the shoulder is con-
sidered the glenohumeral joint, and it is often modeled
as a ball-and-socket joint, with three DOFs (Blanco
Ortega et al., 2022; Freivalds, 2011; Varghese et al.,
2018). These include:
In the sagittal plane, flexion (raising the arm)
and extension (pushing the arm backward),
Figure 4(a);
In the frontal plane, abduction (lifting the arm to
the side) and adduction (lowering the arm to the
midline), Figure 4(b);
In the transverse plane, external rotation (out-
ward rotation around its axis) and internal
rotation (inward rotation around its axis),
Figure 4(c).
The natural ROMs (Forner-Cordero et al., 2008) and
the torques required to perform typical ADLs (Copaci
et al., 2016; Murray and Johnson, 2004) for each move-
ment of the scapular girdle are reported in Table 1.
3.2. Elbow joint
Although the elbow joint, located between the upper
arm and forearm, includes three independent ball-and-
socket joints enclosed in a single capsule, it is normally
simplified into a two DOF joint:
In the sagittal plane, flexion (raising the arm)
and extension (pushing the arm backward),
Figure 5(a);
Figure 2. (a) Main keywords in the field of robotic rehabilitation. (b) Itemization of relevant publications by year.
Figure 3. Definition of anatomical planes of the human body.
4Journal of Intelligent Material Systems and Structures 00(0)
In the frontal plane, pronation (inward rotation
around its axis) and supination (outward rotation
around its axis), Figure 5(b).
The natural ROMs (Forner-Cordero et al., 2008) and
the torques required to perform typical ADLs (Copaci
et al., 2017; Murray and Johnson, 2004; Zwerus et al.,
2019) for each movement of the elbow joint are
reported in Table 2.
3.3. Wrist joint
The wrist is the joint connecting the bones of the fore-
arm with two bones of the hand, and it is formed by
three joints. The movements of the wrist joint include:
In the sagittal plane, flexion (bending the palm
toward the forearm) and extension (bending the
palm away from the forearm), Figure 6(a);
In the frontal plane, abduction (moving the hand
away from the trunk) and adduction (moving the
hand close to the trunk), Figure 6(b).
Figure 4. Schematization of main movements of the shoulder: (a) flexion/extension. (b) Abduction/adduction. (c) Internal/external
rotation.
Table 1. Normal ROMs (Forner-Cordero et al., 2008) and
torque (Copaci et al., 2016; Murray and Johnson, 2004) of the
shoulder joint.
Movement ROM Min/Max Required torque [Nm]
Flexion 0°/130°O170°12 O14.3
Extension 0°/30°O80°
Abduction 0°/;180°4.2 O12
Adduction 0°/50°
Internal rotation 0°/60°O90°3.9
External rotation 0/;90°
Figure 5. Schematization of main movements of the elbow:
(a) Flexion/extension. (b) Supination/pronation.
Table 2. Normal ROMs (Forner-Cordero et al., 2008) and
torque (Copaci et al., 2017; Murray and Johnson, 2004; Zwerus
et al., 2019) of the elbow joint.
Movement Range (min/max) Required torque [Nm]
Flexion 0°/140°O160°2.8 O5.8
Extension ;26°/0°
Pronation 0°/80°3.5
Supination 0°/85°
Garofalo et al. 5
The natural ROMs (Forner-Cordero et al., 2008;
Freivalds, 2011) and the required torque for ADLs and
rehabilitation purposes (Pitzalis et al., 2023) of the
wrist joint are reported in Table 3.
4. Actuators for exoskeletal applications
A multitude of actuators can be conceived and found
on the market. Regardless of the form of energy input
(Sekhar and Uwizeye, 2012; mechanical, electrical, opti-
cal, magnetic, thermal, chemical) that can be used to
classify the device, an actuator can transform generic
energy input into mechanical work, characterized by
force and displacement (Hannaford and Winters, 1990).
Thanks to their energy efficiencies and power out-
put, pneumatic, hydraulic and electric actuators are the
most common, often referred to as conventional actua-
tors (Greco et al., 2022). A large number of upper limb
rehabilitation exoskeletons use this type of actuators, as
shown by several recent works (Bardi et al., 2022; Gull
et al., 2020; Proietti et al., 2022). Their most important
limitations reside in weight, bulk and noise (Kim et al.,
2019; Zaidi et al., 2021).
The current demand for light actuators character-
ized by higher power density, flexibility, adaptability,
and portability is encouraging the recent diffusion of
artificial muscles, otherwise known as soft actuators.
The interest in artificial muscles and smart materials
represents a rapidly growing field with significant scien-
tific interest. Certain types of artificial muscles, such as
Pneumatic Artificial Muscles (PAMs) and Dielectric
Elastomer Actuators (DEAs), are rapidly maturing due
to their wide applicability, as stated by Jing et al.
(2023). Other types of smart materials, such as Shape
Memory Alloys (SMAs), Carbon Nanotubes (CNTs),
graphene, and Twisted and Coiled Artificial Muscles
(TCAMs), are currently under investigation in numer-
ous scientific studies due to their extremely interesting
properties, which will be extensively described later.
For instance, as shown in some studies (Aliseichik
et al., 2022; Gonzalez-Vazquez et al., 2023), artificial
muscles prove to be highly promising actuators in the
realm of soft wearable rehabilitation robots. However,
a better understanding of minimal biomechanical
requirements is necessary to design solutions capable of
providing a reasonable level of assistance for specific
tasks. The integration of artificial muscles into rehabili-
tative exoskeletal structures should be a gradual pro-
cess, initially involving relatively simple tasks before
eventually progressing to more complex movements
that may require higher forces or fast joint movement
speeds with rapid accelerations and decelerations.
There are high expectations regarding the integration
of these innovative actuation technologies, as they are
believed to enhance the quality of life for the elderly
and individuals affected by diseases and injuries that
have rendered them significantly disabled due to muscle
weakness.
According to Mirvakili and Hunter (2018), artificial
muscles belong to the category of actuators capable of
generating a reversible change in their shape (contrac-
tion, expansion, or rotation) when exposed to an exter-
nal stimulus. The main difference with conventional
actuators is that, while the latter need multiple compo-
nents to work, the actuation process of artificial mus-
cles can be designed as a single component.
With a specific focus on exoskeletons accomplish-
ment in the rehabilitation context, the selection of the
actuation mechanism is crucial, especially when aiming
to build a portable system (Manna and Dubey, 2018).
Moreover, the complexity of human joints complicates
the selection process, as different types of actuators
may be suitable depending on the chosen joint.
Although the concept of ‘‘actuator’’ typically refers to
devices capable of generating movement, exoskeletal
actuators can be both active and passive (Bardi et al.,
2022). Active actuators can generate movements from
signals derived, eventually, from the human body, such
Table 3. Normal ROMs (Forner-Cordero et al., 2008;
Freivalds, 2011) and torque (Pitzalis et al., 2023) of the wrist
joint.
Movement Range (min/max) Required torque
[Nm]
Flexion 0°/90°0.35 O3 for
rehabilitation
6O10 for ADLs
Extension 0°/80°
Adduction
(Radial deviation)
0°/30°O40°0.35 O3 for
rehabilitation
8O14 for ADLsAbduction
(Ulnar deviation)
0°/15°
Figure 6. Schematization of main movements of the wrist:
(a) flexion/extension. (b) Adduction/abduction.
6Journal of Intelligent Material Systems and Structures 00(0)
as ElectroMyoGraphic (EMG; Bi et al., 2019) or even
ElectroEncephaloGraphic (EEG; Maura et al., 2023)
signals. On the other hand, passive actuators are joint-
supporting elements that can provide forces without
the need for an external stimulus, using devices such as
springs, cables, and elastic bandages.
Among the several parameters usable to compare
the performances of the different types of actuators
(Zhang et al., 2019), to define a practical guide to select
the most suitable actuator, in the present work the fol-
lowing parameters were chosen:
Operating medium: refers to the type of medium
required to activate the actuator, for example,
compressed air for pneumatic actuators, water
or specific oils for hydraulic actuators, electricity
for electric motors.
Strain (%): represents the maximum percentage
length variation attainable by the actuator.
Force (N): represents the maximum load attain-
able by the actuator; in some circumstances,
when the force is referred to the cross-section of
the actuator, this parameter can be provided as
a stress (MPa).
Torque (Nm): represents the maximum torque
attainable by the actuator; it is mostly used when
the forces are applied at a significant distance
from the anchor points.
Amplitude of the stimulus: refers to the ampli-
tude required to excite the actuator, for example,
pressure for pneumatic and hydraulic actuators,
current/voltage for electric actuators such as
Shape Memory Alloys (SMAs) and Liquid
Crystal Elastomers (LCEs).
Specific power (W/kg): is the power output per
unit mass, which is a measure of power density
of the specific actuator.
Number of cycles: is a parameter quantifying
(typically assuming statistical models) the dur-
ability and reliability of the actuator before
failure.
Bandwidth (Hz): represents the frequency range
within the actuator can work.
4.1. Electric motors
Electric motors, whose working principle is based on
Ampere’s law (Fundamentals of Motor Control, 2023;
WEG, 2023), can be classified as Direct Current (DC)
or Alternating Current (AC) motors. Both types consist
of a stationary part, the stator, and a rotating part, the
rotor (Moyer and Chicago, 2010). The stator generates
a magnetic field (either through permanent magnets or
electromagnets) driving the rotor, which delivers the
mechanical power in the form of a torque (Beaty and
Kirtley, 1998; Chapman, 2012; Moyer and Chicago,
2010). In DC motors, a commutator reverses the cur-
rent flow in the rotor windings to maintain the continu-
ity of the rotation motion. DC motors can be brushed
or brushless, the former use brushes and commutators
for power transfer (Hughes and Drury, 2013), in the
latter the commutation is electronic. Brushless motors
offer higher efficiency and reduced interference (How
Electric Motors Work, 2002) but are more expensive
(Chapman, 2012; Moyer and Chicago, 2010). AC
motors, including synchronous (rotating at the same
speed of the electrical frequency) and induction (rotat-
ing at a speed lower than the electrical frequency)
motors (Beaty and Kirtley, 1998), operate with electri-
cal windings in the stator and permanent magnets in
the rotor (Moyer and Chicago, 2010).
4.1.1. Applications. The high level of adjustability in
terms of speed and torque is one of the most significant
strengths of DC motors, which offer consistent perfor-
mance even at low speeds and are well-suited for inte-
gration into battery-powered mobile systems. In
addition, they possess a high starting torque (Greco
et al., 2022). As a result, DC motors are commonly
applied in several exoskeleton designs, particularly in
portable robotic systems that require lower torque
requirements. On the other hand, due to their capabil-
ity to generate higher forces if compared with DC
motors, AC motors are preferred for industrial robots
(Majidi Fard Vatan et al., 2021).
In the design and accomplishment of exoskeletons,
micro-DC motors are often employed in conjunction
with gearboxes (to amplify torque) and encoders (for
positional feedback; Portescap, 2021). The numerous
models available on the market provide a wide range of
important operative parameters. Torque spans from
5310
23
to 5 Nm, power outputs are typically below
1 kW, lifespan normally does not exceed 500 h, and
supply voltage ranges from 1.2 to 42 V (Johnson
Motor, 2023; ZHAOWEI, 2023). More details are
reported in Supplemental Table S.1.
Electric motors can be used to develop both rigid
and soft exoskeletons, although the high stiffness of the
former type could be critical, because it restricts the
ROMs of the joints, leading to power losses and system
malfunctions (Copaci et al., 2016). To address these
issues, electric motors are often combined with elastic
elements such as springs and hydraulic cylinders, form-
ing Series Elastic Actuators (SEAs). These elements not
only reduce system stiffness, but also contribute to
weight reduction.
In the work by Qian et al. (2023), they introduced
CURER, a lightweight upper limb rehabilitative exos-
keleton with an anthropomorphic mechanical structure
(Figure 7(a)). The device employed Bowden cables,
electric motors, and SEAs, to enhance compliance and
reduce limb weight.
Garofalo et al. 7
Herbin and Pajor (2021) presented a mechatronic
system similar to a wearable robot (Figure 7(d)). Their
system replicated kinematic movements of the human
upper limb using a 7-DOF system actuated by DC
motors and Bowden cables. The cable system was
equipped with pre-tensioned springs to enhance flexibil-
ity and control.
Pan et al. (2022a) proposed NESM-g, an upper limb
exoskeleton for post-stroke rehabilitation (Figure 7(c)).
The exoskeleton featured 4-DOF and utilized DC elec-
tric motors and SEAs to interact with the patient’s
shoulder and elbow. It offered compactness and ease of
maintenance.
