Proceedings of the 20th World Congress

Proceedings of the
The International 20th World
Federation of Congress
Automatic Control
Proceedings
The
Proceedings of
of the
International 20th World
Federation
the 20th Worldof Congress
Automatic Control
Congress
Toulouse, France, July 9-14, 2017 Available online at www.sciencedirect.com
The
The International
Toulouse, France,Federation
International Federation of
of Automatic
Automatic Control
July 9-14, 2017 Control
Toulouse, France, July 9-14, 2017
Toulouse, France, July 9-14, 2017
ScienceDirect
Review
Review of of Upper
Upper Limb Limb HybridHybrid Exoskeletons
IFAC PapersOnLine 50-1 (2017) 15169–15178
Exoskeletons
Review
Review of
of Upper
Upper Limb
Limb Hybrid
Hybrid Exoskeletons
Exoskeletons
Ashley M. Stewart, Christopher G. Pretty, Mark Adams, XiaoQi Chen
Ashley M. Stewart, Christopher G. Pretty, Mark Adams, XiaoQi Chen
 Pretty,
Ashley
Ashley M. M. Stewart,
Stewart, Christopher
Christopher G.  Pretty, Mark
G. Mark Adams,
Adams, XiaoQi XiaoQi Chen Chen
Mechanical Engineering Department,  University of Canterbury,
Mechanical Engineering Department, University of Canterbury,
Mechanical
Mechanical Engineering Department,
Christchurch,
Engineering Department, University
University of
New Zealand of Canterbury,
Canterbury,
Christchurch, New Zealand
Christchurch, New
Christchurch, New Zealand Zealand
Abstract: Hybrid exoskeletons, which combine electrically controlled actuation with functional electrical
Abstract: Hybrid exoskeletons, which combine electrically controlled actuation with functional electrical
stimulation, potentially
Abstract: offer great benefits for muscular rehabilitation. This review aims to provide an
Abstract: Hybrid
stimulation, Hybrid exoskeletons,
exoskeletons,
potentially offer great which
which combine
combine
benefits electrically controlled
controlled actuation
electricallyrehabilitation.
for muscular actuation
This reviewwith
with functional
functional
aims to provide electrical
electricalan
overview
stimulation, of the state of the art of current upper-limb hybrid exoskeletons with a focus on stroke
stimulation,
overview of potentially
potentially
the state of offer
thegreat
offer great
art ofbenefits
currentfor
benefits muscular
muscular rehabilitation.
forupper-limb rehabilitation.
hybrid exoskeletons This
This review with aims
review aims
a focus to
to provide
provide
on stroke an
an
rehabilitation.
overview of Thisstate
the fieldofis still
the veryofnew
art and further
current upper-limbdevelopment
hybrid of the current control
exoskeletons with a methods
focus onused for
stroke
overview of the
rehabilitation. Thisstate
fieldofis still
the artveryofnew current upper-limb
and further hybrid of
development exoskeletons
the current controlwith a methodsfocus onused stroke
for
hybrid exoskeletons
rehabilitation. This is needed. More investigation is needed with regards to the potential benefit of
rehabilitation.
hybrid exoskeletons This field
field is
is still
still very
is needed. very
More new and
and further
newinvestigationfurther development
development
is needed with of
of the
the current
current
regards to control
control methods
methods
the potential used
used for
benefit for
of
hybrid exoskeletons
hybrid exoskeletons
exoskeletons as a patient-monitoring and rehabilitation assist-as-need tool.
hybrid exoskeletons asis isaneeded.
needed. More
More investigation
patient-monitoring investigation is
is needed
and rehabilitation with
with regards
neededassist-as-need regardstool. to
to the
the potential
potential benefitbenefit of of
©hybrid
2017, exoskeletons
hybrid exoskeletons
IFAC (International as
as aa patient-monitoring
patient-monitoring
Federation of Automatic and
and rehabilitation
rehabilitation assist-as-need
by Elseviertool.
assist-as-need
Control) Hosting tool.
Ltd. All rights reserved.
Keywords: Rehabilitation engineering and healthcare delivery, stroke, hybrid exoskeletons, functional
Keywords: Rehabilitation engineering and healthcare delivery, stroke, hybrid exoskeletons, functional
electrical
Keywords: stimulation,
Rehabilitation upperengineering
limb and
Keywords:stimulation,
electrical Rehabilitation upperengineering
limb and healthcare
healthcare delivery,
delivery, stroke,
stroke, hybridhybrid exoskeletons,
exoskeletons, functionalfunctional
electrical stimulation,
electrical stimulation, upper limb upper limb 

