275

The design of a Ž ve-degree-of-freedom powered

orthosis for the upper limb

G R Johnson1*, D A Carus2, G Parrini3, S Scattareggi a Marchese4 and R Valeggi3

1Centre for Rehabilitation and Engineering Studies, University of Newcastle, Newcastle upon Tyne, UK
2Department of Applied Physics, Mechanical and Electrical Engineering, University of Dundee, Scotland, UK
3Scienzia Machinale srl, Pisa, Italy
4Signo Motus srl, Messina, Italy

Abstract: In response to the need for a sophisticated powered upper-limb orthosis for use by people
with disabilities and/or limb weakness or injury, the MULOS (motorized upper-limb orthotic system)
has been developed. This is a Ž ve-degree-of-freedo m electrically powered device having three degrees
of freedom at the shoulder, one at the elbow and one to provide pronation/supination. The shoulder
mechanism consists of a serial linkage having an equivalent centre of rotation close to that of the
anatomical shoulder; this is a self-contained module in which power transmission is provided by
tensioned cables. The elbow and pronation/supination modules are also self-contained.
The system has been designed to operate under three modes of control:
1. As an assistive robot attached directly to the arm to provide controlled movements for people
with severe disability. In this case, it can be operated by a variety of control interfaces, including
a specially designed Ž ve-degree-of-freedo m joystick.
2. Continuous passive motion for the therapy of joints after injury. The trajectory of the joints is
selected by ‘walk-through’ programming and can be replayed for a given number of cycles at a
chosen speed.
3. As an exercise device to provide strengthening exercises for elderly people or those recovering
from injury or surgery. This mode has not been fully implemented at this stage.
In assistive mode, prototype testing has demonstrated that the system can provide the movements
required for a range of simple tasks and, in continuous passive motion (CPM ) mode, the programming
system has been successfully implemented. Great attention has been paid to all aspects of safety.
Future work is required to identify problems of operation, and to develop new control interfaces.

Keywords: powereed orthosis, upper limb, MULOS, modes of control

1 INTRODUCTION paralysis caused by high-level spinal cord injury or mus-

Despite the attractions of providing powered movement
at the upper limb for people, for instance, with high-
level spinal cord injury, there have been relatively few
attempts to produce suitable devices. This probably
results from the inevitable complexities of producing
multiple-degree-of-freedo m systems and the associated
diYculties of control. A number of diVerent applications
can be envisaged for such a system. The possibility of
providing assisted limb motion for people with total
The MS was received on 22 December 1999 and was accepted after
revision for publication on 28 July 2000.
* Correspondin g author: Centre for Rehabilitation and Engineering
Studies (CREST), University of Newcastle, Stephenson Building,
Newcastle upon Tyne NE1 7RU, UK
cular dystrophy has been proposed by Snelson et al. [1].
However, while there are many users of powered wheel-
chairs who can use the upper limb only with the aid of
a powered device, few systems are currently available.
For instance, powered orthoses have been developed by
Engen’s group in Texas to provide grip function for
patients with spinal injury [2]. At the other end of the
technological scale, robotic devices such as MANUS*
can substitute totally for use of the upper limb [3].
The provision of continuous passive motion (CPM )
at one or more joints has been strongly advocated by
Salter in 1989 [4], but has been conŽ ned to the joints of
the lower limb and usually to one joint at a time.
* Exact Dynamics, The Netherlands.