It is evident from the aforementioned applications
that electric motors are commonly used in cable-driven
devices for exoskeletons. Cable-driven actuators employ
flexible cables housed within sheaths (Bowden cables) to
transmit force and motion, reducing friction and provid-
ing protection for the user. These actuators typically con-
sist of an electric motor as the drive unit, benefiting from
increased power and bandwidth (Bardi et al., 2022). The
drive unit applies force to one or more cables connected
to the load, and the movement of the load is generated
by the tension and release of the cables, controlled by the
drive unit. Cable-driven actuators are inherently compli-
ant and flexible, making them suitable for wearable
Figure 7. Some recent applications of cable-driven actuators with electric motors for exoskeleton: (a) sections and details of CURER
structure. Readapted from Qian et al. (2023); (b) elbow exosuit, hand exoskeleton and full device of the device build-up by Noronha et al..
Readapted from Noronha et al. (2022); (c) overview of the NESM-g. Readapted from Pan et al.(2022a); (d) schematic of allowed DOFs of
the exoskeleton developed by Herbin and Pajor. Reproduced with permission from Herbin and Pajor (2021); (e) prototype of soft elbow
exoskeleton for the assistance of flexion/extension and pronation/supination movements. Reproduced from Ismail et al. (2019).
8Journal of Intelligent Material Systems and Structures 00(0)
robots, as skeletal muscles behave similarly with actua-
tor/tendon pairs (Xiloyannis et al., 2022).
One advantage of cable-driven actuators is that the
drive units can be placed away from the robotic devices,
reducing size and weight, which are critical design chal-
lenges in wearable exoskeletons (Ochieze et al., 2023).
However, cable-driven actuators are primarily suitable
for personalized devices as cable routing and anchor
point placement must be carefully tailored to the
patient’s anatomy to ensure proper force transfer (Ang
et al., 2023). Additionally, they suffer from other draw-
backs, including low mechanical efficiency due to fric-
tion losses and cable compliance, as well as high shear
forces on the patient’s skin (Xiloyannis et al., 2022).
Noronha et al. (2022) presented a recent application
of cable-driven actuators in a rehabilitative exoskeleton
for chronic stroke patients. The exoskeleton was
divided into two separate portions, one for the elbow
and one for the hand, both actuated by DC electric
motors driving the Bowden cables. These portions
could work independently or synergistically, to simulta-
neously facilitate rehabilitation of the elbow and hand
(Figure 7(b)).
Ismail et al. (2019) also employed a similar
approach, designing a 2-DOF soft exoskeleton to assist
flexion/extension and pronation/supination movements
of the elbow (Figure 7(e)). The cable-driven actuation
system implemented two coupled DC electric motors
with a worm gear mechanism for converting rotation
into linear motion.
Li et al. (2018) developed a 7-DOF upper limb exos-
keleton for alleviating post-stroke complications. The
exoskeleton was powered by 10 brushless DC electric
motors, with force transmitted to the patient’s joints
through Bowden cables. EMG signals demonstrated a
58% reduction in outgoing muscle signal amplitude due
to the exoskeleton’s utilization.
4.2. Fluid power actuators
Fluid power actuators convert fluid pressure into
mechanical work. They include pneumatic and hydraulic
actuators, which differ in the type of fluid medium used:
hydraulic actuators use liquid (e.g. water or oils
(Johnson, 2002)), while pneumatic actuators use gas
(typically air (Ponomareva, 2006)). They can be classi-
fied as linear or rotary, generating force or torque,
respectively. Hydraulic actuators operate at high pres-
sures, requiring stronger materials and resulting in hea-
vier systems. Pneumatic actuators operate at lower
pressures, using lightweight materials for reduced bulki-
ness. Typical forces and torques vary depending on the
type, pressure, and dimensions (Atos, 2023; FESTO,
2023b, 2023c; Kra
¨mer, 2023). This type of actuators
consist of various components such as compressors or
tanks (pneumatic), pump (hydraulic) and various types
of sensors (Rosenfeld, 2017), which limit portability for
wearable exoskeleton devices. Pneumatic actuators offer
flexibility, moderate specific power, and lower costs, but
experience efficiency losses. Hydraulic actuators provide
higher forces and powers but involve higher costs. For
more detailed information, refer to Supplemental Tables
S.2 and S.3 for pneumatic and hydraulic cylinders,
respectively, provided as Supplemental Material.
Despite their high-power density, due to the reliabil-
ity of pumps and compressors, pneumatic and hydrau-
lic cylinders are mainly used in non-portable devices.
To meet the demand for portable and wearable devices,
compliant materials have been used to develop
Pneumatic Artificial Muscles (PAMs) and Hydraulic
Artificial Muscles (HAMs; Greco et al., 2022). These
actuators consist of an elastomeric chamber wrapped
in interlaced fibers and can expand or contract by
injecting compressible or incompressible fluids (Kalita
et al., 2022; Paterna et al., 2022). Different types of
PAMs exist, with the most common being braided mus-
cles, also known as McKibben muscles. Traditional
PAMs contract when the chamber is pressurized, while
Inverse Pneumatic Artificial Muscles (IPAMs) expand
under pressure (Hawkes et al., 2016). PAMs can exhibit
bending and rotation movements through customized
fiber arrangements (Fu et al., 2022). A recent study
introduced Fluidic Fabric Muscle Sheets (FFMSs; Zhu
et al., 2020), a type of PAM capable of compression,
bending, and adapting to objects, including the human
body. PAMs can generate forces ranging from 120 to
6000 N, but their force-displacement characteristics are
nonlinear, posing challenges for control and modeling
(Kalita et al., 2022). The supply pressure typically
ranges from 1 to 6 bar, and the bandwidth can reach
up to 150 Hz (FESTO, 2023a). PAMs have significant
strokes, usually around 30–50% (Daerden et al., 2001).
They offer advantages such as lightweight design, high
power density, ease of implementation, and compli-
ance. Further information is available in Supplemental
Table S.4.
4.2.1. Applications. Pneumatic actuators are used for
both rigid and soft exoskeletons. Figure 8 shows some
relevant applications. Klein et al. (2008) designed the
BONES exoskeleton for upper limb rehabilitation
(Figure 8(a)). The device, pneumatically actuated,
allows rehabilitation of the shoulder’s three main
DOFs and flexion/extension movement of the elbow.
Four pneumatic cylinders are positioned on the fixed
frame for shoulder control, while a fifth actuator is
used to control elbow movements. The device is capa-
ble of providing torques that significantly exceed the
requirements for ADLs or rehabilitation of a compro-
mised limb.
Lee’s work (Lee, 2014) presents a wearable exoskele-
ton for hand and elbow rehabilitation (Figure 8(c)).
The elbow module consists of a hydropneumatic
Garofalo et al. 9
Table 4. Properties of main conventional actuators and artificial muscles for upper limb exoskeletal applications: green, high applicability; yellow, moderate applicability; red, low
applicability.
Actuator Type(s) of actuation Operating medium
andamplitude of
the stimulus
Force capability
[N]
Torque capability
[Nm]
Specific power
[W/kg]
Strain
[%]
Bandwidth
[Hz]
Number of cycles
[–]
Main advantages Main drawbacks
Electric DC
micromotors
Rotary motion Electric energy
1.2 O42 V
[56,57]
– 0.005 O5
Require systems to
amplify the torque
50 O\1000 10 O40
[20]
–17O500 h
[56]
High efficiency
High precision control
Low torque
High initial costs
Pneumatic cylinder Linear motion
Rotary motion
Compressed air
0.1 O10 MPa
[68]
23 O1870
(Linear cylinder)
[68]
0.2 O112
(Rotary cylinder)
[69]
100 O1000
[20]
10 O100
[20]
–10
7
cycles
[20]
Ease installation
and maintenance
Lightweight
Bulky due to compressors
or tanks
High noise levels
Hydraulic cylinder Linear motion Water, petroleum or
synthetic oils
100 O250 MPa
[67]
3800 O640,000
[67]
– 200 O\10,000
[20]
10 O100
[20]
–10
7
cycles
[20]
High force capability
Ease installation
and maintenance
Bulky due to the need
of a pump
High noise levels
Pneumatic/hydraulic
artificial muscles
(PAMs/HAMs)
Linear motion
Bending motion
Rotary motion
Compressed air
(Pneumatic)
Water or oils
(Hydraulic)
0.1 O0.8 MPa
[76]
120 O6000
[76]
– Up to 10,000
[20]
30 O.40
[77]
Up to 150
[76]
.10
7
cycles
[71]
Lightweight
Low cost
High actuation force
Bulky due to compressors
or tanks
Non-linear force
SMA actuators: wires
and springs
Linear motion
Bending motion
Rotary motion
Electric energy
(Joule effect)
45 O9500 mA
[89]
0.1 O35 (wires)
0.2 O50 (springs)
[89]
– 1000
[21]
5
Suggested \3.5 (wires)
15.2 O32 (springs)
[89]
\3
[89]
.10
5
cycles
@\3.5%
[89]
Lightweight
Noiseless
High force capability
Limited stroke
Hysteresis phenomena
Relatively expensive
Twisted and Coiled
Artificial Muscles
(TCAMs)
Linear motion
Rotary motion
Heat
Up to 130°C
(Polyethylene)
Up to 240°C
(Nylon 6,6)
[94]
.7
[98]
– 27,100
(nylon 6,6)
[94]
Up to 49
(nylon 6)
[94]
Up to 5
(in water)
Up to 7.5
(in helium)
[43]
1.2 310
6
cycles
(nylon 6,6 @10%)
[43]
Low cost
Low voltages
Large power density
Low energy efficiency
Slow heat dissipation
Dielectric Elastomers
(DEAs)
Linear motion Electric Energy
1O10 kV
[106]
Up to 1
(single DEs layer)
Up to 94
(stacked DEs)
[22]
–\200
[47]
1.8 O79 (thickness strain)
12.4 O380
(area strain)
[106]
\1000
[43]
.5310
7
cycles
(stacked DEs)
[22]
Large stoke
Lightweight
Noiseless
High actuation voltage
Low actuation forces
(single DEs layer)
Liquid Crystal
Elastomers
(LCEs)
Linear motion
Bending motion
Electric Energy
1O3V
[108]
Up to 4
[108]
–1
[21]
Up to 40
[107,108]
Up to 100
(electric actuated)
[21]
.10
4
cycles
[21]
Large stroke capability
Low actuation voltage
Lightweight
Noiseless
Low specific power
Low actuation force
Ionic Polymer
Metal Composites
(IPMCs)
Bending motion Electric Energy
0.5 O10
[110]
0.00098 O0.198
[111]
– 2.6
(metallic
electrodes)
244
(non-metallic
electrodes)
[43]
.3
(metallic electrodes)
Up to 8.2
(non-metallic electrodes)
[43]
100
(metallic electrodes)
33 (non-metallic
electrodes)
[73]
76,000 cycles
(metallic electrodes)
[43]
Low actuation voltages
Fast response
Lightweight
Low actuation force
Poor durability in dry
environments
Require humid condition
for the actuation
10 Journal of Intelligent Material Systems and Structures 00(0)
actuator with a movable cylinder, a micro-compressor,
and a controller. Hand rehabilitation is facilitated
through the use of SEAs, which consist of torsion
springs and steel wires for motion transmission.
A significant disadvantage of the two aforemen-
tioned devices is their excessive rigidity, mainly due to
the use of rigid elements such as pneumatic and/or
hydraulic cylinders. This greatly limits the portability
and adaptability of the exoskeletons, often making
them suitable only for use in specific hospital or clinic
environments. To overcome this problem, as mentioned
earlier, soft materials and artificial muscles are today
preferable.
A classic example that supports the aforementioned
statement can be found in the work of Ang and Yeow
(2020). They developed a 3D-printable soft exoskeleton
aimed at assisting patients in the flexion/extension
movement of the elbow. The actuator used is a PAM
with a bellows geometry (Figure 8(b)). The air cham-
bers are optimized to convert pressure variations into
force variations through the design of the bellows.
When the actuator is powered, the bellows deform, gen-
erating flexion in the actuator. This actuator is capable
of providing a force of up to 100 N with a supply pres-
sure of 2 bar. However, the authors have also described
a force model that can customize the dimensions of the
actuator to adjust its force capacity. The soft actuator
is integrated into a wearable textile sheath. The activa-
tion signal is obtained using EMG signals captured by
electrodes placed on the patient’s skin.
Figure 8(d) illustrates the Wearable Upper Limb
Rehabilitation Assistance Exoskeleton System
(WURAES) with 4-DOF, developed by Chiou et al.
(2022). Specifically, the device, designed for rehabilita-
tion of the shoulder and elbow flexion/extension move-
ments of both limbs, is actuated by four torsion spring
mechanisms and PAMs that act through steel wires on
each joint. To capture the user’s movement intentions,
surface electrodes were used to monitor EMG signals,
using two simple bracelets placed on the user’s
Figure 8. Typical applications of fluid power actuators for the development of rigid and soft exoskeletons: (a) different rendered
views of BONES exoskeleton. Readapted from Klein et al. (2008); (b) 3D printed actuator with a bellow type design for the flexion
movement and the prototype of the soft exoskeleton for the assisting of elbow flexion/extension DOF. Readapted from Ang and
Yeow (2020); (c) hand and arm modules of the wearable robot developed by Lee. Readapted from Lee (2014); (d) picture of the
WURAES developed by Chiou et al., where EMG bracelets can be observed placed on the user’s forearms, the PAMs are positioned
on the back of the device. Reproduced from Chiou et al. (2022); (e) soft and lightweight glove actuated by HAMs for assisting home
rehabilitation. Reproduced with permission from Polygerinos et al. (2015).