 occur for exoskeleton users. Thus, it is important that an
1. INTRODUCTION occur for exoskeleton users. Thus, it is important that an
1. INTRODUCTION exoskeleton
occur designed users.
for exoskeleton
exoskeleton for Thus,rehabilitation
it isis importantpurposes
important that an is
1. INTRODUCTION occur for
exoskeleton designed users. for Thus, it
rehabilitation purposes that an
is
Stroke is the second 1. INTRODUCTION
largest cause of disability worldwide lightweight exoskeleton as well as quick and easy to attach as there is no
Stroke is the second largest cause of disability worldwide lightweight exoskeleton as designeddesigned
well as quick for
forandrehabilitation
rehabilitation
easy to attach as purposes
purposes
there is no is
is
after dementia (Worldlargest HeartcauseFederation, Effective point
2015).worldwide in having a well-working exoskeleton if patients refuse
Stroke is
is the
Strokedementia
after the second
second
(Worldlargest Heartcause of
of disability
Federation, disability Effective lightweight
2015).worldwide lightweight
point in havingas
as well
well as
as quick
quick and
a well-working easy
easy to
andexoskeleton
use it.in having a well-working exoskeleton if patients refuse
to attach
attach as
as there is
is no
thererefuse
if patients no
rehabilitation
after after a stroke is important if the costs Effective
of stroke point
after dementia
dementiaafter
rehabilitation (World
(World
a stroke Heart
Heart Federation,
Federation,
is important if the2015).
2015).
costs Effective
of stroke use pointit.in having a well-working exoskeleton if patients refuse
on society areafter
rehabilitation to bea minimized.
stroke is Stoke survivors
important if the costsoftenof suffer use it.
stroke
rehabilitation
on society areafter to bea minimized.
stroke is important if the costs
Stoke survivors often suffer use
of stroke it.
from
on hemiparesis. Regaining strength and movement in the Assessment tools currently used by physicians to monitor
on society
fromsociety are
are to
hemiparesis. to be minimized.
beRegaining
minimized. Stoke
Stoke and
strength survivors
survivors
movementoften
often in suffer
suffer
the Assessment tools currently used by physicians to monitor
affected side takes Regaining
time and can be improved with the use of patient progress tend to be time-consuming, subjective, and
from
from hemiparesis.
hemiparesis.
affected side takes Regaining
time and can strength
strength
be improvedand
and movement
movement
with the use in
in theof Assessment
the Assessment
patient
lacking
progress
in
tools
toolstend
sensitivity
currently
currently used
used by
to be time-consuming,
(Loureiro et
by physicians
al.,
physicians
2011).
to
to monitor
subjective,
When
monitorand
patient
rehabilitation therapy involving repetitive and function-
affected
affected side
side takes
rehabilitation takes time and
and can
time involving
therapy can be improved
improved with
be repetitive and the
with the use
use of
function- of patient
patient progress
lacking progress
in sensitivity tend
tend to to be
be time-consuming,
(Loureiro et al., 2011).subjective,
time-consuming, subjective,
When patient and
and
specific tasks (Senelick,
rehabilitation therapy 2010).
involving repetitive and function- improvements
lacking in are small
sensitivity some ofetthe
(Loureiro al.,more
2011). commonly
When used
patient
rehabilitation therapy
specific tasks (Senelick, 2010). involving repetitive and function- lacking in sensitivity (Loureiro
improvements are small some of the more commonly used et al., 2011). When patient
scales are onlyaremoderately
improvements small reliable (Patton et al., 2008).
specific
specific tasks
tasks (Senelick,
(Senelick, 2010).2010). improvements
scales are onlyaremoderatelysmall some of
of the
somereliable more
more commonly
the (Patton commonly
et al., 2008). used
used
Despite the high requirement and expected increase in scales Roboticaredevices only offer new, more
moderately reliable sensitive,
(Patton and
et consistent
Despite the high requirement and expected increase in Robotic scales aredevices only moderately
offer new, more reliable (Patton and
sensitive, et al.,
al., 2008).
2008).
consistent
demand for at-home rehabilitationand systems and devices which options for monitoring of patient improvement but consistent
they need
Despite
Despite for
demand the high
high requirement
theat-home requirement
rehabilitationand systemsexpected
expected
and devicesincrease
increase which in Robotic
in Robotic for
options devices
devices
monitoring offer
offer of new,
new, more sensitive,
sensitive, and
moreimprovement
patient and
but consistent
they need
support foractivities ofrehabilitation
daily living,systems many and of devices
the currently to be able
options for tomonitoring
provide measures of patient that can be easily
improvement but understood
they
demand
demand for
support at-home
at-homeofrehabilitation
activities daily living,systems many and of devices
the currentlywhich
which to options
be able fortomonitoring
provide measures of patientthat improvement
can be easily they need
butunderstood need
available
support rehabilitation
activities of daily robotic
living, devices
many ofarethe extremely
currently by
to physicians
be able to and related
provide to current
measures that methods.
can be Withunderstood
easily regards to
support activities
available rehabilitationof dailyroboticliving, devices
many ofarethe extremely currently by to be able to provide
physicians and related measures
to currentthat methods.
can be easily Withunderstood
regards to
expensive, lacking in portability, complexareto extremelyuse, and evaluation of the effect of torobotic systems on patient progress
available
available rehabilitation
expensive, rehabilitation
lacking in portability,robotic
robotic devices devices
complexareto extremelyuse, and by by physicians
physicians
evaluation
there is
of the
little
and
consensus
related
andeffect
related of to
and
current
current
robotic
a wide
methods.
methods.
systems
variety
With
With regards
on patient
of
regards
measures
progress
have
to
to
therefore
expensive, only suitable
lacking in to use in clinics
portability, under
complex the tosupervision
use, and evaluation of the effect of robotic systems on patient progress
expensive,only
therefore lacking
suitableintoportability,
use in clinicscomplexunder thetosupervision
use, and there evaluation of consensus
is little the effect of and robotic
a wide systems
varietyonofpatient
measures progress
have
of physicians (Lu et toal.,use2011, Maciejasz al., 2014). been
et supervision used (Sivan et al., 2011, Del-Ama et al., 2012, Lu ethave al.,
therefore
therefore
of only
only suitable
physicians suitable
(Lu et toal.,use2011, in
in clinics
clinics under
under the
Maciejasz the al., 2014). there
et supervision there
been used
2011,
is
is little
little consensus
consensus
(Sivan
Maciejasz
and
and aa Del-Ama
et al., 2011,
et al., 2014).
wide
wide variety
There
variety
is
of
et al.,
also a
measures
of 2012,
measures
lack
Lu ethave
of wide-
al.,
Improvement of patient’s access to at home rehabilitation been
of
of physicians
physicians of(Lu
Improvement (Lu et
et al.,
patient’s 2011,
al., access
2011, to Maciejasz
at home et
Maciejasz al.,
al., 2014).
etrehabilitation
2014). 2011, been used
used (Sivan
(Sivan et
Maciejasz et al., 2011,
et al., 2014).Del-Ama
2011, There is et
Del-Ama al.,
al., a2012,
etalso lack Lu
2012, Lu et
et al.,
of wide- al.,
devices shouldof lead to an increase in the frequency of training 2011, scale assessments
Maciejasz of robotic devices which can likely be
Improvement
Improvement
devices shouldof leadpatient’s
patient’s access
access into
to an increase totheat home
home rehabilitation
at frequency rehabilitation
of training scale Maciejasz et
2011, assessments al.,
et of 2014).
al.,robotic There
There is
2014). devices also
also aa can
is which lack of
of wide-
lack likely wide- be
(Lu et al., 2011). As it has beeninshown repeatedly that an scale contributed to the high
assessments of cost to produce
robotic devices the
which devices
can (Loureiro
likely
devices
devices
(Lu should
should
et al., 2011). lead
leadAs to an
to an increase
increase
it has beeninshownthe frequency
the frequency
repeatedly of training
of training scale assessments
that an contributed to the high of robotic devices the
cost to produce which devices likely be
can (Loureiro be
increased dosage of training is highly positively correlated et al., 2011).to the high cost to produce the devices (Loureiro
(Lu
(Lu etet al.,
increased 2011).
al.,dosage
2011). of As it
it has
Astraininghas been
been shown repeatedly
shownpositively
is highly repeatedly that
that an
correlated an contributed
contributed
et al., 2011).to the high cost to produce the devices (Loureiro
with patient
increased improvement
dosage of we is should then also expect better et al., 2011).
increased
with patientdosage of training
improvement trainingwe is highly
highly
should thenpositively
positively
also expect correlated
better et
correlated al., 2011).
patient
with outcomes with a higher uptake of at-home The ability of a rehabilitation robot to assist-as-need is highly
with patient
patientpatient improvement
improvement
outcomes with we a should
we higher then
should then also
also expect
uptake expect
of at-home better
better The ability of a rehabilitation robot to assist-as-need is highly
rehabilitation devices (Vaca Benitezuptake et al., of2013). To desirable in stroke rehabilitation robotics as it can encourage
patient
patient outcomes
outcomes
rehabilitation devices with
with(Vaca aa higher
higher
Benitezuptake et al., of2013). at-home
at-home To The The ability
abilityinof
desirable
the patient to
aa rehabilitation
ofstroke rehabilitation
make
rehabilitation
the effort
robot
robot to
to assist-as-need
robotics
to
assist-as-need
achieve
is
is highly
as it can encourage
movement
highly
rather
encourage
rehabilitation a higher
devices uptake (Vacaof at-home
Benitez rehabilitation
et al., devices
2013). To desirable in stroke rehabilitation robotics as it can encourage
rehabilitation devices (Vaca Benitez
encourage a higher uptake of at-home rehabilitation devices the et al., 2013). To desirable in stroke rehabilitation robotics
patient to make the effort to achieve movement rather as it can encourage
there needs to be a focus on designing rehabilitation devices, than learning to rely on the robot to perform the movement
encourage
there needsaatohigher
encourage higher uptake
uptakeonof
be a focus at-home
ofdesigning rehabilitation
at-home rehabilitation devices the
devices
rehabilitation devices, thanpatient
the patient
learning to
to tomake
makerely the
the effort
effort
on the robot to
to toachieve
achieve
perform movement
movement
the movement rather
rather
especially exoskeletons, which are cheap, portable, and easy than (Lu etlearning
al., 2011, to Jarrassé
rely et al., 2014, Maciejasz et movement
al., 2014).
there
there needs to
to bebe aa focus
needsexoskeletons,
especially focuswhichon
on designing
designing
are cheap, rehabilitation
rehabilitation
portable, and devices,
devices,
easy (Lu thanetlearning
al., 2011, rely on
to Jarrassé on the
the robot
robot
et al., 2014, to perform
perform the
to Maciejasz theet movement
al., 2014).
to use. Appropriately timed action is more important than strength
especially
especially
to use. exoskeletons,
exoskeletons, which which are are cheap,
cheap, portable,
portable, and easy (Lu
and easy (Lu et et al.,
al., 2011,
Appropriately 2011, Jarrassé
Jarrassé
timed actionet
et al.,
al., 2014, Maciejasz
2014, important
is more Maciejasz than et
et al., 2014).
al.,strength
2014).
to use. for functional gains
Appropriately timed however
action isrepetitive
more practice than
important whichstrength
builds
to use. Appropriately
for functional gains timedhowever
action isrepetitive
more importantpractice than whichstrength
builds
A study (Biddiss and Chau, 2008) which surveyed a range of strength for without
functional gainsspecific
howeverfunctional
repetitive application
practice can
which stillbuilds
help
A study (Biddiss and Chau, 2008) which surveyed a range of strength for functionalwithout gains however
specific repetitive
functional practice can
application which stillbuilds
help
prosthetic hand users indicated that the most important to diminish impairment (Patton et al., 2008). The ability of
A
A study
study (Biddiss
prosthetic hand and
(Biddiss usersChau,
and Chau, 2008)
2008) which
indicated which
that the surveyed
most aaimportant
surveyed range
range of of strength
strength
to diminish without
without specific functional
specific (Patton
impairment functional application
application
et al., 2008). The can still
still help
can ability help
of
requirement as defined by the users was a reduction in the rehabilitation robots to adapt to different users and even to
prosthetic
prosthetic hand
requirement hand users
users by
as defined indicated
indicated
the users that
thatwas the most
most important
thea reduction in the to
important to diminish
diminish impairment
rehabilitation
the same
impairment
user
robots to adapt
on
(Patton
(Patton
different
to et
days
et al.,
al., 2008).
different
or
2008).usersThe
throughout
Theandability
the
even of
ability
same
of
to
weight of the device. A low-cost prosthetic was ranked as the
requirement
of theas
requirement
weight defined
asdevice.
defined by
by the
A low-costthe users
users was
was aawas
prosthetic reduction
ranked in
reduction the rehabilitation
as the
in the same userrobots
rehabilitation robots to
to adapt
on differentadapt daysto
to different
different users
users and
or throughout andtheeven
evensame to
to
third
weight most
of the important
device. A feature.
low-cost Both weightwas
prosthetic andrankedcost aswerethe session
the same is also
user important
on different withdays regardsor to minimising
throughout the set-up
same
weight of the device. A low-cost prosthetic
third most important feature. Both weight and cost were session was ranked as the the same user on different days or
is also important with regards to minimising set-up throughout the same
considered more important than precise movement of the session time andiscost also of rehabilitation (Maciejasz et al., 2014). set-up
third
third most
most more
considered important
important feature.
feature.
important thanBoth
Both weight
weight
precise and
and cost
movement costofwere
were
the time session alsoofimportant
andiscost important
rehabilitation with regards
regards to
with(Maciejasz toetminimising
minimising
al., 2014). set-up
device. While this important
study was than of prosthetic hand users and not
considered
considered
device. While more
more this important
study was than precise
precise hand
of prosthetic movement
movement
users and of the time and cost of rehabilitation (Maciejasz et
the
of not time and cost of rehabilitation (Maciejasz et al.,
al., 2014).
2014).
exoskeleton
device. While users
this it is quite
study was likely
of that a hand
prosthetic similar
users trend
and will
not
device. Whileusers
exoskeleton this study was of
it is quite prosthetic
likely that a hand users
similar andwill
trend not
exoskeleton
exoskeleton users users it it is
is quite
quite likely
likely thatthat aa similar
similar trend
trend willwill
Copyright
2405-8963 ©© 2017,
2017 IFAC 15734
IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
Copyright © 2017 IFAC 15734
Peer review
Copyright © under
2017 responsibility
IFAC of International Federation of Automatic
15734Control.
Copyright © 2017 IFAC
10.1016/j.ifacol.2017.08.2266 15734
Proceedings of the 20th IFAC World Congress
15170
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