H06899 © IMechE 2001 Proc Instn Mech Engrs Vol 215 Part H
276 G R JOHNSON, D A CARUS, G PARRINI, S SCATTAREGGIA MARCHESE AND R VALEGGI
Recently, Carus et al. [5] have demonstrated the possibil-
ity of using these methods in the rehabilitation of the
hand. Cogemo* have also produced systems to provide
single-axis motion at the shoulder or elbow. However,
there has been virtually no work on the extension of this
technology to provide multiple-degree-of-freedo m move-
ments at the wrist, elbow or shoulder despite the need
to rehabilitate the upper limb as a whole.
Another potential application of a powered exo-
skeleton is for the maintenance of upper-limb function
in the elderly which is dependent upon coordination,
range of motion and muscle strength. While the use of
CPM can enhance range of motion, there is still the need
to improve or maintain strength. At present, while Ž tness
equipment is readily available for the young and Ž t it is
available only in the gymnasium and is not appropriate
to the needs of the elderly.
Finally, perhaps the most challenging application of
such a system would be its use as a therapy device for
patients after stroke. This idea has been explored by a
number of researchers. At the level of a single joint,
Cozens [6 ] has developed a controlled robotic arm which
assists the patient to move the joint to a sequence of
target positions deŽ ned by illuminated lights. Krebs et al.
[7] have developed the MIT-MANUS system, having
three degrees of freedom and using impedance control
to provide therapy to the upper limb. Similar approaches
have been investigated by Riekensmeyer [8] and Lum
et al. [9]. However, the lack of powered devices with
more than three degrees of freedom of movement to the
joints of the arm leads to a perceived need for equipment
providing a powered orthosis for the upper limb to pro-
vide movements of up to six degrees of freedom using
control modalities to satisfy all of the requirements
identiŽ ed above.
There have been a number of attempts to design such
equipment, although it appears that much more work
has been done on powered upper-limb prostheses (e.g.
the Utah arm described by Jacobsen et al. [10] in 1982 )
than on orthoses or assistive devices. Early attempts at
designing orthotic systems (e.g. Rancho Los Amigos
seven-degree-of-freedo m manipulator in reference [11])
were generally unsuccessful due to their high complexity,
high cost and diYculty of control. There were also severe
limitations in the interface control systems; for instance,
on the Rancho Los Amigos device, each joint was con-
trolled directly by a tongue switch. More recently the
Toronto orthosis was developed by Salter [4] and then
improved by Romilly et al. [12] to become the UBC
orthosis. This was a Ž ve-degree-of-freedo m device
intended for users with conditions such as muscular
dystrophy, poliomyelitis or amyotropic lateral sclerosis.
However, it appears not to have been developed to com-
mercial production.
Probably because of the limitations of these devices,
* Cogemo SA, Tournes, France.
Proc Instn Mech Engrs Vol 215 Part H
work has subsequently concentrated on the design of
manipulators that attempt to replace the function of the
arm rather than provide direct assistance. A notable
recent example of such a system is the MANUS wheel-
chair-mounte d robot [3]. However, this approach
ignores those users who do retain some upper limb
function, such as people with muscular dystrophy, and
cannot, of course, provide the therapeutic or exercise
applications discussed above.
It was the above situation that led a European
Consortium to propose an entirely new design of a multi-
degree-of-freedo m powered orthosis for the upper limb.
This paper describes the mechanical design and develop-
ment of a Ž ve-degree-of-freedo m motorized upper limb
orthotic system (MULOS ). The targeted primary users
of this system are elderly and disabled people and it has
been designed primarily for use as an assistive orthosis
or as a CPM and exercise device. While it has not been
designed primarily for providing neurorehabilitation ,
these applications may well be possible in the future.
2 APPROACH
The overall concept of the MULOS system is a Ž ve-
degree-of-freedo m modular powered articulated struc-
ture which can be attached to the user’s arm in order to
provide powered movement. At the design planning
stage it was necessary to consider the potential uses and
the modes of control which might be required. After
discussions of feasibility within constraints of time and
Ž nance, it was decided to concentrate on the develop-
ment of three control modes as follows:
1. Assistive mode. To provide powered movement under
control from a suitable interface device.
2. Continuous passive motion. To provide cyclical
motion of selected joints during rehabilitation after
joint injury or total joint replacement.
3. Exercise mode. To provide a preselected resistance at
a speciŽ ed joint.
The technical challenges were, therefore, to design a
powered mechanism which would Ž t to and provide
motion of the complete upper limb including the
shoulder complex but excluding the hand. The major
task has been to design a mechanical power transmission
system to allow powered movement while maintaining
minimum bulk and mass together with a control system
to permit use under the chosen control modes.
3 MECHANICAL DESIGN
3.1 Overall design
It was decided to design the system in such a way that
individual modules corresponding to particular joints
H06899 © IMechE 2001
 THE DESIGN OF A FIVE-DEGREE-OF-FREEDOM POWERED ORTHOSIS FOR THE UPPER LIMB
could be used in isolation. This called for the design of
self-contained assemblies to power the shoulder, the
elbow and pronation/supination of the forearm. The
associated control system should also be modular and
the whole assembly should be constructed in such a way