Garofalo et al. 11
forearms. Their operation is based on different encoded
gestures that are appropriately converted into corre-
sponding rehabilitative movements.
In the work of Ma et al. (2020), an innovative wear-
able exoskeleton for upper limb rehabilitation is pro-
posed. The device consists of a 4-DOF system for
rehabilitating the shoulder’s abduction/adduction,
elbow flexion/extension, wrist flexion/extension, and
hand movements. The actuation of elbow, wrist, and
hand movements is achieved through appropriately
positioned PAMs along the limb. These PAMs are
made of flexible and compliant materials such as sili-
cone, PLA, and Kevlar fibers, which allow for adjusta-
ble actuator movements according to the user’s needs.
The abduction/adduction movement of the shoulder is
achieved using a soft spherical actuator made of PVC.
Specifically, it is positioned under the patient’s armpit,
and, when pressurized, it lifts the patient’s arm, gener-
ating the shoulder abduction movement.
Another typical application of McKibben muscles is
proposed by Tschiersky et al. (2020), who developed a
wearable robot for upper limb assistance. The device is
actuated by flexible pneumatic actuators, consisting of
thin McKibben muscles connected in series with elastic
bands. The device can operate passively, providing
gravitational compensation, as well as actively, enabling
the rehabilitation of shoulder abduction/adduction
movement. The authors demonstrated that the exoske-
leton, with an actuator width of 80 mm, is capable of
generating a torque of approximately 4 Nm with a 90°
arm elevation.
Another interesting application is shown in Figure
8(e), featuring a soft robotic glove designed for hand
rehabilitation by Polygerinos et al. (2015). The device is
actuated by HAMs positioned on each finger. Each
HAM consists of an elastomeric chamber externally
reinforced with fibers strategically positioned for gener-
ating flexion, torsion, or extension motions. The glove,
including the pressurized fluid (water), weighs a total
of 285 g. It is composed of various HAMs appropri-
ately positioned using specific supports, along with a
neoprene structure placed on the back of the hand,
which is anchored to the wrist using a strap. All power
supply components, including the battery, water reser-
voir, sensors, and controller, are positioned on a waist
belt weighing approximately 3.3 kg.
4.3. Shape memory alloys (SMAs)
Materials that can return to their original shape after
exposure to external stimuli, such as electricity, heat,
magnetic fields, light, or changes in pH, are known as
shape memory materials. This unique property, termed
the Shape Memory Effect (SME), is performed by some
metal alloys called Shape Memory Alloys (SMAs) and
certain polymers called Shape Memory Polymers
(SMPs). SMPs can undergo deformations ranging from
200 O800%, but they typically exert relatively low
forces, usually in the range of a few tenths of a Newton
(Liu et al., 2007). Consequently, they are not suitable
for use in exoskeletons and will not be discussed in
detail.
SME can manifest in three primary forms (O
¨zkul
et al., 2019): One-Way SME (OWSME), Two-Way
SME (TWSME), and PseudoElasticity (PE). OWSME
materials can recover their original shape only in the
high-temperature phase, achieved through repetitive
cycles of deformation and heating. In contrast,
TWSME materials can return to their initial configura-
tion both at high and low temperatures, transitioning
between shapes through cycles of heating and cooling.
PE materials, on the other hand, can recover significant
deformations without the need for an external stimulus.
However, it’s important to note that SME is not an
inherent property of materials but rather a result
achieved through appropriate training techniques.
In SMAs, the SME occurs due to the transition
between two crystalline phases: martensite (low-tem-
perature phase, relatively soft) and austenite (high-tem-
perature phase, relatively hard), which depends on
temperature and stress levels (Mavroidis et al., 1999).
Classic examples of SMAs include Ag-Cd, Au-Cd, Cu-
Al-Ni alloys, and many others, although Ni-Ti-based
alloys (Ni-Ti, Ni-Ti-(X)) are the most popular due to
their significantly better thermomechanical perfor-
mance compared to others (Hamid et al., 2023).
SMAs are promising for actuation due to their
lightweight nature and high force capabilities, offering
greater specific power compared with most commonly
used actuators. For instance, commercially available
Ni-Ti-based SMA wire actuators (SAES Getters,
2023) can produce forces ranging from 0.1 to 35 N
depending on their diameter and require current
inputs between 45 and 9500 mA. They also hold
appeal for shaping custom textile structures, expand-
ing possibilities in wearable soft structures (Fu et al.,
2022). However, SMA actuators have limitations,
including restricted strain capabilities (typically
around 5%), and limited bandwidth due to heating
and cooling cycles. Cooling times, influenced by fac-
tors like temperature and wire diameter, can be
lengthy, constraining their use in robotics (Kumar
et al., 2020). More information about SMA actuators
are reported in Supplemental Table S.5.
4.3.1. Applications. SMAs, especially in the form of wires
or springs, are widely applied in the development of
wearable exoskeletal systems for upper limb rehabilita-
tion. An intriguing application is exemplified by Copaci
et al. (2017), who have created a wearable exoskeleton
based on filamentary SMA actuators (Figure 9(a)).
Specifically, the device, weighing approximately 0.6 kg
(excluding electronics), facilitates the rehabilitation of
12 Journal of Intelligent Material Systems and Structures 00(0)
elbow flexion/extension movements and is composed of
only six rigid structural components manufactured via 3D
printing for attachment to the patient’s limb. The SMA
actuators consist of three fundamental elements: Bowden
cables, to simplify the mechanical transmission of move-
ments while minimizing mechanical losses; Teflon sheaths,
to insulate the SMA wires from the Bowden cables; SMA
wires, connected to a power supply for energy delivery
through the Joule heating effect. The proposed device is
highly customizable according to individual needs and
can significantly reduce costs and production times due to
the utilization of 3D printing technology.
Park et al. (2023) have recently introduced an Upper
Limb Exosuit (ULE) designed to assist elbow flexion/
extension movements, employing a fabric woven actua-
tor composed of springs made from Ni-Ti-based SMA
wires with a diameter of 40 mm (Figure 9(b)). This con-
figuration significantly enhances the force capacity of
SMA actuators and overcomes their primary limitation,
that is, their limited deformation capability, enabling
contractions ranging from 50% to 30% with variable
loads between 4 and 10 kg. The proposed ULE is
exceptionally lightweight, weighing only 540 g (includ-
ing electronics and the battery), and experimental
results highlight its ability to reduce muscle activity by
approximately 50%–70% with loads up to 10 kg, while
keeping limb ROMs unaffected.
Another application of SMA actuators is presented
by Jeong et al. (2019), who developed an assistive exos-
keleton for wrist flexion/extension and adduction/
abduction movements, called the Soft Wrist Assist
(SWA) and shown in Figure 9(c). In this case, SMA
spring-like actuators are employed to maximize their
force and deformation capabilities. These actuators
have demonstrated a force capacity of 10 N and a 40%
contraction. Furthermore, to improve the bandwidth
of the actuators, the SMA springs incorporate a cool-
ing system composed of polymeric tubes made with
Ecoflex 00–30, cooled by mineral oil. The robotic
device weighs only 151 g and, thanks to its soft nature,
allows wearing times \2 min, simplifying the execu-
tion of many ADLs.
Figure 9. Exoskeletal applications of SMA actuators: (a) wearable exoskeleton for elbow rehabilitation with SMA actuators.
Reproduced from Copaci et al. (2017); (b) ULE for elbow support built with SMA-based fabric muscles. Readapted from Park et al.
(2023); (c) different wrist motions supplied by SWA: flexion/extension and adduction/abduction of the wrist. Reproduced from Jeong
et al. (2019); (d) soft wearable robot with SMA actuators for elbow flexion/extension and pronation/supination. Readapted from
Jeong et al. (2022).
Garofalo et al. 13
Finally, another notable application has been devel-
oped by Jeong et al. (2022), who have proposed a soft
wearable robot for assisting elbow flexion/extension
and pronation/supination movements using SMA
alloys (Figure 9(d)). The artificial muscle has been
designed using a bundle of helical springs incorporated
within an extensible tube, into which a cooling fluid is
injected. Experimental trials have revealed that the
robot can generate a torque exceeding 3.2 Nm to assist
elbow flexion/extension movements. Furthermore, it
has demonstrated the capability to increase the ROM
for pronation/supination by an average of 15.7°
through the application of a torque of approximately
0.5 Nm. Specifically, the mechanism for pronation/
supination of the elbow, which is challenging to design
with a simple linear actuator due to the human body’s
anatomy, was realized using torsional bands directly
connected to the SMA actuators and wrapped around
the forearm at a 45°, to maximize the torsional torque.
4.4. Twisted and coiled artificial muscles (TCAMs)
In 2014, Haines et al. (2014) successfully demonstrated
how certain high-strength precursor polymer fibers,
such as nylon 6, nylon 6,6, and polyethylene, could be
effectively utilized for the construction of Twisted and
Coiled Artificial Muscles (TCAMs). TCAMs, also
known as Twisted and Coiled Polymers (TCPs), are
fabricated using cost-effective high-strength polymer
fibers like fishing line and sewing thread. These fibers
are transformed through the introduction of twist to
create muscles capable of contraction and expansion.
The twisting of the fibers generates coils capable of
achieving wide excursions and lifting heavy loads
through tensile and torsional actuation, with minimal
(or even negligible) hysteresis phenomena (Yuan and
Poulin, 2014).
These muscles can be actuated through various
methods, such as chemical, photonic, hydrothermal, or
electrothermal activation. The last method is often pre-
ferred for its versatility and simplicity (Suzuki and
Kamamichi, 2018). Nevertheless, since polymers are
non-conductive materials, Joule heating activation
requires the use of appropriate conductive elements.
For this reason, TCAMs are frequently coated with
metallic wires, sprinkled with sheets of Carbon
NanoTubes (CNTs), or incorporate conductive ele-
ments either internally or externally within the coiled
muscle fibers. In any case, the temperature at which
maximum deformation occurs depends on the precur-
sor material: for instance, in the case of nylon 6,6, it is
around 240°C, whereas in polyethylene, it is around
130°C (Haines et al., 2014).
The process of manufacturing TCAMs involves
twisting polymer fibers under the influence of an appro-
priate load to achieve a structure similar to that of a
spring (Saharan and Tadesse, 2019). The precursor
fiber is anchored at one end to a motor that is set in
rotation at a specific speed, while a weight is applied to
the other end to keep the fiber under tension during the
process. The amount of load to be applied depends on
the dimensions and properties of the precursor fiber
used, and it is crucial for the proper fabrication of
TCAMs.
The fiber can be wound upon itself or around a
mandrel. In the former case, TCAMs with higher load
capacities but reduced stroke capabilities are obtained,
whereas in the latter case, the opposite features are
achieved. Once the desired structure is obtained, an
annealing treatment is required to remove internally
induced residual stresses from the coiled structure and
prevent spontaneous unwinding of the actuator.
Subsequently, a training process is carried out using a
test load to refine the performance.
TCAMs can also be combined to create multi-plies
structures, which increases the overall strength of the
resulting actuator, or they can be used to create wear-
able woven fabrics with actuation capabilities.
A distinctive feature of TCAMs is their chirality,
that is, the option to match or mismatch the direction
and number of twists and coils. Specifically, if the direc-
tion of fiber twisting is the same as that of coiling, it is
referred to as homochirality, and the artificial muscle
contracts when activated. Conversely, if the directions
of twisting and coiling are opposite, it is termed hetero-
chirality, and the actuator extends when powered.
Another parameter that impacts the performance of
TCAMs is the spring index C, which is the ratio
between the mean diameter of the coiling and that of
the precursor fiber. It provides an estimate of the level
of coiling of the artificial muscle. A higher C results in a
greater deformation capacity of the actuator, whereas,
a C close to 1 implies a stiffer artificial muscle, that
means higher forces with smaller displacements
(Saharan and Tadesse, 2019).
The use of TCAMs offers significant advantages
compared to the majority of other conventional actua-
tors and artificial muscles. Due to their lightweight
nature and strength, they can achieve tremendous spe-
cific power outputs, reaching up to 27.1 kW/kg (Haines
et al., 2014). Additionally, the precursor fibers are made
from inexpensive polymeric materials (about e5/kg),
making them highly competitive for applications in
robotics and biomedicine. The only drawback of
TCAMs is their very low energy efficiency (’1%),
which makes them energy-intensive and costly in terms
of power consumption. Further information is avail-
able in Supplemental Table S.6.