Compliance of assistive robots is another important factor to With these features, hybrid exoskeletons have potentially
consider as the robot should aid and encourage but not limit great benefits for upper-limb stroke rehabilitation. However,
the movement of the patient (Maciejasz et al., 2014, Loureiro there is little literature reporting the use of such devices. This
et al., 2011). This is important for safety as well. The paper reviews the state of the field and proposes that there is
exception to this is robots deliberately used to temporarily a need for a portable, hybrid, assist-as-need exoskeleton for
constrain certain joints or for resistive training, however upper-limb stroke rehabilitation. This paper is organized in
consideration of the robot’s interaction with the user in a safe four sections. Section 2 describes the scope of this review.
manner is still the most important feature. Above all else the Section 3 details the literature search methodology and
robot should pose no harm to the user or nearby individuals. presents the results. Section 4 discusses the current research
gaps, and outlines potential future areas of research relating
Actively actuated exoskeletons offer advantages in their to upper-limb hybrid exoskeletons.
ability to repetitively and precisely provide assistance or
2. SCOPE OF THIS REVIEW
resistance to a user. However, portability is often limited
because the actuators which produce the required forces tend A review has previously been conducted on hybrid
to be heavy and have high power requirements. In addition, exoskeletons for lower limb rehabilitation of spinal injury
exoskeletons do not help to combat muscle atrophy if the patients (Del-Ama et al., 2012). No reviews could be found
patient provides no force themselves. Thus exoskeletons are that cover current actuated upper limb hybrid exoskeletons. It
more useful for the rehabilitation of patients who already should be noted that in stroke rehabilitation “often, more
have some ability to move their limb (Hara, 2013). severe impairments are found in the upper limb” (Lu et al.,
2011) yet rehabilitation research focus is largely placed on
Functional Electrical Stimulation (FES) allows for a much recovery of function in the lower limb.
lighter device which is also better suited to reducing muscle
atrophy in patients with no or extremely limited movement. A recent review was conducted on hybrid robotic systems for
However, precise control of FES is extremely difficult and upper limb stroke rehabilitation (Resquín et al., 2016),
producing the specific, repetitive, and functional movement however it predominately covered passive devices and
desired for effective stroke rehabilitation is not easily actively actuated end-effector devices. No actively actuated
accomplished. Furthermore extended use of FES is limited by exoskeletons were reviewed. This review focuses on
the introduction of muscle fatigue caused by the unnatural exoskeletons as opposed to end-effector devices and also
motor unit recruitment order (Doucet et al., 2012). The forces only looks at hybrid active devices. That is those that are
required for large movements, such as shoulder abduction, capable of actively moving the patient’s limb and in this case
are too great to be provided by the use of FES which is much involve the use of both FES and an electrically controlled
better suited to smaller movements such as finger extension actuator. The FES and actuator do not necessarily need to be
(Schill et al., 2011, Pylatiuk et al., 2009a). Some patients also operating on the same joint. Papers that were not written in
find the use of FES painful. English were excluded from this review.

Hybrid exoskeletons, as defined in this paper, are 3. METHOD AND RESULTS

exoskeletons which use an electrically controlled actuator
(e.g. electric motor, pneumatics, hydraulics, e.t.c) in A search was conducted in the IEEE, PubMed,
combination with FES to provide active assistance/resistance ScienceDirect, and Google Scholar databases for relevant
to the user. A hybrid exoskeleton can potentially result in a papers added prior to October 2016. A total of 26 articles
device which is overall lighter (and thus more portable) than were reviewed covering 12 different exoskeletons. Table 1
a non-hybrid exoskeleton and which can produce greater presents a summary of these 12 reviewed upper limb hybrid
forces and more precise control than FES on its’ own. As exoskeletons.
some of the forces are provided by the actuator the FES The results are discussed in six subsections based on the key
should not need to be as intense to produce the same features of exoskeletons in stroke rehabilitation. The
movement thus reducing the amount of pain at the electrode subsections are as follows: subsection 3.1 discusses the
sites and delaying muscle fatigue (Doucet et al., 2012). suitability of hybrid exoskeletons to aid movement in the
different joints. Subsection 3.2 presents the need and ability
The combination of devices also potentially offers a greater of hybrid-exoskeletons to provide portable aid.
ability to precisely monitor patient progress and Subsection 3.3 addresses hybrid exoskeletons as a patient
automatically fine tune the amount of assistance given to the monitoring tool. Subsection 3.4 compares the types of
patient. Electromyography (EMG) can provide information actuators used. Subsection 3.5 discusses the various control
on the muscle fatigue of the patient which, when combined schemes used. Subsection 3.6 presents an evaluation of the
with the consistency in movement produced by the actuators, usability of the current upper-limb hybrid exoskeletons and
could be useful for tracking patient progress across several the clinical evidence regarding the benefits and costs of
sessions (Cifrek et al., 2009). This information can then be hybrid exoskeletons.
used to control the amount of support which the exoskeleton
gives to the patient, as well as the FES parameters, and ratio
of FES support to mechanical actuator support, with the aim
of giving the patient an appropriately challenging workout.