as to permit mounting either on a powered wheelchair
or on any other appropriate Ž xture. This paper is primar-
ily concerned with the design of the mechanical compo-
nents and will concentrate particularly on the shoulder
mechanism which presented a major challenge.
3.2 Shoulder
The diYculty of designing a mechanism to move the
shoulder, relates to the need to allow a large range of
motion and the necessary moments. An initial simpliŽ ed
modelling study showed that a relatively high torque
(around 20 N m) would be required to elevate the arm
in  exion or abduction. This led to a need for a low-
mass powered mechanism which would permit the
shoulder to move through the required workspace. Due
to the range of possible applications it was felt necessary
to maintain this space as large as was practically poss-
ible. For instance, while the range of shoulder elevation
required for everyday tasks is usually less than 90ß [13],
greater ranges may be required for CPM applications.
Therefore, the problem was to design a kinematic chain
intimately connected to the humerus, which could move
with the shoulder through a large range of motion with-
out causing injury or dislocation of the relatively delicate
natural shoulder joint. This led to the investigation of
possible designs in which it was assumed that, initially,
the shoulder joint could be considered as a spherical
joint. A conceptual design was considered in which a
second spherical joint, having its centre of rotation
coincident with that of the shoulder, resulted in two
linked kinematic chains (the manipulator and the
shoulder) which were equivalent and possessed the same
workspace. This was realized by a kinematically equival-
ent mechanism having a sequence of three revolute joints
with intersecting axes. Furthermore, since the shoulder
complex does not move as a pure spherical joint, and,
in addition, the user’s position is not Ž xed, the rotation
centre of the manipulator must always be, to some
extent, diVerent from that of the shoulder. However,
provided that this misalignment is small, the two kinem-
atic chains are nearly coincident and, assuming that they
are not rigidly connected (because of the presence of soft
tissues), they can move together as a mechanism. It is
important to point out here that the translations of the
glenoid would be expected to be around 20 mm [14] and
that they would have to be accommodated even when
the arm was being moved passively [15]. Greater mis-
alignments were possible by inserting compliances or
passive degrees of freedom between the two structures.
Having decided to use a mechanism of this type, it
was necessary to determine its spatial position and the
H06899 © IMechE 2001
277
order of the joints of the structure. This problem is com-
plex because it is necessary to Ž nd the optimum relation-
ship between the key variables (e.g. orientation of axes
with respect to a Ž xed frame, link lengths and geometry)
leading to the maximum workspace of the two linked
kinematic chains. Due to this complexity, an experimen-
tal approach was chosen. The problem was addressed
using a passive three-degree-of-freedo m linkage having
ten available adjustments of link lengths and orientation
of all axes of rotation. Initially this was assembled in
such a way as to avoid collision with user body segments.
For a number of conŽ gurations, the compatibility of the
workspace of the two kinematic chains was experimen-
tally evaluated according to the  ow diagram in Fig. 1.
This optimization was performed in two stages. Firstly,
the kinematic chain of the shoulder mechanism was
adjusted and deemed to be satisfactory when it allowed
access to the required workspace without problems of
singularities or of collision, with either a part of the
structure or with the user. When this stage of optimiz-
ation had been achieved, it was possible to proceed to
more detailed design of the powered structure. However,
the integration of the main mechanical components (e.g.
motors, power transmission and sensors) unavoidably
changed the volumes occupied by the system and pos-
itions at which collision occurred, making it necessary
to perform a further series of iterative adjustments of
the design conŽ guration. Finally, experimental studies
have shown that the user is able to access all of the
workspace required for identiŽ ed tasks. The chosen
design conŽ guration is shown in Fig. 2, and the corre-
sponding Denavit–Hartenberg parameters are shown
in Fig. 3.
In order to design the mechanical drives and to select
the appropriate electric motors, it was necessary to
develop a computer model of the combined arm and
structure. This model, described in an earlier paper by
Buckley and Johnson [16 ], allowed the computation of
joint torques and power consumption at any chosen pos-
ition of the limb and while performing certain deŽ ned
tasks (see Table 1).
The associated drive mechanism was required to be of
minimum mass in order to minimize the required torques
and hence motor size and power consumption. It was
decided at an early stage that an electrical system (rather
than, for instance, a hydraulic system) should be used
because of the availability of compact electric motors
and the ease of control. Since a high drive torque was
required at the joint providing elevation of the arm, it
was of paramount importance not to increase this any
more than necessary by the Ž tting of electrical motors
and power transmission. This led to a requirement to
stow the electric motors as near as possible to the Ž rst
joint, and to use a lightweight power transmission. While
it has been shown in the design of the MANUS robot
that systems using long transmission shafts and gear-
boxes can be successful, this option was rejected on the
Proc Instn Mech Engrs Vol 215 Part H
278 G R JOHNSON, D A CARUS, G PARRINI, S SCATTAREGGIA MARCHESE AND R VALEGGI