4.4.1. Applications. Despite TCAMs being exceptional
candidates for the development of wearable rehabilita-
tive exoskeletons, their application in the exoskeletal
context is currently quite limited. In fact, as described
14 Journal of Intelligent Material Systems and Structures 00(0)
in the previous section, this type of actuator is rela-
tively new, therefore, their use as actuators in robotic
applications requires careful and complex analyzes.
Nevertheless, the clear advantages of TCAMs can
enable the creation of wearable exoskeletons, minimiz-
ing the size and weight associated with the components,
and promoting their ergonomic wearability without
excessively disturbing the user’s psychological well-
being. In the following, the only two relevant exoskele-
tal applications identified in the literature are
discussed.
In Figure 10, TCAMs-Exo is shown, a 2-DOF wrist
rehabilitation exoskeleton designed by Greco et al.
(2023) and based on the use of TCAMs. The authors,
aiming to overcome the main limitations associated
with the use of conventional actuators such as, for
instance, the noise and bulk of electric motors and
pneumatic actuators, and the hysteresis and cost of
SMA actuators, have proposed a lightweight soft exos-
keleton (0.135 kg). This exoskeleton enables the rehabi-
litation of wrist flexion/extension and abduction/
adduction movements due to its bioinspired nature. In
particular, the device allows for a ROM of 130°and 79°
for wrist flexion/extension and abduction/adduction,
respectively. It consists of two Non-Deformable Links
(NDLs) positioned on the forearm and hand, and eight
thermally-actuated TCAMs. These TCAMs are made
using carbon fiber and silicone rubber (CF/SR
TCAMs), capable of exerting a force greater than 7 N.
They are indeed an innovative type of actuators, capa-
ble of lifting up to 12,660 times their weight and provid-
ing displacements up to 25% with a voltage on the
order of 0.2 V/cm (Lamuta et al., 2018). Furthermore,
to ensure the safe use of the device, the authors have
proposed the use of an isolation system made of a
stretchable polymer tube, whose diameter has been
sized to minimize heating of the outer layer in contact
with the user’s skin.
In the work of Saharan et al. (2017), they propose a
hand orthosis called iGrab, actuated by TCAMs made
of nylon 6,6 coated with a thin layer of silver (100 nm)
for electrothermal activation. The device, designed and
developed through additive manufacturing, uses six 2-
ply artificial muscles with a diameter of 1.35 mm and a
length of 380 mm, each capable of generating approxi-
mately 3 N of force. Specifically, one 2-ply artificial
muscle is used for each finger of the hand, except for
the thumb, which uses two 2-ply artificial muscles. The
iGrab device allows for flexion/extension movements of
all fingers of the hand and weights approximately
100 g, excluding the power supply. To compensate for
the main drawback of TCAMs, namely their low energy
efficiency, the authors have proposed the use of a lock-
ing system with three linear TCP actuators to maintain
significant positions and thus mitigate energy consump-
tion. Furthermore, to increase the excitation frequency
of the actuators, pulsed power supplies and active
cooling systems are employed, allowing for bandwidths
ranging from 1 to 7.5 Hz. The authors conducted vari-
ous tests by mounting the exoskeletal device on a bio-
mimetic hand to verify its ability to handle different
objects and perform ADLs.
4.5. Electro-active polymers (EAPs)
Electro-Active Polymers (EAPs) are materials capable
of providing mechanical actuation induced by an exter-
nal electric field. Due to their lightweight nature, cost-
effectiveness, ease of processing, and, most notably,
their ability to emulate biological muscles, EAPs are
particularly promising in a variety of emerging applica-
tions, such as micro-robotics and micro-sensing.
The term EAPs includes various types of actuators,
commonly categorized into electronic EAPs and ionic
EAPs. In electronic EAPs, actuation is achieved
through the separation of opposite charges, while in
ionic EAPs, there is a diffusion of ions into and out of
the material. Consequently, electronic EAPs exhibit a
significantly better dynamic behavior due to their
shorter switching times (Grellmann et al., 2022).
Notable examples of electronic EAPs include
Electrostrictive Polymers (EPs), Ferroelectric Polymers
(FPs), Liquid Crystal Elastomers (LCEs), and
Dielectric Elastomers (DEs). On the other hand, key
ionic EAPs include Conductive Polymers (CPs), Ionic
Polymer Gels (IPGs), and Ionic Polymer Metal
Composites (IPMCs). Several of these materials include
a high degree of innovation, and their performances
continue to improve thanks to the ongoing scientific
research. The following sections describe well-
established EAP features and their typical applications
in the field of exoskeletons.
4.5.1. Dielectric elastomers (DEs). Among the various
types of EAPs, DEs unquestionably exhibit the highest
deformation capabilities when exposed to an electric
Figure 10. Exoskeletal application of TCAM actuators: CF/SR
TCAMs used for a wearable exoskeleton for wrist rehabilitation.
Reproduced from Greco et al. (2023).
Garofalo et al. 15
field. Therefore, they are considered the most attrac-
tive, primarily due to their remarkable energy efficiency
(Shankar et al., 2007a). DEs can be seen as parallel-
plate capacitors, consisting of a three-layer structure: a
polymeric film sandwiched between two compliant elec-
trodes. Their operating mechanism can be easily
described using Coulomb’s force definition. By apply-
ing an electric field between the electrodes, an attractive
force, also known as Maxwell stress (1), is generated,
causing the two electrodes to approach each other
(Shankar et al., 2007b). If the dielectric material is
assumed to be incompressible, the reduction in film
thickness results in the expansion of the material in
other directions, namely length and width, depending
on the boundary conditions. The magnitude of the
actuation pressure is provided by the following equa-
tion (Pelrine et al., 2000):
sM=ErE0
V
t
2
=ErE0E2ð1Þ
where e
r
is the relative dielectric constant, e
0
is the
vacuum dielectric constant, Vis the applied voltage, tis
the thickness of the polymeric layer (dielectric mate-
rial), and Eis the applied electric field.
Common dielectric materials used in DEs include
silicone-based elastomers, acrylates, polyurethane, and
elastomeric copolymers. Depending on the material
used as dielectrics, the resulting properties of DE actua-
tors will vary. For instance, polyurethane and acrylates
exhibit high deformation capabilities, up to
400 O500%, although they are subject to strong vis-
coelastic behavior, leading to significant electromecha-
nical losses, with consequent inefficient and slow
actuation. In contrast, silicone elastomers are stiffer
and capable of generating smaller deformations (up to
100%), but they are considerably faster and exhibit
lower losses (Hau et al., 2018; Rizzello, 2023).
As for the electrodes, they should exhibit good elec-
trical conductivity and be capable of maintaining their
conductive properties for millions of cycles (Shankar
et al., 2007a). Furthermore, in order not to compromise
the electromechanical efficiency of the actuator, it is
advisable to minimize the intrinsic stiffness of the used
electrodes. For this reason, electrodes are often made
from carbon-based materials such as carbon powders,
carbon grease, Single-Walled Carbon Nanotubes
(SWCNTs), or metallic materials. Metals, although
their superior conductors compared to carbon, gener-
ally produce small deformations due to their high rigid-
ity (Mirvakili and Hunter, 2018).
DEs can be customized to form different layouts to
enable various actuation modes, typically involving
thickness contraction and expansion of the dielectric
membrane. This flexibility allows for the creation of
two distinct types of actuators based on the chosen
actuation mode (Rizzello, 2023). Thickness contraction
is used to develop actuators with transverse shrinkage.
Since the reduction in thickness of a single DE layer is
typically on the order of a few micrometers, multiple
layers are usually stacked together to achieve a signifi-
cantly greater macroscopic stroke. These are known as
stacked DEAs, capable of generating very high actua-
tion forces, up to 94 N, but with relatively limited dis-
placements, typically not exceeding 10% (Pan et al.,
2022b). As apparent from above, stacked DEAs offer
remarkable versatility, as the force and displacement
capabilities can be adjusted by modifying the number
of layers and the electrode surfaces.
As mentioned earlier, membrane expansion of the
polymeric material can also be utilized as an actuation
mode. However, in this case, unlike with stacked
DEAs, the actuation mode is elongation rather than
contraction. In these cases, actuators are often divided
into two subcategories: in-plane actuators, character-
ized by bidimensional motion, and out-of-plane actua-
tors, characterized by displacement fields occurring
orthogonally to the elastomer, which is facilitated by
elastic elements. Naturally, single-layer DEs are pre-
ferred when a higher stroke capacity is required, but in
this case the amount of the force does not exceed a few
Newtons.
In general, DEs offer significant advantages when
used as actuators, such as the ability to achieve high
deformations, high energy efficiency and density, cost-
effectiveness, flexibility, and lightweight properties.
However, they do have important limitations, espe-
cially in the context of wearable exoskeleton applica-
tions. In fact, their actuation requires extremely high
electric fields (and thus voltages), ranging from 8 to
420 V/mm (with voltages on the order of dozens of kV)
depending on the thickness of the dielectric membranes
(Brochu and Pei, 2010). Additionally, the mechanical
behavior of these materials often exhibits non-linearity,
making their modeling and control exceptionally chal-
lenging. More information about DE actuators are
reported in Supplemental Table S.7.
4.5.2. Liquid crystal elastomers (LCEs). Liquid Crystal
Elastomers (LCEs) combine the typical elastic proper-
ties of elastomeric polymers, that is, elastic behavior
over a wide range of deformation, with those typical of
liquid crystals, such as anisotropy and phase transfor-
mations (Brochu and Pei, 2010; Greco et al., 2022).
LCEs consist of specific groups of liquid crystals,
known as mesogens, flexibly bonded to cross-linked
polymer chains. The cross-linking of chains allows the
mesogens to partially reorient within the polymer struc-
ture when activated, generating forces and displace-
ments. This type of material is soft and highly
compliant and, thanks to the intrinsic properties of
elastomers, has a glass transition temperature (T
g
)
lower than room temperature (Wang et al., 2022). This
16 Journal of Intelligent Material Systems and Structures 00(0)
property enables LCEs to exhibit remarkable deforma-
tive capabilities, especially when compared to many
other types of soft materials.
In response to temperature changes, LCEs rearrange
their internal structure into two distinct phases
(Grellmann et al., 2022). At low temperatures, the
structure is known as the nematic phase, in which the
mesogens align with the polymer chains. A temperature
increase allows the mesogens to switch from an ordered
state to a disordered state, known as the isotropic
phase. This phase change is entirely reversible through
the application and removal of an external stimulus.
The reorientation of mesogens can be achieved
through a wide range of external stimuli, including var-
iations in pH and, notably, the use of electric, magnetic,
or light fields. The main types of LCEs are essentially
two: thermally responsive LCEs and photosensitive
LCEs (Wang et al., 2022). Thermally responsive LCEs
are often integrated with resistive elements such as resis-
tors or photothermal materials. Common actuators
based on LCEs are made of thin films (about 1 mm
thick) with flexible resistors inside to provide the neces-
sary heat for actuation. A recent application of thermo-
responsive LCEs was developed by He et al. (2019),
where the authors demonstrated that a single layer is
capable of generating deformations of up to 40% and
actuation forces of approximately 4 N, with very low
voltages (1 O3 V). These characteristics make LCEs
particularly interesting materials for applications in the
field of soft and micro-robotics. On the other hand,
photosensitive LCEs are highly customizable as they
can exhibit different behaviors depending on the actua-
tor’s geometry and the angles of illumination. However,
for more practical applications, thermoresponsive LCEs
are still preferred, mainly due to their ease of control.
The primary advantage of LCEs is that, differently
from DEAs, they require very low actuation voltages,
typically on the order of a few volts. However, they
have limited power densities and relatively modest
actuation forces. At present, the application of LCEs is
quite limited due to the complex preparation process.
In fact, only a few types of monomeric liquid crystals,
such as RM257 and RM82, are commercially available
and can be used to create LCEs (Wang et al., 2022).
Furthermore, the actuation capabilities of LCEs stem
from the anisotropic arrangement of mesogens. In fact,
without any specific preparation process, mesogens
arrange themselves entirely randomly within the mate-
rial, and in this configuration, LCEs do not exhibit any
kind of actuation response when exposed to external
stimuli. More information about LCEs typical proper-
ties are reported in Supplemental Table S.8.
4.5.3. Ionic polymer metal composites (IPMCs). Ionic
Polymer Metal Composites (IPMCs) consist of an ion-
exchange polymer membrane, commonly known as
ionomer, placed between two metallic or non-metallic
electrodes (Brochu and Pei, 2010). An ionomer is a
thermoplastic material with a high capacity for ion
exchange, thus providing channels for the migration of
ions and affecting the performance of IPMCs (Bernat
et al., 2023). As for most of thermally responsive
EAPs, the electrodes are used to apply voltage to the
surface of the membrane. Operating like the plates of a
capacitor, they induce the migration of ions within the
membrane, resulting in flexural actuation. The electro-
des predominantly used are precious metals such as
gold, silver, platinum, and palladium. However, nowa-
days, highly flexible and conductive non-metallic elec-
trodes such as CNTs and graphene are also widely
employed (Zhang et al., 2023).