15735
 Table 1: Reviewed Upper Limb Hybrid Exoskeletons
System Supported Portable Stage of FES Actuator Main Control Control Method Feedback Advantages Possible Improvements Type Weight
Movements Development Input
Orthojacket Shoulder, elbow Fixed to Investigative study Motion Stepper sEMG and/or sEMG proportional Speed, Modularity. Full device has not yet been SCI 33.2 g –
(Schill et al., (actuator only) wheelchair on 3 healthy stim 8 Motors joystick control position FES electrode array. tested. Only testing of the 700 g
2011) Wrist and fingers Elbow & subjects & 1 SCI (wrist & (Shoulder) Can be worn under modules have been done. elbow
(FES only) wrist sections subject (elbow fingers) Fluidic clothing. No fatigue monitoring. section
can be made only) actuators
portable (elbow)
Haptic Robotic Fingers, thumb No Prototype Motion Firgelli linear Key/button Button proportional Position Allows independent No fatigue monitoring. Stroke 56 g
Glove(Hartopanu stim 8 actuators press control control of fingers. Response time is slow. actuator
et al., 2015)
Toulouse, France, July 9-14, 2017

FES/Robot Hand Fingers No Pilot clinical study 2 channel Firgelli linear sEMG sEMG proportional sEMG Shown to improve Some patients had trouble Stroke Not
(Rong et al., 2013) on 4 stroke subjects stimulator actuators control movement accuracy as with the weight of the stated
over 20 sessions well as voluntary effort device.
and subject focus. No fatigue monitoring.
Wearable Shoulder, fingers No Prototype Beurer Pneumatic Key/button PID control of a Position Demonstrates that Predefined trajectory. Stroke Not
Proceedings of the 20th IFAC World Congress

Rehabilitation EM41 muscles press predefined hybrid exoskeletons can No fatigue monitoring. stated
Robot (Tu et al., trajectory reduce the weight of the
2012) device with no
performance loss.
EXOSLIM (Serea Shoulder, elbow No Prototype Motion DC motors Key/button Button proportional Position, No fatigue monitoring. SCI 3 kg
et al., 2015) stim 8 press control acceleration

BCI-Controlled Elbow, wrist Yes Investigative study Reha Stim BLDC motor EEG (Emotiv State machine with Position Easy to put on (<30 s Predefined trajectory. Stroke 1 kg
Exoskeleton (actuator only) on 9 stroke subjects 1 (elbow) EPOC proportional control with help) No fatigue monitoring.
(Elnady et al., Fingers (FES over 2 sessions 7 (fingers) Brushed DC neuroheadset) of predefined
2015) only) days apart motor (wrist) trajectories
MUNDUS Shoulder, elbow No. Fixed to Investigative study Reha Stim DC motors EEG (other Intergral control, or Position, RFID tags for object Trial was unsuccessful for SCI 2.2 kg
(Pedrocchi et al., (FES only) wheelchair on 1 SCI subject for 1 (fingers) control options Biomimetic force, recognization. the hybrid part.
2013, Ambrosini Fingers (actuator Restricted to the hybrid part (shoulder available as feedforward object FES electrode array Takes 35-40 minutes to set

15736
et al., 2009, only or FES table (unsuccessful) & elbow) well, e.g. sEMG control, or Adaptive location Modularity. up and calibrate.
CROSTA, 2012) only, not both) or head/eye feedforward control
movement)
BCI-Controlled Elbow, wrist Yes Investigative study EMPI300 BLDC motor EEG State machine with Position, Easy to put on (<30 s Predefined trajectory. SCI Not
Wearable Robot (actuator only) on 5 healthy (fingers) (elbow, wrist) proportional control acceleration with help) Not backdrivable. stated
(Looned et al., Fingers (FES subjects of predefined No fatigue monitoring.
2014) only) trajectories
RUPERT Shoulder Unknown Investigative study Reha Stim Pneumatic sEMG sEMG PID control sEMG, No fatigue monitoring. Stroke Not
(Tu et al., 2014) (actuator only) (tethered?) on 3 healthy 2 muscles position, stated
Elbow subjects pressure,
Fingers (FES Kinect
only)
THINK2GRASP Elbow (actuator No An investigative Motion Electric motor EEG + joystick State machine for EEG, Neoprene sleeve with Subject required trunk SCI Not
(Rohm et al., only and not all study was done on stim 8 (elbow) type of movement joystick Velcro attached restraint in order to produce stated
Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

2012, Rohm et al., users) hybrid portion of (fingers) (EEG) electrodes. normal correct movement.
2013, Rupp et al., Fingers (FES the device with 2 Proportional control Provides visual Takes at least an hour to set
2013) only) SCI subjects for degree of feedback to the patient. up and the device caused
movement Modular. mental fatigue due to
(joystick) complex control. Control is
switched between elbow
and hand.
Early Hybrid Elbow (actuator Fixed to Investigative study 2 channel Electric motor Voice Neural network Force, Force feedback is Requires movement in SCI Not
Exoskeleton only) wheelchair on 5 quadriplegic stimulator (elbow) processes pretrained position, provided to users. the shoulder. stated
(Varoto et al., Fingers (FES subjects voice commands 3D visual
2008) only) motion
capture
ALEx Shoulder, elbow No Investigative study Reha Stim Electric motor GUI - key press Impedance control Position Back-drivable. No fatigue monitoring. Stroke Not
(Crema et al., (actuator only) on 1 healthy patient (shoulder, Adjustable Possible misalignments stated
2015) Fingers (FES of the elbow and elbow) assistance/resistance. with the shoulder of the
only) shoulder modules subject and not adaptable to
different sized patients.
15171
Proceedings of the 20th IFAC World Congress
15172
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

so the actuators did not need to be situated on the subject’s

3.1 Supported Movements hand (Hartopanu et al., 2015). Initial validation of the Haptic
Robotic Glove appears to have been successful with regards
Of the 12 upper-limb hybrid exoskeletons reviewed in this to performing the desired movement however the device was
paper only five use FES and an electrically controlled limited by its slow response time. Firgelli linear actuators
actuator for the same joint. With one exception, in cases were also used for the actuation of the FES/Robot Hand
where FES and the actuator are used for different joints the (Rong et al., 2012, Rong et al., 2013) and in this case some
FES is used for the fingers and the actuator is used for the subjects found the device too heavy due to the placement of
elbow and shoulder. FES induced fatigue is considerably less the actuators on the hand. Despite this the FES/Robot Hand
pronounced for movement of smaller limbs, such as the was able to demonstrate a decrease in the muscular effort
fingers. Table 2 shows the distribution of the actuation type required to perform tasks and a decrease unnecessary
by joint for the reviewed hybrid exoskeletons. muscular activity. The Wearable Rehabilitation Robot (Tu et
al., 2012) successfully demonstrated the ability of hybrid
Table 2: Distribution of actuation type by joint. exoskeletons to reduce the weight and improve the accuracy
when compared respectively to an actuator or FES alone for
Actuation Type Fingers Wrist Elbow Shoulder grasping. Actual weight reduction of the exoskeleton has yet
Hybrid 3 0 2 2 to be done however.