Fig. 1 Flow diagram to illustrate the decision-making process in designing the shoulder mechanism

Fig. 2 Diagrammatic representation of the MULOS structure

Proc Instn Mech Engrs Vol 215 Part H H06899 © IMechE 2001
 THE DESIGN OF A FIVE-DEGREE-OF-FREEDOM POWERED ORTHOSIS FOR THE UPPER LIMB 279

arrangement can be seen in Fig. 4. Position transducers

were Ž tted at each joint to ensure direct feedback, and
overload clutches were provided at each joint for safety
purposes.
The use of further cable drives for the elbow and fore-
arm mechanisms was rejected for two reasons. Firstly,
the required torque and, therefore, the motor size was
considerably reduced at these joints. Second, the need
for modularity and for adjustment to Ž t the user would
have led to considerable design complexities. The elbow
joint mechanism was designed as a single-degree-of-free -
dom joint with a separate powered unit at the wrist to
provide pronation and supination. The drive at the
elbow joint was provided by a Minimotor motor and
gearbox assembly, connected to a brake and an encoder.
The drive assembly included a ‘double-back’ drive
arrangement to reduce its overall length because of the
reduced space available between the distal end of the
shoulder mechanism and the elbow joint. A slip clutch
was inserted between the drive chain and the elbow joint
gearbox, in order to decouple the elbow joint from the
drive motor in the event of a user spasm. The Ž nal torque
output was provided by a bevel gearbox driving an
output arm Ž xed to a thermoplastic elbow orthosis.
The pronation/supination module was operated by
moving an inner ring, Ž tted to the user, within a Ž xed
concentric outer ring attached to the MULOS structure
(see Fig. 4). The size of the inner ring was determined
by the size of the user’s wrist; initially a ring of 70 mm
Fig. 3 Kinematic conŽ guration of the MULOS system. X, Y, internal diameter was used. Movement was provided by
Z is the world reference frame and x , y , z , etc., are the axes
1 1 1 a timing belt attached to the inner ring, and moved by
corresponding to each link. It should be noted that joint 1 is
a pulley attached to the motor on the Ž xed ring. The
rotated by 35ß around the Y axis. The corresponding Denavit–
Hartenberg parameters are shown below: required maximum speed for the joint was chosen to be
60–75ß /s on the basis of simple experimental studies. The
Link a (mm)
i
á (deg)
i
d (mm)
i
õ
i available range of motion was between 75ß pronation
1 0 õ
and 140ß supination measured from the ‘handshake’
Õ 90 0
2 0 Õ 90 0 õ
1 position, with the palms of the hands facing each other.
2
3 0 0 135 õ
3

grounds of cost and complexity. It was decided, instead, 4 CONTROL

to use cable drives which have previously been described
by Bergamasco et al. [17] as a means of providing a
high-stiVness, low-mass power transmission system for
the three degrees of freedom at the shoulder. In this
arrangement, one electric motor was attached to the
Ž xed structure to provide power at the Ž rst joint, and
the other two were stowed inside the Ž rst link. This
While this paper concentrates primarily on the mechan-
ical aspects of the MULOS design it is important
to include a description of the control systems. In a
multiple-degree-of-freedo m structure of this kind, the
sensors, the control architecture and the related software
are strictly interdependent. In addition, the choice to

Table 1 Peak torque and power recorded at the shoulder mechanism joints

Peak torque (N m) Peak power ( W )

Simulation Joint 0 Joint 1 Joint 2 Elbow Joint 0 Joint 1 Joint 2 Elbow

Azimuth 0.68 12.47 14.95 — 0.38 4.3 7.44 —

Elevation 0.93 11.05 14.69 — 0.14 4.33 8.26 —
Rotation 0.45 11.68 10.10 — 0.27 3.34 3.98 —
Eating 2.25 8.14 11.20 7.20 1.21 2.25 7.21 12.23