The operating principle of IPMCs is similar to that
of DEAs, although in this case the actuation movement
is flexural rather than linear. In fact, the application of
a potential difference between the two electrodes, typi-
cally in the range of 0.5–10 V, allows for the breaking
of molecular bonds in the polymer, releasing mobile
ions into a hydrated domain. Consequently, cations
migrate to the cathode, while anions migrate to the
anode. However, because the diffusion of anions to the
anode is not proportional to that of cations to the cath-
ode, the resulting actuation motion is flexural
(Grellmann et al., 2022).
Regarding polymer membranes, two frequently used
perfluorinated ionomers are Nafion (perfluorosulfonic
acid) and Flemion (perfluorocarboxylic acid), due to
their excellent performance and commercial availability
(Zhang et al., 2023). The electromechanical perfor-
mance of IPMCs is strongly influenced by the proper-
ties of Nafion, that is, conductivity, internal hydration,
and elastic modulus of the membrane. However, the
most commonly used Nafion membrane is produced by
the U.S. company Dupont and commercialized in the
form of thin layers whose thickness spans between 50
and 200 mm. Nevertheless, due to the limited force cap-
abilities of such membranes, which are not higher than
a few mN, practical applications of Nafion are cur-
rently extremely restricted (He et al., 2022). In fact, the
actuation force that a generic IPMCs can generate is
proportional to the characteristic dimensions of the
actuator, as suggested by the following equation
(Mirvakili and Hunter, 2018):
Fb}
wt2
lð2Þ
where w,t, and lrepresent the width, thickness, and
length of the actuator, respectively.
As mentioned, Flemion is also a typical ion-
exchange membrane widely used in the fabrication of
IPMCs, and in many aspects, it proves to be even more
efficient than Nafion. In fact, some experimental stud-
ies (Nemat-Nasser and Wu, 2003) have shown that
Garofalo et al. 17
Flemion ionomers have a higher ion exchange capacity,
greater stiffness, and increased hydration compared to
Nafion. Additionally, at the same voltage, they gener-
ate greater flexural deformative capacities.
Ion exchange in IPMCs is facilitated by membrane
hydration, enabling the diffusion of ions. However, in
conventional Nafion-based IPMCs, water, used as a
solvent, naturally tends to evaporate during actuation.
This leads to a significant reduction in ion flow, com-
promising the stability and durability of the actuator
(Zhang et al., 2023).
Classic variables influencing the mechanical perfor-
mance of IPMCs include dimensions, the type and con-
centration of cations, water content, membrane surface
preparation, the electrode-to-polymer matrix interface
area, as well as electrode conductivity (He et al., 2022).
The primary advantages of these artificial muscles are
undoubtedly their high deformative capabilities, light-
weight nature, high specific energy, and ease of actua-
tion using low voltages. However, as previously
mentioned, one of their major limitations, especially in
dry environments, is their poor durability, primarily
due to solvent loss that occurs during extended actua-
tion. Other useful information about IPMCs actuators
are reported in Supplemental Table S.9.
4.5.4. Applications. The application in exoskeletons of
EAP actuators is currently limited due to their recent
development, which still makes conventional actuators
as preferred option. Nevertheless, several studies can be
found in the literature dealing with various types of
EAP actuators applied in the field of wearable robotics,
including some prototyping efforts.
In the work of Lidka et al. (2018) a development
and analysis study of DEs is carried out to assist wrist
movements. The authors created a single-layered DE,
using a thin silicone film (40 mm thick) as the polymeric
membrane and a mixture of graphite, isopropanol, and
isooctane as electrodes. Force tests highlighted that the
fabricated DE actuator is capable of lifting a mass of
up to 95 g, corresponding to a force of 0.932 N.
However, this force is inadequate to move the wrist,
which would require a force of approximately 9.7 N. A
stack of 11 DEs could therefore be sufficient to deliver
the required force, provided that the force increases lin-
early with the number of connected DEs.
To support the aforementioned points and the asser-
tions made by colleagues Lidka et al., Behboodi and
Lee (2019) conducted evaluation tests on Stacked
Dielectric Elastomer Actuators (SDEAs) commercia-
lized by the company CTsystems (CTSystems AG,
2023; Figure 11(a)). The study revealed that this type
of actuators is capable of generating forces up to
21.74 N and exhibits a specific power of approximately
426 W/kg, with a steady-state current consumption of
only 39 mA. Furthermore, the analyzed SDEAs
demonstrate very high durability, approaching to
5310
7
cycles. All these factors make SDEAs excel-
lent candidates for potential exoskeletal applications.
The only drawbacks are represented by the limited
deformation capacity (’3%) and, as usual, the high
supply voltages. However, the authors note that their
low power consumption, primarily due to extremely
low operating currents, allows for the safe use of these
actuators, provided that appropriate insulation precau-
tions are taken into account.
A typical application of SDEAs for the rehabilita-
tion of finger movements has been proposed by Carpi
et al. (2008). The authors demonstrated how this type
of actuators can effectively replace passive elements,
such as springs and elastic bands. In particular, the DE
actuator was fabricated using a silicone substrate, as
the dielectric, and a pair of compliant electrodes, made
of silicone and black carbon, that is, a specific form of
amorphous carbon commonly used as filling and rein-
forcing agent in many polymeric materials. Figure 11(b)
illustrates the operating principle of the exoskeleton:
when a voltage is applied, it contracts, allowing the lift-
ing of the finger.
In a more recent study, Carpi et al. (2014) demon-
strated that orthoses made using DEs allow for signifi-
cant customization by modulating the stiffness of the
actuator continuously during the rehabilitation of fin-
ger movements (Figure 11(c)). Many of the existing
exoskeletal technologies today are indeed based on the
use of passive elements characterized by a single intrin-
sic stiffness. The use of DEs enables the modulation of
the stiffness of the actuating element, creating an ortho-
sis that is both active and versatile, allowing different
usage based on the motor abilities of the patient.
A recent application of SDEAs is proposed by
Duduta et al. (2019), who used stacked DEs to create a
prototype of an exoskeleton devoted to the rehabilita-
tion of elbow flexion/extension movements (Figure
11(f)). The primary limitation of single-layered DEs,
namely their extremely limited force capacity, has been
overcome by stacking multiple layers using ultra-thin
electrodes based on CNTs, significantly enhancing their
mechanical performance. This allowed the creation of
an actuator capable of generating forces exceeding
10 N and contractions of approximately 24%, with an
applied voltage of around 3.75 kV. Again, the critical
factor limiting the application in wearable robots is the
high activation voltage, which requires careful analysis
for safety issues.
Another recent application of DEs as actuators for
the rehabilitation of elbow flexion/extension move-
ments has been proposed by Behboodi et al. (2020). In
their work, the authors first analyzed the actuation cap-
abilities of a stack of 3 34 DEs connected in series.
Experimental tests demonstrated that this artificial
muscle is capable of generating a force of 30.47 N, with
a maximum contraction of 5.3 mm. Subsequently, a
18 Journal of Intelligent Material Systems and Structures 00(0)
Figure 11. Applications of EAPs for the development of wearable exosuits: (a) SDEA developed by CTsystems and used by Behboodi
and Lee as an alternative actuator for rehabilitation devices. Readapted from Behboodi and Lee (2019); (b) hand orthosis actuated by
DEs, the actuator is placed over the wrist inside the transparent case. Readapted from Carpi et al. (2008); (c) experimental set-up for
modulating the stiffness of the DE actuator during the rehabilitation of finger movements. Reproduced with permission from Carpi
et al. (2014); (d) hand orthosis actuated by spring-roll DE actuators, the actuator is placed inside the yellow case over the forearm.
Reproduced from Amin et al. (2020); (e) elbow joint rehabilitation prototype actuated by a 3 35 stacked DE actuator. Reproduced
from Behboodi et al. (2020); (f) Stacked DEs as actuators for orthoses. The application of voltage allows the flexion of the forearm.
Reproduced with permission from Duduta et al. (2019).
Garofalo et al. 19
stack of 3 35 DEs was mounted on a mannequin,
and its effectiveness in a potential rehabilitation appli-
cation was verified (Figure 11(e)). The accomplished
actuator was able to articulate the elbow at an angular
velocity of 16.2°/s with a tensile load of 1 N. However,
the test results suggest that the use of these actuators
without appropriate transmission systems is quite lim-
ited, primarily due to their size and dimensions.
In Figure 11(d), the exoskeleton prototype and
actuation technology proposed by Amin et al. (2020)
are depicted. The authors introduced a hand-finger
rehabilitation device actuated by spring-roll DEs and a
cable-driven mechanism for power transmission. The
DE actuator is composed of multiple layers of VHB
4910 elastomer sheets and conductive electrodes made
of carbon grease. Pre-strain of the dielectric membrane
improved the actuator’s performance and its failure
resistance. Experimental tests demonstrated that such
an actuator is capable of generating a force of 4.35 N
and a maximum displacement of 11.3 mm, making it
suitable for hand rehabilitation. The final exoskeletal
device consists of four DE independent actuators, one
for each finger but the thumb.
The work by Lee et al. (2012) focused on the
dynamics of pinching in a finger exoskeleton using
IPMCs. The authors analyzed the electromechanical
characteristics of IPMC actuators through mathemati-
cal modeling and experimental analyzes. The IPMC
actuators were fabricated using Nafion membranes and
platinum electrodes. Despite the inherently limited
force capacity of IPMCs, the authors demonstrated
how the arrangement of the actuators significantly
influences the applied force, showing that, by a proper
design, a reduction of up to 50% in the force required
for finger actuation is attainable.
5. Critical analysis and comparison of
actuators
The literature discussed in the present review paper high-
lights the existence of numerous exoskeletons for the
upper limbs and related studies in the field of robotics.
However, there are still several critical aspects to deal
with for the widespread establishment of wearable
robotics. Its development is primarily constrained by the
choice of actuators, as well as the development of spe-
cific support mechanisms necessary for their ergonomic
and flexible use (Manna and Dubey, 2018). In fact, the
design of a wearable exoskeleton should not only con-
sider the technical specifications for the proper function-
ing of the device, with the respect to the DOFs and
ROMs of the joints, but should also aim at the develop-
ment of ergonomic and user-friendly devices. These
should ensure continuous use without fatigue, and gen-
erate a sense of psychophysical well-being rather than
rejection in the user (Faria-Fortini et al., 2011).
Despite the emphasized critical issues, the analysis
of the state-of-the-art conducted in this work allows for
the identification of several fundamental characteristics
for the selection of the most suitable actuator for a
wearable exoskeleton for the upper limbs. These key
features are schematically presented in Figure 12. As it
can be noticed, the selection of the most suitable actua-
tion device should consider four fundamental aspects:
the operation tasks, technical capabilities, economic
advantages, and mental well-being.
Following the scheme, it is apparent that the initial
step is to define the scope of the exoskeletal structure.
Nowadays, there are a multitude of contexts in which
exoskeletons can be employed, ranging from the bio-
medical sector, addressed to rehabilitation and assis-
tance, to the industrial sector, for enhanced strength
capabilities and fatigue reduction (Ang et al., 2023;
Rosen and Ferguson, 2020). Depending on the applica-
tion context, the exoskeletal device will have a set of
characteristics that distinguish it in terms of functional-
ity and capabilities.
Once the application is identified, it is possible to
define a series of technical specifications. This step is
crucial because operational specifications allow for the
definition of the exoskeleton’s capabilities. Fundamental
aspects in this context include:
Wearability, unlike conventional (or ground-
based) exoskeletons, exosuits have to be worn
and ensure pleasant and ergonomic use, regard-
less of their function (Bardi et al., 2022;
Xiloyannis et al., 2022);
DOFs and ROMs capabilities, which depend on
the type of actuators, transmission systems, and
fastening mechanisms used;
Specific power, high force capabilities and light-
weight will limit the sense of fatigue during use;
Size, less bulky designs are generally more com-
fortable for users to wear for extended periods
and allow for greater freedom of movement;
Energy efficiency, high consumption could limit
or even prevent the use of the device.
In addition to all these, economic aspects play an
important role. These mainly include the purchase,
maintenance and management costs, and the durability
of the device. Regarding the purchase cost, this can
vary widely based on factors such as the complexity of
the device, the materials used, and the level of imple-
mented technology. Clearly, customization to meet spe-
cific needs could imply higher costs compared to
standard or widely used models. Nevertheless, the use
of innovative materials (e.g. artificial muscles) com-
bined with alternative production techniques (e.g. 3D
printing), can lead to the design of highly versatile
robotic devices, allowing a significant reduction in pro-
duction costs and, consequently, the purchase cost. The
20 Journal of Intelligent Material Systems and Structures 00(0)
selection of actuation technologies also influences main-
tenance costs. Indeed, wearable exoskeletons for the
upper limbs, like any other mechanical device, require
periodic inspections and regular maintenance to ensure
optimal functioning. Hence the use of more reliable
materials can significantly reduce these cost aspects that
make the device more accessible. Definitely, the dur-
ability of the device is certainly an aspect that influences
its cost-effectiveness. It depends significantly on the
quality of materials and the final design of the device.