Electrically 1 2 6 3 Three exoskeletons use hybrid actuation for the shoulder

controlled and/or elbow. The Wearable Rehabilitation Robot (Tu et al.,
actuator only 2012) successfully demonstrated the advantages of hybrid
actuation for shoulder movement with regards to weight
FES only 7 1 2 1 reduction and energy requirements. The EXOSLIM (Serea et
al., 2013, Serea et al., 2014, Serea et al., 2015) and RUPERT
(Tu et al., 2014) use hybrid actuation for the shoulder and
The exoskeletons which use FES only for grasping and an elbow, and the elbow respectively. Both are exoskeletons
actuator only for shoulder/elbow movements vary in their which have been developed very recently and as such only
treatment of the wrist joint. The wrist joint may be fixed, limited testing has been performed so far.
actuated, or neither. Despite the differences in wrist support
all four exoskeletons which report results on reach and grasp In the less distal joints fatigue from FES is a much bigger
performance using this combination, enabled the users to issue, as is the case with actuator weight on more distal
achieve successful reach and hand-opening movements or joints. MUNDUS (Pedrocchi et al., 2013, Ambrosini et al.,
reaching and grasping of objects such as a cup or pretzel 2009, CROSTA, 2012) is unique in its use of FES only for
(Varoto et al., 2008, Elnady et al., 2015, Looned et al., 2014, actuation of the shoulder and elbow joints in combination
Rohm et al., 2012, Rohm et al., 2013, Rupp et al., 2013). with use of an actuator only for finger movements. It is
important to note however that FES for the shoulder and
The OrthoJacket (Wiegand et al., 2011, Schill et al., 2011, elbow is not used entirely on its own but is done in
Schill et al., 2009, Schulz et al., 2011, Schulz et al., 2009, combination with passive elements (springs and elastic wires)
Pylatiuk et al., 2009a), which uses FES only for the fingers or electric lockable/unlockable breaks which provide support
and an actuator only for the shoulder and elbow, did not against gravity. Passive hybrid exoskeletons are an
undergo testing of the grasping part of the exoskeleton, alternative method to full hybrid exoskeletons and have been
however the authors did propose and perform some reviewed elsewhere (Resquín et al., 2016). The use of FES in
investigation into the use of electrode arrays in an attempt to combination with passive components for the elbow and
achieve more reliable grasping patterns (Pylatiuk et al., shoulder did result in a reduction in the energy required from
2009b). ALEx (Crema et al., 2015), is also designed to use the patient to perform tasks. However, the testing of the
electrode arrays for application of the FES however, as for grasping component of this exoskeleton was not successful
the OrthoJacket, the device appears to have only undergone due to the weight of the actuators.
testing of its elbow and shoulder components without the
addition of the grasping component. It is possible that 3.2 Portability
electrode arrays may provide a better alternative to
electrically controlled actuators with regards to improving Half of the reviewed exoskeletons are aimed at stroke
control of FES for finger movements but more investigation patients. The rest are aimed primarily at spinal cord injury
is required. (SCI) patients with some accounting for other neuromuscular
and neurodegenerative dieses as well. Portability for SCI
Three exoskeletons use hybrid actuation for the fingers. The patients beyond the attachment to the patient’s wheelchair is
Haptic Robotic Glove (Hartopanu et al., 2013, Hartopanu et not often considered a high priority as without a second
al., 2014) initially used Futaba S3306 MG Maxi servos for exoskeleton for the legs the patient is not able to locomote
the actuation of the fingers however these were found to be independently anyway. On the other hand, stroke patients,
too bulky to fit on the subject’s hand. Firgelli linear actuators who may have sufficient lower limb function but be lacking
were then used instead in conjunction with artificial tendons in upper limb function, can desire assistance with activities of

15737
Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178 15173

daily living (ADL) which demand a higher level of achieving functional movement rather than on patient
portability. This kind of portability is severely lacking with monitoring. The ability of fatigue monitoring to be performed
regards to the exoskeletons reviewed here. Only the BCI- in combination with tracking and control of patient
controlled (Brain-Computer Interface) Exoskeleton (Elnady movements offers many advantages if fed back to the
et al., 2015) and the BCI-controlled Wearable Robot offer physiotherapist and could be extremely useful when selecting
full portability (Looned et al., 2014), with the OrthoJacket training programs. In addition an exoskeleton could also be
(Wiegand et al., 2011, Schill et al., 2011, Schill et al., 2009, designed to adjust its assistance in response to a combination
Schulz et al., 2011, Schulz et al., 2009, Pylatiuk et al., 2009a) of patient performance and physiotherapist commands.
having the ability to be made fully portable for some joints.
Indeed, even portability with the use of a wheelchair is So far little focus has been placed on muscular fatigue
somewhat lacking. The entire OrthoJacket exoskeleton is monitoring let alone incorporating it into the feedback and
portable with the use of a wheelchair. MUNDUS (Pedrocchi control of the exoskeleton. Two of the exoskeletons,
et al., 2013, Ambrosini et al., 2009, CROSTA, 2012) and the MUNDUS (Pedrocchi et al., 2013, Ambrosini et al., 2009,
Early Hybrid Exoskeleton (Varoto et al., 2008) also fall into CROSTA, 2012) and THINK2GRASP (Rohm et al., 2013,
this category. Overall less than half of the reviewed Rupp et al., 2013, Rohm et al., 2012), monitor the muscular
exoskeletons offer any kind of portability in their present fatigue of the subject in order to test the performance of the
state. exoskeleton but none incorporate fatigue monitoring
feedback in the control or user interface of the exoskeleton.
Many of the exoskeletons are not portable because they are While some of the exoskeletons are modular and can be
tethered to either a power/air supply or to a computer as is adapted to the different patients none are capable of
often the case with early prototypes. Therefore it would not automatically adjusting their assistance in response to muscle
be difficult to improve the portability of these exoskeletons fatigue.
with wireless communications or a battery pack fixed on the
user/wheelchair and in many cases future developments are An alternative form of patient monitoring with the use of an
planned to do so. Another important feature with regards to exoskeleton is regular observation of the ability of a patient
portability is the ability of the patient to take the device to track or perform a defined movement either by
home. Even if the device has to be plugged into a wall or measurement of the average error or speed. In cases where
computer the ability to take it home enables the patient to the speed of the patient is monitored, such as the BCI-
perform rehabilitation at home, potentially reducing travel controlled Exoskeleton (Elnady et al., 2015) and the BCI-
costs, and increasing the frequency and repetitiveness of controlled Wearable Robot (Looned et al., 2014), the timing
training. All of these exoskeletons reviewed here, with the is not done by the exoskeleton itself but rather by the
possible exceptions of The Wearable Rehabilitation Robot observer. Often where error measurement is done for tracking
(Tu et al., 2012), and RUPERT (Tu et al., 2014) which both it is not fed back to the user but is rather used simply for
require air supplies and are bulky in size, are small enough testing of the exoskeleton. Only in two cases, The FES/Robot
that transporting them to a patient’s home would likely not be Hand (Rong et al., 2012, Rong et al., 2013), and ALEx
difficult. (Crema et al., 2015), is the tracking task feedback an inbuilt
function designed to be presented to the patient and/or
Limited information could be found on the weight of the therapist as a means of patient monitoring. The FES/Robot
exoskeletons. In some cases only information on the weight Hand also monitors muscle activation levels however it is not
of the actuators was given, rather than the weight of the entire known as to whether this is just for testing purposes of the
exoskeleton. The BCI-controlled Exoskeleton (Elnady et al., exoskeleton or if it is intended to provide patient monitoring
2015), which targeted the elbow joint and fingers, was one of in a clinical setting. In the case of ALEx, the impedance of
the few fully portable exoskeletons and had a total weight of the exoskeleton can be adjusted and monitored which when
approximately 1 kg. The EXOSLIM (Serea et al., 2013, Serea compared with the tracking error feedback can help the
et al., 2014, Serea et al., 2015), which targeted the shoulder physiotherapist to assess and adjust the difficulty of the tasks
and elbow joints, was not fully portable but this was due to thus improving their ability to track patient progress. None of
tethering and not the weight. The weight of EXOSLIM was the exoskeletons are capable of automatically adjusting their
approximately 3 kg. None of the exoskeletons which level of assistance based on the user’s ability.
primarily targeted finger actuation gave the weight of the
overall device. Instead where weight was specified most gave Although not directly related to patient monitoring it is worth
the weight of the individual actuators. For these exoskeletons briefly mentioning here other forms of feedback which are
even actuators weighing as little 56 g were found to be too provided and are useful to the user. The THINK2GRASP
heavy to successfully be placed on the hand. (Rohm et al., 2013, Rupp et al., 2013, Rohm et al., 2012)
allows the patient to see the activation level and current state
3.3 Patient Monitoring of the exoskeleton which can improve their ability to control
the exoskeleton. The Early Hybrid Exoskeleton (Varoto et al.,
The vast majority of these hybrid exoskeletons are aimed 2008) provides haptic force feedback to the user which helps
purely at ADL assistance or a combination of ADL assistance them relearn the force required to pick up objects and can
and rehabilitation rather than rehabilitation alone. Thus, for especially aid patients who lack proprioception.
these hybrid exoskeletons, more focus has been placed on