In this context, actuators can effectively enhance the
lifespan of exoskeletons, allowing for extended usage
time without the need for maintenance and partial or
even total replacement.
The last key point for the selection of an exoskeletal
actuation device is associated with psychophysical well-
being. In this context, a challenging aspect is related to
the psychological acceptance of these devices. From
this perspective, wearable exoskeletons, unlike ground-
based ones, can positively influence mental well-being
by restoring a sense of independence in individuals with
motor disfunctions. The ability to move and perform
ADLs can significantly boost self-confidence.
Furthermore, the design of a wearable exoskeleton
should follow a user-friendly approach. This would
contribute to making devices easy to wear, comforta-
ble, and versatile, promoting mental and emotional
acceptance. Reducing the stigma associated with
robotic aids can positively impact body image and,
consequently, mental well-being. This goal can also be
achieved through user engagement and feedback.
A compact overview of various actuator types, along
with their main advantages and disadvantages, is pre-
sented in Table 4. Based on their degree of applicability
in wearable robotics for the upper limb, the actuator
types are ranked with three different colors: green, high
applicability; yellow, moderate applicability; red, low
applicability. This ranking was built by analyzing the
spider chart shown in Figure 13. The comparison is
made taking into account eight parameters considered
fundamental for a potential wearable exoskeletal
Figure 13. Comparative analysis of main actuators in upper
limb exoskeleton using a spider web chart. The scale is chosen
arbitrarily from 0 (poor capability) to 8 (excellent capability).
Figure 12. Main features for the selection of actuators for wearable upper limb exosuits.
Garofalo et al. 21
application: bulkiness, noise, cost, output power, power
consumption, stroke capability, lifespan, and band-
width. Analyzing the chart, electric actuators exhibit
the best combination of the considered parameters.
Despite their significant output power, the main limita-
tion is primarily related to weight and size, with conse-
quent limitations in the field of wearable robotics
(Copaci et al., 2016). Similar issues can be raised for
fluid power actuators, for which, compared to electrical
actuators, the bulkiness is further worsened by the pres-
ence of compressors and tanks (for pneumatic devices)
or pumps (for hydraulic devices). Additionally, these
devices are quite noisy, making them unsuitable or gen-
erally less used in wearable robotics, due to the acoustic
stress generated.
In an opposite position to fluid power actuators are
SMA actuators. Indeed, they exhibit excellent perfor-
mance in terms of size, noise level, and output power
(SAES Getters, 2023), but they suffer energy efficiency,
lifespan, cost, and especially stroke capability (Fu
et al., 2022; Kumar et al., 2020). Nevertheless, there are
numerous upper limb exoskeletal applications imple-
mented using this technology, and despite its limita-
tions, the results are noteworthy.
Among artificial muscles, TCAMs have recently
attracted particular attention. They show an excellent
combination of the considered technical parameters.
Their only limitation, as extensively discussed earlier,
lies in their high power consumption and, consequently,
low energy efficiency (Haines et al., 2014). This recent
type of artificial muscle has the potential to be success-
fully applied in the development of wearable exoskele-
tons for the upper limbs. However, there are still few
exoskeletal prototypes actuated by TCAMs, making
this technology a highly active field, open to further
experimental analysis and scientific research.
Lastly, the characteristics of EAPs are also outlined.
Despite their exceptional properties in terms of energy
consumption, noise level, and bandwidth (Grellmann
et al., 2022), EAPs are still relatively underutilized in
wearable robotics, mainly due to their limited force and
stroke capabilities. However, among all available
EAPs, DEs are particularly well-suited for the intended
purpose, thanks to the possible combination in stack
form, which can overcome their primary limitations
(Shankar et al., 2007a).
6. Conclusions
This article provides a detailed survey of the intricate
and complex field of wearable robotics, with a specific
focus on the most suitable actuation devices for the
development of a wearable upper limb exoskeleton in
the biomedical domain. Thanks to the comprehensive
overview conducted in the intricate realm of actuation,
it has been possible to thoroughly analyze the
advantages, disadvantages, and operational principles
of the main actuation technologies. Furthermore, some
typical applications primarily aimed at rehabilitation
have been examined. The focus has been primarily
directed toward a particular type of exoskeletons,
namely those wearable for the upper limb. This is
because, given the growing interest in soft robotics and
the increasingly widespread use of artificial muscles,
the wearability of exoskeletal devices is becoming
increasingly crucial for the development of new rehabi-
litation technologies. Additionally, it has been observed
how the upper limbs are the most involved in ADLs,
and therefore, the ability to wear an exoskeletal device
allows the user to participate successfully in daily activ-
ities while simultaneously improving their mental
health. Motor dysfunctions in the upper limbs signifi-
cantly impact the quality of life by limiting or com-
pletely hindering the ability to perform simple ADLs.
For instance, shoulder and elbow joints facilitate arm
movements in three-dimensional space, while wrist
joints allow for more precise movements by controlling
hand positions. In any case, all these joints are more or
less involved in the execution of any type of activity.
The selection of an exoskeletal actuator obviously
must consider the type of joints to be addressed, as they
significantly influence the final design of the device.
For shoulder and elbow joints, the main technical speci-
fications are essentially related to DOFs, ROMs and
force. Therefore, actuators must ensure lightness and
minimal dimensions while allowing the correct move-
ment of the joints. On the other hand, in the case of the
wrist, forces, DOFs, and ROMs are considerably lower,
and consequently, greater precision in movements may
be required. This is especially true when the exoskeleton
aims to assist in ADLs rather than rehabilitation.
The critical review revealed the transition from con-
ventional actuators to more recent artificial muscles sig-
nificantly improves ergonomics, portability, flexibility,
and, above all, lightness. By mimicking the contraction
and expansion of natural muscles, artificial muscles
enable smoother and more natural movement for the
wearer, allowing individuals to perform tasks that were
previously challenging. The wide overview shows that
exoskeletons equipped with artificial muscles are often
designed to reduce the physical strain on users by pro-
viding support and assistance during repetitive tasks.
This not only enhances the resilience of patients and
shortens their recovery time but also reduces the risk of
musculoskeletal fatigue and its related accidents.
Furthermore, the inherent flexibility of artificial mus-
cles allows for the creation of exosuits that can be tai-
lored to the specific needs and preferences of individual
users. By adjusting parameters such as stiffness, actua-
tion force, and response time, designers can optimize
the performance of the exoskeleton to suit different
tasks and user requirements. Despite the obvious and
important advantages described so far, these actuators
22 Journal of Intelligent Material Systems and Structures 00(0)
still have numerous limitations to overcome for full
application in wearable robotics, mostly related to low
energy efficiencies and force and displacement capabil-
ities. Indeed, artificial muscles need to closely mimic
the behavior of natural muscles to provide effective
support and assistance during rehabilitation. Achieving
biomechanical compatibility involves optimizing fac-
tors such as actuation speed, force generation, and
range of motion to ensure that the exosuit interacts
seamlessly with the user’s body and movement patterns.
Moreover, efficient energy utilization is critical for pro-
longed use of rehabilitation exosuits, particularly for
individuals with mobility impairments who may rely on
them for extended periods. Consequently, artificial
muscles should be designed to minimize power con-
sumption while still delivering the required levels of
assistance, which can be challenging given the varying
energy requirements of different rehabilitation tasks
and movements. Last but not least, precise control
algorithms are essential for coordinating the activation
of artificial muscles in response to the user’s movements
and intentions. This requires robust sensing systems
capable of accurately detecting biomechanical signals
such as muscle activity, joint angles, and forces exerted
by the user. Real-time feedback and adaptive control
strategies are necessary to ensure that the exosuit
responds appropriately to changes in the user’s beha-
vior and environment. Consequently, conventional
actuators, such as electric motors and pneumatic cylin-
ders, whose usage is well-established, are still often pre-
ferred. However, the need for lighter robots and, more
importantly, the possibility of providing rehabilitative
assistance outside hospitals with a consequent lower
psychological impact, increasingly drive the develop-
ment of wearable exosuits that can be worn even under
clothing. The study reported in the present paper high-
lights how this task can be effectively accomplished
through a series of new artificial muscles, such as
SMAs, TCAMs, and EAPs. As observed, many of these
actuators have already been employed in some exoske-
letal prototypes, but further studies would be necessary
to fully validate and extend their use. Indeed, each of
these actuators presents some weaknesses that must be
addressed before their integration into exoskeletal sys-
tems can be assured. For instance, it has been observed
how SMAs and TCAMs suffer particularly from low
energy efficiency and lifespan; or how EAPs have lim-
ited force and displacement capabilities despite their
remarkable performance in terms of energy consump-
tion and bandwidth. The integration of these actuation
systems into a wearable exoskeletal device must there-
fore necessarily take into account a series of fundamen-
tal parameters such as: actuation force and control,
biomechanical compatibility, energy efficiency, durabil-
ity and reliability, cost and accessibility. However, it is
believed that this review, along with the descriptive
tables of actuators provided as Supplemental Material,
could serve as a guideline tool for selecting the most
suitable actuation device for the development of future
innovative wearable exoskeletons.
Acknowledgements
We acknowledge co-funding from Next Generation EU, in the
context of the National Recovery and Resilience Plan,
Investment PE8 – Project Age-It: ‘‘Aging Well in an Aging
Society,’’ CUP H23C22000870006. This resource was co-
financed by the Next Generation EU [DM 1557 11.10.2022].
The views and opinions expressed are only those of the authors
and do not necessarily reflect those of the European Union or
the European Commission. Neither the European Union nor
the European Commission can be held responsible for them.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The authors received no financial support for the research,
authorship, and/or publication of this article.
ORCID iD
Luigi Bruno https://orcid.org/0000-0002-6745-0466
Data availability statement
Data sharing not applicable to this article as no datasets were
generated or analyzed during the current study.
Supplemental material
Supplemental material for this article is available online.
References
Aliseichik AP, Gribkov DA, Efimov AR, et al. (2022) Artifi-
cial muscles (review article). Journal of Computer and Sys-
tems Sciences International 61(2): 270–293.
Amin H, Assal SFM and Iwata H (2020) A new hand rehabi-
litation system based on the cable-driven mechanism and
dielectric elastomer actuator. Mechanical Sciences 11(2):
357–369.
Ang BWK and Yeow C-H (2020) Design and modeling of a
high force soft actuator for assisted elbow flexion. IEEE
Robotics and Automation Letters 5(2): 3731–3736.
Ang BWK, Yeow C-H and Lim JH (2023) A critical review
on factors affecting the user adoption of wearable and soft
robotics. Sensors 23(6): 3263.
Aria M and Cuccurullo C (2017) Bibliometrix : An R-tool for
comprehensive science mapping analysis. Journal of Infor-
metrics 11(4): 959–975.
Atos (2023) Cylinders catalog. Available at: https://atos.com.
Balasubramanian S (2015) Motor impairments following
stroke. Human Systems Neuroscience 1–30. http://dx.doi.
org/10.13140/RG.2.1.3349.6804
Garofalo et al. 23
Bardi E, Gandolla M, Braghin F, et al. (2022) Upper limb soft
robotic wearable devices: A systematic review. Journal of
Neuroengineering and Rehabilitation 19(1): 87.
Beaty HW and Kirtley JL (1998) Electric Motor Handbook.
New York: McGraw-Hill Handbooks. McGraw-Hill.
Behboodi A, DeSantis C, Lubsen J, et al. (2020) A mechan-
ized pediatric elbow joint powered by a de-based artificial
skeletal muscle. In: 42nd Annual International Conference
of the IEEE Engineering in Medicine & Biology Society
(EMBC), Montreal, QC, Canada, July 2020, pp. 4930–
4935. IEEE.
Behboodi A and Lee S (2019) Benchmarking of a commer-
cially available stacked dielectric elastomer as an alterna-
tive actuator for rehabilitation robotic exoskeletons. In:
2019 IEEE 16th International Conference on Rehabilitation
Robotics (ICORR), Toronto, ON, Canada, June 2019, pp.
499–505. IEEE.
Bernat J, Gajewski P, Ko1ota J, et al. (2023) Review of soft
actuators controlled with electrical stimuli: IPMC, DEAP,
and MRE. Applied Sciences 13(3): 1651.
Bi L, Feleke A, Guan C, et al. (2019) A review on EMG-based
motor intention prediction of continuous human upper
limb motion for human-robot collaboration. Biomedical
Signal Processing and Control 51: 113–127.