15738
Proceedings of the 20th IFAC World Congress
15174
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

3.4 Actuation of FES invoked fatigue. In some hybrid exoskeletons, like

THINK2GRASP (Rohm et al., 2012, Rohm et al., 2013,
Electric motors are the most commonly used actuator among Rupp et al., 2013), when muscle fatigue is noticed the
the reviewed hybrid exoskeletons and among upper-limb stimulation intensity is simply increased to compensate. In all
exoskeletons in general (Maciejasz et al., 2014, Lo and Xie, cases where the FES device has been specified a commercial
2012), especially for movement of wrist, elbow, and shoulder device has been used instead of one being designed
joints. Two hybrid exoskeleton, the Haptic Robotic Glove specifically for use in the hybrid exoskeleton. This makes it
(Hartopanu et al., 2013, Hartopanu et al., 2014), and easier to incorporate hybrid exoskeletons into a clinical
MUNDUS (Pedrocchi et al., 2013, Ambrosini et al., 2009, setting as most rehabilitation clinics already use FES,
CROSTA, 2012), attempted to use electric motors for finger however it limits the flexibility and controllability of the
actuation and neither was successful. Electric motors are hybrid exoskeletons possibly limiting their potential
capable of larger forces and are easy to make compact and advantages.
portable however they often suffer from a lack of back
drivability and a high weight. Linear actuators, used by two
3.5 Intention Estimation and Control
hybrid exoskeletons here, the Haptic Robotic Glove
(Hartopanu et al., 2013, Hartopanu et al., 2014) and the The simplest method used to estimate the intention of the
FES/Robot Hand (Rong et al., 2012, Rong et al., 2013), are user is the use of a key/button press or joystick. This method
slightly more suited for actuation of the fingers however even is used in 50 % of these hybrid exoskeletons either as the sole
these can be too heavy as was seen with the FES/Robot method of intention estimation or in combination with other
Hand. One solution to this may be the use of artificial methods. The downside to using a key/button press or
tendons as was the case for the Haptic Robotic Glove. This joystick is that often then the patient’s movement is limited in
allows placement of the actuator closer to the centre of the some way. If the key/button press must be done with the
user’s body where the weight of the actuator is easier to other hand then the patient is unable to take part in bilateral
support. training or activities which require the use of both arms
without the presence of another person. Other methods for
Pneumatics, used by two hybrid exoskeletons in this review, obtaining user intention include the monitoring of speech or
the Wearable Rehabilitation Robot (Tu et al., 2012), and eye movement. One exoskeleton, the Early Hybrid
RUPERT (Tu et al., 2014), can be light weight and are very Exoskeleton (Varoto et al., 2008), used voice recognition as
compliant. However, they are difficult to control due to a its method of intention estimation. This method was correct
non-linear performance, can be slow to operate, and are only in recognizing a command 92% of the time. The exoskeleton,
capable of forces in one direction so must work in pairs. They MUNDUS (Pedrocchi et al., 2013, Ambrosini et al., 2009,
also require a nearby air supply which is difficult to make CROSTA, 2012), which monitors eye movement was
portable without adding significant weight. However, the successful in obtaining correct user intention. However, the
combination has been shown to be capable of reducing the eye-tracking unit used is mounted to a table and this method
overall weight of the exoskeleton (Tu et al., 2012). The is thus limited in its portability.
OrthoJacket is unique in its use of flexible fluidic actuators
which may be actuated either by air or by liquid (Pylatiuk et The use of EMG or electroencephalography (EEG), as used
al., 2009a). These are lightweight and compliant and in this by several of these hybrid exoskeletons including RUPERT,
case actuated by an air supply stored below the wheelchair. are by far the best but also the most complex methods used to
They are more compact than the pneumatics used for the obtain user intention. The reason for this is that actuation is
Wearable Rehabilitation Robot, and RUPERT exoskeletons. achieved in a very similar way to how a patient would
normally actuate their arm. When a mental attempt by a
The parameters used for FES vary depending on the specific patient to move their arm results in correct actuation this can
muscle and patient and have been discussed elsewhere result in the formation or reestablishment of neural pathways
(Doucet et al., 2012). Little feedback control of FES speeding up the recovery of the patient and helping them to
parameters has been performed in these hybrid exoskeletons. relearn self-movement (Maffiuletti et al., 2011). The use of
Parameters are often adjusted until desired movement is EMG is particularly useful for fatigue monitoring and does
achieved and then fixed at a set level. The OrthoJacket (Schill not require the training that EEG does. EMG signals are more
et al., 2011), ALEx (Crema et al., 2015), and MUNDUS closely related to the intended movement and are easier to
(Pedrocchi et al., 2013, Ambrosini et al., 2009, CROSTA, read, however EMG signals are not present in all individuals
2012) do make some attempt to optimize the electrode or can be very weak immediately after a stroke (Kiguchi and
placement but beyond this, closed-loop control with regards Hayashi, 2013). Thus, sometimes the use of EEG is a better
to fatigue management of FES pulses is severely lacking for option as it is applicable to a larger variety of users.
all hybrid exoskeletons.
Proportional or proportional-integral-derivative (PID) control
Cycling pulses on and off has been shown to improve muscle
(Fig. 1) is used for most of these hybrid exoskeletons with the
recovery, reduce fatigue, and increase patient comfort
input being either an EMG signal, a button count, or the
(Doucet et al., 2012). The FES/Robot Hand (Rong et al.,
degree of joystick movement. The two exceptions are The
2012, Rong et al., 2013) does compare different combinations
Early Hybrid Exoskeleton (Varoto et al., 2008) which uses a
of FES and actuator support however this is with the aim of
neural network to recognize and respond to predefined voice
reducing tracking error and not with regards to management

15739
Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178 15175

commands, and ALEx (Crema et al., 2015) which uses positive note some exoskeletons proved useful even for
impedance control in combination with a predefined patients with proprioception problems (Hartopanu et al.,
trajectory. THINK2GRASP (Rohm et al., 2012, Rohm et al., 2014), and others, such as the FES/Robot Hand (Rong et al.,
2013, Rupp et al., 2013), uses a state machine to select the 2012, Rong et al., 2013), reduced the energy required of the
type of movement and proportional control for the degree of user for a particular task. Lack of back drivability was an
movement. Other exoskeletons (Tu et al., 2012, Hartopanu et issue for some exoskeletons like the BCI-controlled Wearable
al., 2014, Looned et al., 2014) use pre-defined trajectories Robot (Looned et al., 2014) whereas others, such as the
where either EEG or a key/button press are used to initiate OrthoJacket (Wiegand et al., 2011, Schill et al., 2011, Schill
the movement. Of those using pre-defined trajectories two et al., 2009, Schulz et al., 2011, Schulz et al., 2009, Pylatiuk
(Tu et al., 2012, Hartopanu et al., 2014) are capable of et al., 2009a) and ALEx, were more compliant.
multiple different movements which are selected by the use
of a state machine. Proportional, PID, impedance, and Hybrid exoskeletons are a very recent development and thus
admittance control are simple to implement but more robust only limited testing has been done to date. In some cases the
control schemes, such as artificial neural networks (ANNs) or entire exoskeleton has not even been tested as a whole, such
other types, may be better suited to take full advantage of the as can be seen with the OrthoJacket (Wiegand et al., 2011,
hybrid combination and to perform fatigue management, Schill et al., 2011, Schill et al., 2009, Schulz et al., 2011,
patient improvement monitoring, or assist-as-need. Schulz et al., 2009, Pylatiuk et al., 2009a), the Haptic Robotic
Glove (Hartopanu et al., 2013, Hartopanu et al., 2014), and
ALEx (Crema et al., 2015). In many cases the exoskeleton
has only been tested on the designer themselves or a couple
of healthy subjects. None of these exoskeletons have been
tested on more than 10 subjects or over more than 20 sessions
and only one hybrid exoskeleton, the FES/Robot Hand (Rong
et al., 2012, Rong et al., 2013), has been tested where
established clinical measures, such as the Fugl-Meyer scale
(Gladstone et al., 2002), have been performed on patients as
an evaluative measure of the exoskeleton’s ability to improve
rehabilitation. In most cases only functional testing has been
performed.
There are currently no commercially available upper-limb
hybrid exoskeletons available and predicted prices for
exoskeletons which do have a long term view of
commercialization are not expected to be cheap (Pedrocchi et
al., 2013). Due to cost of hybrid exoskeletons and indeed
exoskeletons in general wide scale evaluation is difficult
(Loureiro et al., 2011). Devices which can prove beneficial to
a wider range of subjects will likely be easier to
Fig. 1. An example of closed loop PID exoskeleton control. commercialize. Modularity, which has been attempted in
some of these exoskeletons, may help to lessen the barriers to
3.6 Usability and Trials commercialization.

The preparation time of the hybrid exoskeletons ranges from 4. DISCUSSION

the very quick (<30 s with help) as seen with the BCI-
controlled Exoskeleton (Elnady et al., 2015) and BCI- Many of the reviewed hybrid exoskeletons are only in the
controlled Wearable Robot (Looned et al., 2014), to the initial stages of development with limited trials conducted
excessive (up to 1 hour) for the THINK2GRASP exoskeleton thus far. More success has been had with hybrid exoskeletons
(Rohm et al., 2012, Rohm et al., 2013, Rupp et al., 2013). focused on the less distal joints. Indeed, hybrid exoskeletons
However many of these exoskeletons do not detail the may be more useful for the shoulder and elbow joints while
preparation time so it is difficult to fully evaluate the current FES alone may be successfully used for grasping. However
status of this area. Some exoskeletons were difficult to use, the potential benefits of hybrid exoskeletons have yet to be
cause mental fatigue, and/or required several training fully taken advantage of. Many of the control schemes used
sessions prior to their use, as was the case with are simple in nature and none perform involved fatigue
THINK2GRASP. In certain cases, like the Early Hybrid management. The assist-as-need paradigm has only been
Exoskeleton (Varoto et al., 2008), patients did not have implemented in one hybrid exoskeleton. The usability of
sufficient limb movement to use the device, or in the case of current upper-limb hybrid exoskeletons is mixed with many
THINK2GRASP, were not sufficiently able to use the control exoskeletons performing quite well in some areas but lacking
interface. ALEx (Crema et al., 2015) was only capable of in others. Modularity can be a trade off with complexity and
fitting patients of a smaller and thinner stature and suffered predicted commercial prices are not expected to be cheap.
from possible joint misalignment at the shoulder. On a more However preliminary results are promising and the ability of

15740
Proceedings of the 20th IFAC World Congress
15176
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

hybrid exoskeletons to improve the energy efficiency of 2013 17th International Conference, 2013. IEEE,
exoskeletons has been successfully demonstrated in 687-692.
prototypes. It is proposed that a novel hybrid exoskeleton for HARTOPANU, S., POBORONIUC, M., SEREA, F.,
rehabilitation be designed which is capable of performing IRIMIA, D. & LIVINT, G. New Issues on FES and
patient fatigue monitoring and adjusting its assistance as Robotic Glove Device to Improve the Hand
needed. Rehabilitation in Stroke Patients. 6th International
Conference on Modern Power System, 2015.
REFERENCES HARTOPANU, S., POBORONIUC, M., SEREA, F. &
LIVINT, G. Towards Human Arm Rehabilitation in
AMBROSINI, E., FERRANTE, S., GFOELER, M.,
Stroke Patients by Means of a Hybrid FES &
REICHENFELSER, W., KARNER, J., SCHAUER,
Robotic Glove. 2014 International Conference and
T., KLAUER, C., FERRIGNO, G. & PEDROCCHI,
Exposition on Electrical and Power Engineering
A. 2009. A hybrid AssistiveSystem to Support Daily
(EPE) 2014. IEEE, 148-152.
Upper Limb Activities. Proceedings of the 17th
JARRASSÉ, N., PROIETTI, T., CROCHER, V.,
Annual International FES Society Conference (pp.
ROBERTSON, J., SAHBANI, A., MOREL, G. &
9-12).
ROBY-BRAMI, A. 2014. Robotic Exoskeletons: A
BIDDISS, E. & CHAU, T. 2008. Dielectric Elastomers as
Perspective for the Rehabilitation of Arm
Actuators for Upper Limb Prosthetics: Challenges
Coordination in Stroke Patients. Frontiers in human
and Opportunities. Medical Engineering & Physics,
neuroscience, 8, p.947, 8.
30, 403-418.
KIGUCHI, K. & HAYASHI, Y. Motion Estimation Based on
CIFREK, M., MEDVED, V., TONKOVIĆ, S. & OSTOJIĆ,
EMG and EEG Signals to Control Wearable Robots.
S. 2009. Surface EMG Based Muscle Fatigue
Systems, Man, and Cybernetics (SMC), 2013 IEEE
Evaluation in Biomechanics. Clinical Biomechanics,
International Conference on, 2013. IEEE, 4213-
24, 327-340.
4218.
CREMA, A., MANCUSO, M., FRISOLI, A., SALSEDO, F.,
LO, H. S. & XIE, S. Q. 2012. Exoskeleton Robots for Upper-
RASCHELLA, F. & MICERA, S. A hybrid NMES-
limb Rehabilitation: State of the Art and Future
exoskeleton for Real Objects Interaction. Neural
Prospects. Medical Engineering & Physics, 34, 261-
Engineering (NER), 2015 7th International
268.
IEEE/EMBS Conference on, 2015. IEEE, 663-666.
LOONED, R., WEBB, J., XIAO, Z. G. & MENON, C. 2014.
CROSTA, E. 2012. An Assistive Device Based on the
Assisting Drinking with an Affordable BCI-
Detection of the User's Intention From Residual
controlled Wearable Robot and Electrical
EMG to Drive an Upper Limb Neuroprosthesis.
Stimulation: A Preliminary Investigation. Journal of
DEL-AMA, A. J., KOUTSOU, A. D., MORENO, J. C., DE-
Neuroengineering and Rehabilitation, 11, 1-13.
LOS-REYES, A., GIL-AGUDO, Á. & PONS, J. L.
LOUREIRO, R. C., HARWIN, W. S., NAGAI, K. &
2012. Review of Hybrid Exoskeletons to Restore
JOHNSON, M. 2011. Advances in Upper Limb
Gait Following Spinal Cord Injury. Journal of
Stroke Rehabilitation: A Technology Push. Medical
Rehabilitation Research Department of Veterans
& Biological Engineering & Computing, 49, 1103-
Affairs, 49, 497-514.
1118.
DOUCET, B. M., LAM, A. & GRIFFIN, L. 2012.
LU, E. C., WANG, R. H., HEBERT, D., BOGER, J.,
Neuromuscular Electrical Stimulation for Skeletal
GALEA, M. P. & MIHAILIDIS, A. 2011. The
Muscle Function. The Yale Journal of Biology and
Development of an Upper Limb Stroke
Medicine, 85, 201.
Rehabilitation Robot: Identification of Clinical
ELNADY, A. M., ZHANG, X., XIAO, Z. G., YONG, X.,
Practices and Design Requirements Through a
RANDHAWA, B. K., BOYD, L. & MENON, C.
Survey of Therapists. Disability and Rehabilitation:
2015. A single-session Preliminary Evaluation of an
Assistive Technology, 6, 420-431.
Affordable BCI-Controlled Arm Exoskeleton and
MACIEJASZ, P., ESCHWEILER, J., GERLACH-HAHN,
Motor-proprioception Platform. Frontiers in Human
K., JANSEN-TROY, A. & LEONHARDT, S. 2014.
Neuroscience, 9.
A Survey on Robotic Devices for Upper Limb
GLADSTONE, D. J., DANELLS, C. J. & BLACK, S. E.
Rehabilitation. Journal of Neuroengineering
2002. The Fugl-Meyer Assessment of Motor
Rehabilitation, 11, 10.1186.
Recovery After Stroke: A Critical Review of its
MAFFIULETTI, N. A., MINETTO, M. A., FARINA, D. &
Measurement Properties. Neurorehabilitation and
BOTTINELLI, R. 2011. Electrical Stimulation for
Neural Repair, 16, 232-240.
Neuromuscular Testing and Training: State-of-the
HARA, Y. 2013. Rehabilitation with Functional Electrical
art and Unresolved Issues. European Journal of
Stimulation in Stroke Patients. International Journal
Applied Physiology, 111, 2391-2397.
of Physical Medicine and Rehabilitation, 1, 2.
PATTON, J., SMALL, S. L. & ZEV RYMER, W. 2008.
HARTOPANU, S., POBORONIUC, M., SEREA, F.,
Functional Restoration for the Stroke Survivor:
IRIMIA, D. & LIVINT, G. Design of a Hybrid FES-
Informing the Efforts of Engineers. Topics in Stroke
mechanical Intelligent Haptic Robotic Glove.
Rehabilitation, 15, 521-541.
System Theory, Control and Computing (ICSTCC),