Blanco Ortega A, Isidro Godoy J, Szwedowicz Wasik DS, et
al. (2022) Biomechanics of the upper limbs: A review in the
sports combat ambit highlighting wearable sensors. Sen-
sors 22(13): 4905.
Brochu P and Pei Q (2010) Advances in dielectric elastomers
for actuators and artificial muscles. Macromolecular Rapid
Communications 31(1): 10–36.
Canning CG, Ada L, Adams R, et al. (2004) Loss of strength
contributes more to physical disability after stroke than
loss of dexterity. Clinical Rehabilitation 18(3): 300–308.
Carpi F, Frediani G, Gerboni C, et al. (2014) Enabling
variable-stiffness hand rehabilitation orthoses with dielec-
tric elastomer transducers. Medical Engineering & Physics
36(2): 205–211.
Carpi F, Mannini A and De Rossi D (2008) Elastomeric contrac-
tile actuators for hand rehabilitation splints. In: Bar-Cohen Y
(ed.) SPIE Smart Structures and Materials +Nondestructive
Evaluation and Healt Monitoring. San Diego, CA, p.692705.
Chapman SJ (ed.) (2012) Electric Machinery Fundamentals,
5th edn. New York: McGraw-Hill.
Chen Y, Yang Y, Li M, et al. (2021) Wearable actuators: An
overview. Textile 1(2): 283–321.
Chiou S-J, Chu H-R, Li IH, et al. (2022) A novel wearable
upper-limb rehabilitation assistance exoskeleton system dri-
ven by fluidic muscle actuators. Electronics 12(1): 196.
Cieza A, Causey K, Kamenov K, et al. (2020) Global esti-
mates of the need for rehabilitation based on the Global
Burden of Disease study 2019: A systematic analysis for
the Global Burden of Disease Study 2019. Lancet
396(10267): 2006–2017.
Copaci D, Caballero Antonio S, Blanco Rojas F, et al. (2016)
Shoulder exoskeleton for rehabilitation actuated with
shape memory alloy. In: Open Conference on Future Trends
in Robotics, Madrid, Spain, 2016, pp. 37–44.
Copaci D, Cano E, Moreno L, et al. (2017) New design of a
soft robotics wearable elbow exoskeleton based on shape
memory alloy wire actuators. Applied Bionics and Biome-
chanics 2017: 1605101–1605111.
CTSystems AG (2023) CTstack and The Transducer Technol-
ogy. Available at: https://ct-systems.ch.
Daerden F, Lefeber D, Verrelst B, et al. (2001) Pleated pneu-
matic artificial muscles: compliant robotic actuators. In:
Proceedings 2001 IEEE/RSJ International Conference on
Intelligent Robots and Systems. Expanding the Societal Role
of Robotics in the ‘The Next Millennium’, Maui, HI, USA,
2001, pp. 1958–1963. IEEE.
de Miguel-Ferna
´ndez J, Lobo-Prat J, Prinsen E, et al. (2023)
Control strategies used in lower limb exoskeletons for gait
rehabilitation after brain injury: A systematic review and
analysis of clinical effectiveness. Journal of Neuroengineer-
ing and Rehabilitation 20(1): 23.
Duduta M, Hajiesmaili E, Zhao H, et al. (2019) Realizing the
potential of dielectric elastomer artificial muscles. Proceed-
ings of the National Academy of Sciences 116(7):
2476–2481.
Faria-Fortini I, Michaelsen SM, Cassiano JG, et al. (2011)
Upper extremity function in stroke subjects: Relationships
between the international classification of functioning, dis-
ability, and health domains. Journal of Hand Therapy
24(3): 257–264.
Feigin VL, Brainin M, Norrving B, et al. (2022) World Stroke
Organization (WSO): Global stroke fact sheet 2022. Inter-
national Journal of Stroke 17(1): 18–29.
FESTO (2023a) Fluidic Muscle DMSP. Available at: https://
www.festo.com.
FESTO (2023b) Round Cylinders DNSU. Available at:
https://www.festo.com.
FESTO (2023c) Semi-Rotary Drives DRRD, Twin Pistons.
Available at: https://www.festo.com.
Forner-Cordero A, Pons JL, Turowska EA, et al. (2008)
Kinematics and dynamics of wearable robots. In: Pons
Jose
´L. (ed.) Wearable Robots, Madrid, Spain, 1st edn.
Wiley. pp.47–85.
Freivalds A (ed.) (2011) Biomechanics of the Upper Limbs,
2nd edn. Boca Raton, Florida: CRC Press.
Fu C, Xia Z, Hurren C, et al. (2022) Textiles in soft robots:
Current progress and future trends. Biosensors and Bioe-
lectronics 196: 1–25.
Fundamentals of Motor Control (2023) STSPIN. Available
at: https://www.st.com.
Gassert R and Dietz V (2018) Rehabilitation robots for the
treatment of sensorimotor deficits: A neurophysiological
perspective. Journal of Neuroengineering and Rehabilitation
15(1): 46.
Gonzalez-Vazquez A, Garcia L, Kilby J, et al. (2023) Soft
wearable rehabilitation robots with artificial muscles based
on smart materials: A review. Advanced Intelligent Systems
5(4): 1–21.
Greco C, Kotak P, Pagnotta L, et al. (2022) The evolution of
mechanical actuation: From conventional actuators to
artificial muscles. International Materials Reviews 67(6):
575–619.
Greco C, Weerakkody TH, Cichella V, et al. (2023) Light-
weight bioinspired exoskeleton for wrist rehabilitation
powered by twisted and coiled artificial muscles. Robotics
12(1): 27.
Grellmann H, Lohse FM, Kamble VG, et al. (2022) Funda-
mentals and working mechanisms of artificial muscles with
textile application in the loop. Smart Materials and Struc-
tures 31(2): 023001.
24 Journal of Intelligent Material Systems and Structures 00(0)
Gull MA, Bai S and Bak T (2020) A review on design of
upper limb exoskeletons. Robotics 9(1): 16.
Haines CS, Lima MD, Li N, et al. (2014) Artificial muscles
from fishing line and sewing thread. Science 343(6173):
868–872.
Hamid QY, Wan Hasan WZ, Azmah Hanim MA, et al.
(2023) Shape memory alloys actuated upper limb devices:
A review. Sensors and Actuators Reports 5: 1–23.
Hannaford B and Winters J (1990) Actuator properties and
movement control: Biological and technological models.
In: Winters JM and Woo SL-Y (eds) Multiple Muscle Sys-
tems. New York, NY: Springer New York. pp.101–120.
Hau S, Rizzello G and Seelecke S (2018) A novel dielectric
elastomer membrane actuator concept for high-force appli-
cations. Extreme Mechanics Letters 23: 24–28.
Hawkes E, Christensen D and Okamura A (2016) Design and
implementation of a 300% strain soft artificial muscle. In:
2016 IEEE International Conference on Robotics and Auto-
mation (ICRA), Stockholm, Sweden, May 2016, pp. 4022–
4029. IEEE.
He Q, Wang Z, Wang Y, et al. (2019) Electrically controlled
liquid crystal elastomer-based soft tubular actuator with
multimodal actuation. Science Advances 5(10): 1–8.
He Q, Yin G, Vokoun D, et al. (2022) Review on improve-
ment, modeling, and application of ionic polymer metal
composite artificial muscle. Journal of Bionic Engineering
19(2): 279–298.
Herbin P and Pajor M (2021) Human-robot cooperative con-
trol system based on serial elastic actuator Bowden cable
drive in ExoArm 7-DOF upper extremity exoskeleton.
Mechanism and Machine Theory 163: 1–13.
How Electric Motors Work (2002) Available at: http://www.
stefanv.com/rcstuff/qf200212.html.
Huamanchahua D, Castaneda-Vasquez C, Vasquez-Espinoza
A, et al. (2021) Robotic Devices Types Exoskeletons for
Elbow Rehabilitation: A Technological Review. In: 2021
IEEE 12th Annual Ubiquitous Computing, Electronics &
Mobile Communication Conference (UEMCON),New
York, NY, USA, 1 December 2021, pp. 0791–0796. IEEE.
Hughes A and Drury B (2013) Electric motors – the basics.
In: Hughes A and Drury B (eds) Electric Motors and
Drives. Elsevier. pp.1–38.
Ismail R, Ariyanto M, Perkasa IA, et al. (2019) Soft elbow
exoskeleton for upper limb assistance incorporating dual
motor-tendon actuator. Electronics 8(10): 1184.
Janssen MM, Bergsma A, Geurts AC, et al. (2014) Patterns
of decline in upper limb function of boys and men with
DMD: An international survey. Journal of Neurology
261(7): 1269–1288.
Jeong J, Hyeon K, Jang S-Y, et al. (2022) Soft wearable robot
with shape memory alloy (SMA)-Based artificial muscle
for assisting with elbow flexion and forearm supination/
pronation. IEEE Robotics and Automation Letters 7(3):
6028–6035.
Jeong J, Yasir IB, Han J, et al. (2019) Design of shape mem-
ory alloy-based soft wearable robot for assisting wrist
motion. Applied Sciences 9(19): 4025.
Jing Y, Su F, Yu X, et al. (2023) Advances in artificial mus-
cles: A brief literature and patent review. Frontiers in
Bioengineering and Biotechnology 11: 1–23.
Johnson CO, Nguyen M and Roth GA; GBD 2016 Stroke
Collaborators (2019) Global, regional, and national
burden of stroke, 1990-2016: A systematic analysis for the
Global Burden of Disease Study 2016. Lancet Neurology
18(5): 439–458.
Johnson J (2002) Introduction to Fluid Power. Albany, NY:
Delmar Thomson Learning.
Johnson Motor (2023) Standard DC Series Motors Stock
Products. Available at: https://www.johnsonelectric.com.
Kalita B, Leonessa A and Dwivedy SK (2022) A review on
the development of pneumatic artificial muscle actuators:
Force model and application. Actuators 11(10): 288.
Kapsalyamov A, Hussain S and Jamwal PK (2020) State-of-
the-art assistive powered upper limb exoskeletons for
elderly. IEEE Access 8: 178991–179001.
Kim J, Kim JW, Kim HC, et al. (2019) Review of soft actua-
tor materials. International Journal of Precision Engineer-
ing and Manufacturing 20(12): 2221–2241.
Klein J, Spencer S, Allington J, et al. (2008) Biomimetic
orthosis for the neurorehabilitation of the elbow and
shoulder (BONES). In: 2008 2nd IEEE RAS & EMBS
International Conference on Biomedical Robotics and Bio-
mechatronics, Scottsdale, AZ, USA, October 2008, pp.
535–541. IEEE.
Kra
¨mer KHP (2023) Chapter 7: The Pneumatic Cylinder-Part
1. Kornwestheim, Germany: Hafner-Pneumatik KG,
pp.1–11.
Kumar D, Daudpoto J and Chowdhry BS (2020) Challenges
for practical applications of shape memory alloy actuators.
Materials Research Express 7(7): 073001–073012.
Lamuta C, Messelot S and Tawfick S (2018) Theory of the
tensile actuation of fiber reinforced coiled muscles. Smart
Materials and Structures 27(5): 055018.
Lee SJ, Kim YJ, Jeong GH, et al. (2012) Computational analy-
ses of pinching dynamics of a finger exoskeleton composed
of IPMC actuators. International Journal of Precision Engi-
neering and Manufacturing 13(12): 2135–2141.
Lee YK (2014) Design of exoskeleton robotic hand/arm sys-
tem for upper limbs rehabilitation considering mobility
and portability. In: 2014 11th International Conference on
Ubiquitous Robots and Ambient Intelligence (URAI),
Kuala Lumpur, Malaysia, November 2014, pp. 540–544.
IEEE.
Lidka M, Price A and Luisa Trejos A (2018) Development
and evaluation of dielectric elastomer actuators for assis-
tive wearable devices. In: 2018 IEEE Canadian Conference
on Electrical & Computer Engineering (CCECE), Quebec
City, QC, May 2018, pp. 1–4. IEEE.
Li N, Yang T, Yu P, et al. (2018) Bio-inspired upper limb soft
exoskeleton to reduce stroke-induced complications. Bioin-
spiration & Biomimetics 13(6): 066001.
Liu C, Qin H and Mather PT (2007) Review of progress in
shape-memory polymers. Journal of Materials Chemistry
17(16): 1543.
Ma J, Chen D, Liu Z, et al. (2020) A soft wearable exoskeleton
with pneumatic actuator for assisting upper limb. In: 2020
IEEE International Conference on Real-time Computing and
Robotics (RCAR), Asahikawa, Japan, 28 September 2020,
pp. 99–104. IEEE.
Majidi Fard Vatan H, Nefti-Meziani S, Davis S, et al. (2021)
A review: A comprehensive review of soft and rigid wear-
able rehabilitation and assistive devices with a focus on
the shoulder joint. Journal of Intelligent & Robotic Systems
102(1): 9.
Garofalo et al. 25
Manna SK and Dubey VN (2018) Comparative study of
actuation systems for portable upper limb exoskeletons.
Medical Engineering & Physics 60: 1–13.
Maura RM, Rueda Parra S, Stevens RE, et al. (2023) Litera-
ture review of stroke assessment for upper-extremity phys-
ical function via EEG, EMG, kinematic, and kinetic
measurements and their reliability. Journal of Neuroengi-
neering and Rehabilitation 20(1): 21.
Mavroidis C, Charles P and Michael M (1999) Conventional
actuators, shape memory alloys, and electrorheological
fluids. In: Bar-Cohen Y (ed.) Invited Chapter in Automa-
tion, Miniature Robotics and Sensors for Non-Destructive
Testing and Evaluation. pp. 1–26.
Mirvakili SM and Hunter IW (2018) Artificial muscles:
Mechanisms, applications, and challenges. Advanced
Materials 30(6): 1–28.
Moyer E and Chicago U (2010) Basics on Electric Motors.
University of Chicago, pp.1–11.
Murray IA and Johnson GR (2004) A study of the external
forces and moments at the shoulder and elbow while per-
forming every day tasks. Clinical Biomechanics 19(6):
586–594.
Nemat-Nasser S and Wu Y (2003) Comparative experimental
study of ionic polymer–metal composites with different
backbone ionomers and in various cation forms. Journal
of Applied Physics 93(9): 5255–5267.
Noronha B, Ng CY, Little K, et al. (2022) Soft, lightweight
wearable robots to support the upper limb in activities of
daily living: A feasibility study on chronic stroke patients.
IEEE Transactions on Neural Systems and Rehabilitation
Engineering 30: 1401–1411.
Ochieze C, Zare S and Sun Y (2023) Wearable upper limb
robotics for pervasive health: A review. Progress in Biome-
dical Engineering 5(3): 032003.
O
¨zkul Kurgun MA, Kalay E, et al. (2019) Shape memory
alloys phenomena: Classification of the shape memory
alloys production techniques and application fields. Eur-
opean Physical Journal Plus 134(12): 585.
Pang Z, Wang T, Wang Z, et al. (2020) Design and analysis of
a wearable upper limb rehabilitation robot with character-
istics of tension mechanism. Applied Sciences 10(6): 2101.
Pan J, Astarita D, Baldoni A, et al. (2022a) NESM-g:An
upper-limb exoskeleton with compliant actuators for clini-
cal deployment. IEEE Robotics and Automation Letters
7(3): 7708–7715.
Pan M, Yuan C, Liang X, et al. (2022b) Soft actuators and
robotic devices for rehabilitation and assistance. Advanced
Intelligent Systems 4(4): 1–25.
Park SJ, Choi K, Rodrigue H, et al. (2023) Soft exosuit based
on fabric muscle for upper limb assistance. IEEE/ASME
Transactions on Mechatronics 28(1): 26–37.
Paterna M, De Benedictis C and Ferraresi C (2022) The
research on soft pneumatic actuators in Italy: Design solu-
tions and applications. Actuators 11(11): 328.
Pelrine R, Kornbluh R, Joseph J, et al. (2000) High-field
deformation of elastomeric dielectrics for actuators. Mate-
rials Science and Engineering C 11(2): 89–100.
Pitzalis RF, Park D, Caldwell DG, et al. (2023) State of the
art in wearable wrist exoskeletons part I: Background
Needs and Design Requirements. Machines 11(4): 458.
Polygerinos P, Wang Z, Galloway KC, et al. (2015) Soft
robotic glove for combined assistance and at-home rehabi-
litation. Robotics and Autonomous Systems 73: 135–143.
Ponomareva E (2006) Hydraulic and pneumatic actuators
and their application areas. In: Mechatronics - Foundations
and Applications. pp.1–24.
Portescap (2021) Motors for Exoskeleton Applications. Avail-
able at: https://www.portescap.com.
Proietti T, Ambrosini E, Pedrocchi A, et al. (2022) Wearable
robotics for impaired upper-limb assistance and rehabilita-
tion: State of the art and future perspectives. IEEE Access
10: 106117–106134.
Qian W, Liao J, Lu L, et al. (2023) CURER: A lightweight
cable-driven compliant upper limb rehabilitation exoskele-
ton robot. IEEE/ASME Transactions on Mechatronics
28(3): 1730–1741.
Rizzello G (2023) A review of cooperative actuator and sensor
systems based on dielectric elastomer transducers. Actua-
tors 12(2): 46.
Rosenfeld S (2017) Advantages and Drawbacks of Pneumatic,
Hydraulic, and Electric Linear Actuators. pp.1–5.
Rosen J and Ferguson PW (eds) (2020) Chapter 1 - Upper
Limb Exoskeleton Systems - Overview. In: Wearable
Robotics: Systems and Applications. San Diego, CA, USA:
Academic Press, an imprint of Elsevier.
SAES Getters (2023) Shape Memory Alloys/Nitinol. Available
at: https://www.saesgetters.com.
Saharan L, de Andrade MJ, Saleem W, et al. (2017) iGrab:
Hand orthosis powered by twisted and coiled polymer
muscles. Smart Materials and Structures 26(10): 105048.
Saharan L and Tadesse Y (2019) Chapter 3 - novel twisted
and coiled polymer artificial muscles for biomedical and
robotics applications. In: Alina-Maria H and Alexandru
Mihai G (eds) Materials for Biomedical Engineering. Else-
vier. pp.45–75.
Sekhar PK and Uwizeye V (2012) Review of sensor and
actuator mechanisms for bioMEMS. In: Bhansali S and
Abhay V (eds) MEMS for Biomedical Applications. Else-
vier. pp.46–77.
Shankar R, Ghosh T and Spontak R (2007a) Electroactive
nanostructured polymers as tunable actuators. Advanced
Materials 19(17): 2218–2223.
Shankar R, Ghosh TK and Spontak RJ (2007b) Electrome-
chanical response of nanostructured polymer systems with
no mechanical pre-strain. Macromolecular Rapid Commu-
nications 28(10): 1142–1147.
Suzuki M and Kamamichi N (2018) Control of twisted and
coiled polymer actuator with anti-windup compensator.
Smart Materials and Structures 27(7): 075014.
Theadom A, Rodrigues M, Roxburgh R, et al. (2014) Preva-
lence of muscular dystrophies: A systematic literature
review. Neuroepidemiology 43(3-4): 259–268.
Tschiersky M, Hekman EEG, Brouwer DM, et al. (2020) A
compact McKibben muscle based bending actuator for
close-to-body application in assistive wearable robots.
IEEE Robotics and Automation Letters 5(2): 3042–3049.
Varghese RJ, Freer D, Deligianni F, et al. (2018) Wearable
robotics for upper-limb rehabilitation and assistance. In:
Kai-Yu Tong R (ed.) Wearable Technology in Medicine
and Health Care. London: Elsevier. pp.23–69.
26 Journal of Intelligent Material Systems and Structures 00(0)
Wang Z, Yang J and Shengqiang C (2022) Liquid crystal elas-
tomers for soft actuators. Materials Lab 1: 1–4.
Weber LM and Stein J (2018) The use of robots in stroke
rehabilitation: A narrative review. NeuroRehabilitation
43(1): 99–110.
WEG (2023) Specification on Electric Motors. Available at:
https://www.weg.net.
Xiloyannis M, Alicea R, Georgarakis A-M, et al. (2022) Soft
robotic suits: State of the art, core technologies, and open
challenges. IEEE Transactions on Robotics 38(3):
1343–1362.
Yuan J and Poulin P (2014) Materials science. Fibers do the
twist. Science 343(6173): 845–846.
Zaidi S, Maselli M, Laschi C, et al. (2021) Actuation technol-
ogies for soft robot grippers and manipulators: A review.
Current Robotics Reports 2(3): 355–369.
Zhang H, Lin Z, Hu Y, et al. (2023) Low-voltage driven ionic
polymer-metal composite actuators: Structures, materials,
and applications. Advanced Science 10(10): 1–20.
Zhang J, Sheng J, O’Neill CT, et al. (2019) Robotic artificial
muscles: Current progress and future perspectives. IEEE
Transactions on Robotics 35(3): 761–781.
Zhao KD, Robinson CA and Hilliard MJ (2019) Biomecha-
nics of the upper limb. In: Webster JB and Murphy DP
(eds) Atlas of Orthoses and Assistive Devices. Elsevier.
pp.127–133.
ZHAOWEI (2023) Gear Motors. Available at: https://www.
zwgearbox.com.
Zhu M, Do TN, Hawkes E, et al. (2020) Fluidic fabric muscle
sheets for wearable and soft robotics. Soft Robotics 7(2):
179–197.
Zwerus EL, Willigenburg NW, Scholtes VA, et al. (2019) Nor-
mative values and affecting factors for the elbow range of
motion. Shoulder & Elbow 11(3): 215–224.
Appendix
Glossary
Alternative Current (AC): flow of electric charge in a
circuit that periodically reverses direction in a regular
manner
Activities of Daily Living (ADLs): routine tasks that
individuals typically perform as part of their daily self-
care and independent living
Carbon NanoTubes (CNTs): cylindrical nanostructures
composed of carbon atoms arranged in a hexagonal
lattice patter, forming seamless tubes with diameters on
the order of nanometers. These structures can be
Single-Walled (SWCNTs), consisting of a single layer
of carbon atoms, or Multi-Walled (MWCNTs), com-
prising multiple concentric layers of carbon sheets
Direct Current (DC): flow of electric charge in a circuit
in one direction with constant magnitude
Degrees Of Freedom (DOFs): number of independent
movements along which a joint can move
Electro-Active Polymers (EAPs): class of smart materi-
als that exhibit significant changes in shape, size, or
mechanical properties in response to electrical stimula-
tion. Examples of this kind of materials are
Electrostrictive Polymers (EPs), Ferroelectric Polymers
(FPs), Liquid Crystal Elastomers (LCEs), Dielectric
Elastomers (DEs), Conductive Polymers (CPs), Ionic
Polymer Gels (IPGs) and Ionic Polymer Metal
Composites (IPMCs)
ElectroEncephaloGraphy (EEG): electrical impulses
generated by brain’s neurons and recorded from elec-
trodes placed on the scalp to provide valuable insights
into brain function, neural dynamics and cognitive
processes
ElectroMyoGraphy (EMG): electrical signals produced
by skeletal muscles during muscle contractions and
used to record and analyze muscle function, movement
pattern and neuromuscular control
Fluidic Fabric Muscle Sheets (FFMS): soft robotic
actuators that consist of flexible, fabric-like materials
embedded with fluidic channels that are filled with a
fluid and can be selectively pressurized to induce defor-
mation and generate motion
Hydraulic Artificial Muscles (HAMs): actuators that
utilize the principle of fluid mechanics (liquid such as
water or oils) to generate motion and force
Inverse Pneumatic Artificial Muscles (IPAMs): actua-
tors that operate on the principal of pneumatic power,
but with a configuration opposite to that of traditional
PAMs, expanding when actuated
One-Way Shape Memory Effect (OWSME): phenom-
enon that allows certain materials to recover their origi-
nal shape from a deformed state upon heating, but
cannot reverts back to the deformed shape upon subse-
quent cooling
Pneumatic Artificial Muscles (PAMs): actuators that
mimic the function of biological muscles using com-
pressed air or gas as a power source, contracting when
supplied
PseudoElasticity (PE): unique property observed in cer-
tain material that enables them to undergo large and
reversible deformations while maintaining their original
shape upon unloading, and for that it is also known as
SupereElasticity (SE)
Range of Motions (ROMs): expressed freedom in
degrees that each joint can perform in space
Series Elastic Actuators (SEAs): type of actuator com-
monly used in robotics that incorporate an elastic ele-
ment (spring) in series with a motor and a load
Shape Memory Alloys (SMAs): class of metallic materi-
als that exhibit the unique property known as SME
Shape Memory Effect (SME): phenomenon observed in
certain materials that allows them to ‘‘remember’’ and
recover their original shape after being deformed
Garofalo et al. 27
Shape Memory Polymers (SMPs): class of smart mate-
rials that possess the ability to change shape in response
to external stimuli, such as change in temperature, light
or humidity
Twisted and Coiled Artificial Muscles (TCAMs): also
known as Twisted and Coiled Polymers (TCPs), they are
a type of soft robotic actuator and consist of twisted
and coiled fibers made from smart materials that exhibit
actuation behavior when subjected to external stimuli
Two-Way Shape Memory Effect (TWSME): property
observed in certain shape memory materials that allows
them to undergo reversible shape transformations
between two distinct shapes upon changes in
temperature
Upper Limb Exosuit (ULE): wearable robotic device
designed to assist or augment the movement and func-
tion of the human upper limbs
28 Journal of Intelligent Material Systems and Structures 00(0)
A preview of this full-text is provided by SAGE Publications Inc.
Content available from Journal of Intelligent Material Systems and Structures
This content is subject to copyright.