15741
Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178 15177

PEDROCCHI, A., FERRANTE, S., AMBROSINI, E., Medical and Biological Engineering, 2009.
GANDOLLA, M., CASELLATO, C., SCHAUER, Springer, 1816-1819.
T., KLAUER, C., PASCUAL, J., VIDAURRE, C. & SCHILL, O., WIEGAND, R., SCHMITZ, B., MATTHIES,
GFÖHLER, M. 2013. MUNDUS Project: R., ECK, U., PYLATIUK, C., REISCHL, M.,
MUltimodal Neuroprosthesis for Daily Upper Limb SCHULZ, S. & RUPP, R. 2011. OrthoJacket: An
Support. Journal of Neuroengineering Active FES-hybrid Orthosis for the Paralysed Upper
Rehabilitation, 10, 66. Extremity. Biomedizinische Technik/Biomedical
PYLATIUK, C., KARGOV, A., GAISER, I., WERNER, T., Engineering, 56, 35-44.
SCHULZ, S. & BRETTHAUER, G. Design of a SCHULZ, S., PYLATIUK, C., KARGOV, A., GAISER, I.,
Flexible Fluidic Actuation System for a Hybrid SCHILL, O., REISCHL, M., ECK, U. & RUPP, R.
Elbow Orthosis. 2009. ICORR 2009. IEEE Design of a Hybrid Powered Upper Limb Orthosis.
International Conference on Rehabilitation Robotics, World Congress on Medical Physics and Biomedical
2009a. IEEE, 167-171. Engineering, September 7-12, 2009, Munich,
PYLATIUK, C., MÜLLER-RIEDERER, M., KARGOV, A., Germany, 2009. Springer, 468-471.
SCHULZ, S., SCHILL, O., REISCHL, M. & SCHULZ, S., SCHMITZ, B., WIEGAND, R., PYLATIUK,
BRETTHAUER, G. Comparison of Surface EMG C. & REISCHL, M. The Hybrid Fluidic Driven
Monitoring Electrodes for Long-term Use in Upper Limb Orthosis-Orthojacket. 2011.
Rehabilitation Device Control. 2009. ICORR 2009. Myoelectric Symposium.
IEEE International Conference on Rehabilitation SENELICK, R. C. 2010. Technological Advances in Stroke
Robotics, 2009b. IEEE, 300-304. Rehabilitation—High Tech Marries High Touch. US
RESQUÍN, F., GÓMEZ, A. C., GONZALEZ-VARGAS, J., Neurology, 6, 102-4.
BRUNETTI, F., TORRICELLI, D., RUEDA, F. M., SEREA, F., POBORONIUC, M., HARTOPANU, S. &
DE LA CUERDA, R. C., MIANGOLARRA, J. C. & IRIMIA, D. Towards Clinical Implementation of an
PONS, J. L. 2016. Hybrid Robotic Systems for FES & Exoskeleton to Rehabilitate the Upper Limb
Upper Limb Rehabilitation After Stroke: A Review. in Disabled Patients. 2015 20th International
Medical Engineering & Physics, 38, 1279-1288. Conference on Control Systems and Computer
ROHM, M., SCHNEIDERS, M., KREILINGER, A., Science (CSCS), 2015. IEEE, 827-832.
MÜLLER-PUTZ, G. & RUPP, R. First Evaluation SEREA, F., POBORONIUC, M., HARTOPANU, S. &
Results of a BCI-controlled Hybrid Neuroprosthesis OLARU, R. Preliminary Tests on a Hybrid Upper
for Restoration of Grasping in a High Spinal Cord Arm Exoskeleton for Upper Arm Rehabilitation for
Injured Individual. Proceedings of TOBI Workshop Disabled Patients. 2014 International Conference
lII, 2012. 8-9. and Exposition on Electrical and Power Engineering
ROHM, M., SCHNEIDERS, M., MÜLLER, C., (EPE), 2014. IEEE, 153-157.
KREILINGER, A., KAISER, V., MÜLLER-PUTZ, SEREA, F., POBORONIUC, M., IRIMIA, D.,
G. & RUPP, R. 2013. Hybrid Brain–computer HARTOPANU, S. & OLARU, R. Preliminary
Interfaces and Hybrid Neuroprostheses for Results on a Hybrid FES-exoskeleton System
Restoration of Upper Limb Functions in Individuals Aiming to Rehabilitate Upper Limb in Disabled
with High-level Spinal Cord Injury. Artificial People. System Theory, Control and Computing
Intelligence in Medicine, 59, 133-142. (ICSTCC), 2013 17th International Conference,
RONG, W., TONG, K., HU, X. & HO, N. Combined 2013. IEEE, 722-727.
Electromyography (EMG)-driven Robotic System SIVAN, M., O'CONNOR, R. J., MAKOWER, S.,
with Functional Electrical Stimulation (FES) for LEVESLEY, M. & BHAKTA, B. 2011. Systematic
Rehabilitation. 2012 38th Annual Northeast Review of Outcome Measures Used in the
Bioengineering Conference (NEBEC), 2012. IEEE, Evaluation of Robot-assisted Upper Limb Exercise
313-314. in Stroke. Journal of Rehabilitation Medicine, 43,
RONG, W., TONG, K. Y., HU, X. L. & HO, S. K. 2013. 181-189.
Effects of Electromyography-driven Robot-aided TU, X., HE, J., WEN, Y., HUANG, J., HUANG, X.,
Hand Training with Neuromuscular Electrical HUANG, H., GUO, M. & YUAN, Y. Cooperation
Stimulation on Hand Control Performance After of Electrically Stimulated Muscle and Pneumatic
Chronic Stroke. Disability and Rehabilitation: Muscle to Realize RUPERT Bi-directional Motion
Assistive Technology, 10, 149-159. for Grasping. 2014 36th Annual International
RUPP, R., ROHM, M., SCHNEIDERS, M., WEIDNER, N., Conference of the IEEE Engineering in Medicine
KAISER, V., KREILINGER, A. & MÜLLER- and Biology Society (EMBC), 2014. IEEE, 4103-
PUTZ, G. 2013. Think2grasp-BCI-controlled 4106.
Neuroprosthesis for the Upper Extremity. TU, X., HUANG, J., YU, L., XU, Q. & HE, J. Design of a
Biomedical Engineering/Biomedizinische Technik. Wearable Rehabilitation Robot Integrated with
SCHILL, O., RUPP, R. & REISCHL, M. Signal Processing Functional Electrical Stimulation. 2012 4th IEEE
Concepts for Optimal Myoelectric Sensor Placement RAS & EMBS International Conference on
in a Modular Hybrid FES Orthosis. 4th European Biomedical Robotics and Biomechatronics
Conference of the International Federation for (BioRob), 2012. IEEE, 1555-1560.

15742
Proceedings of the 20th IFAC World Congress
15178
Toulouse, France, July 9-14, 2017 Ashley M. Stewart et al. / IFAC PapersOnLine 50-1 (2017) 15169–15178

VACA BENITEZ, L. M., TABIE, M., WILL, N.,

SCHMIDT, S., JORDAN, M. & KIRCHNER, E. A.
2013. Exoskeleton Technology in Rehabilitation:
Towards an EMG-based Orthosis System for Upper
Limb Neuromotor Rehabilitation. Journal of
Robotics, 2013.
VAROTO, R., BARBARINI, E. S. & CLIQUET JR, A.
2008. A Hybrid System for Upper Limb Movement
Restoration in Quadriplegics. Artificial Organs, 32,
725-729.
WIEGAND, R., SCHMITZ, B., PYLATIUK, C. &
SCHULZ, S. 2011. Mechanical Performance of
Actuators in an Active Orthosis for the Upper
Extremities. Journal of Robotics, 2011.
WORLD HEART FEDERATION. 2015. Stroke [Online].
World Heart Federation. Available:
http://www.world-heart-
federation.org/cardiovascular-health/stroke/
[Accessed July 2016].

15743