H06899 © IMechE 2001 Proc Instn Mech Engrs Vol 215 Part H
280 G R JOHNSON, D A CARUS, G PARRINI, S SCATTAREGGIA MARCHESE AND R VALEGGI
Cable Drive
Shoulder Axis 2
Shoulder Axis 3
Elbow Drive
Fig. 4 Arrangement of electric motors and cable drives at the shoulder mechanism
implement three diVerent modalities (assistive, CPM and
exercise) with the same hardware aVects quite heavily
the control architecture and, hence, the related software.
In the chosen mechanical design, the control has had to
take into account the transformation s between the
orthosis joint space and the human arm joint space, in
order to allow a compound movement. The control of
MULOS is, depending upon the modality, a velocity
control or a force control on Ž ve degrees of freedom,
with particular care being devoted to aspects of safety.
This control is provided by a PC104 66 MHz 486 com-
puter having 8 MB RAM. This computer provides both
low-level PID control of the electric motors as well as
performing the necessary Jacobian calculations for kine-
matic control. This computer and all of the electric
motors are powered by a 24 V d.c. battery to allow com-
patibility with powered wheelchairs and to ensure electri-
cal safety of the user.
In assistive mode, the user can control the mechanism
through a four-plus-one degree-of-freedo m set of com-
mands, representing basic shoulder or elbow speeds. This
command set can be generated by various kinds of user
interface, taking into account the individual user’s dis-
abilities. Special care has been devoted to the user
interface connection, providing the system with a well-
deŽ ned protocol over a general-purpose electrical
interface. In this way, many kinds of switches, joy-
sticks (or a combination of these), and other standard
interfaces can be easily connected.
Proc Instn Mech Engrs Vol 215 Part H
Shoulder Axis 1
Pronation Supination Drive
In CPM mode, the required movement sequences can
be selected using ‘walk-through programming’ in which
the user moves the system by hand through the required
motions. The control software records these movements
and can then reproduce them for a chosen number of
cycles at a selected speed. The exercise mode software
has not been implemented at this stage of the project.
The safety of the user has been a major consideration
of the control design. Mechanisms have been provided
at mechanical, electrical and software levels in order to
avoid unexpected or dangerous behaviour of the orthosis
during use. Particular care has been taken to ensure that
it will react appropriately and safely to unexpected
user spasms.
5 CONTROL INTERFACES
Communication and control represent major challenges
in the introduction of high technology for the use of
people with disabilities. While development of such
interface controls did not fall within the remit of this
project, it was necessary to have some device to allow
prototype user testing and demonstration. It was with
this in mind that the Consortium performed a design
and evaluation of a specialist joystick for the control of
MULOS. This device combined the normally used side-
to-side movements of a joystick together with axial
movement and rotation about the long axis [18]. It was
H06899 © IMechE 2001
 THE DESIGN OF A FIVE-DEGREE-OF-FREEDOM POWERED ORTHOSIS FOR THE UPPER LIMB
demonstrated that potential users found this design
intuitive to use and were able perform a number of
tracking tasks created for assessment purposes.
6 THE FINAL DESIGN
The prototype system was mounted on a powered
wheelchair and utilized the same battery power supply.
The joystick was incorporated into the armrest of the
wheelchair on the opposite side from the orthosis. The
principal components can be seen in Fig. 5 and are
described below.
1. The shoulder mechanism. The uppermost electric motor
can clearly be seen and the other two motors are stowed
in the moving structure. The drive cables are guided by
idler pulleys to conform to the form of the links. A
diVerential arrangement has been employed so that two
of the drives can be transmitted at a common shaft
Fig. 5 Photograph of the MULOS system Ž tted to a powered wheelchair
H06899 © IMechE 2001
281
assembly. No adjustment for size is required but it is
necessary to ensure that the user’s shoulder is appropri-
ately aligned with the kinematic centre of the linkage
as discussed above. This adjustment is provided at the
point of attachment to the wheelchair.
2. The elbow and forearm assembly. The elbow mechan-
ism is attached to the last link of the shoulder mech-
anism by an adjustable link which allows Ž tting to
users of diVerent sizes. The electric motor and associ-
ated drive system are concealed within the lightweight
housing. The forearm mechanism has a small electric
motor which transmits torque to the ring surrounding
the wrist through an integral reduction gearbox and
toothed belt drive.
3. The four-degree-of-freedo m joystick. This can be
clearly seen in Fig. 5.
4. Controlling computer. This is mounted at the back of
the wheelchair.
5. Power supply. Electrical power is derived from the
wheelchair battery beneath the seat. In the event of
Proc Instn Mech Engrs Vol 215 Part H
282 G R JOHNSON, D A CARUS, G PARRINI, S SCATTAREGGIA MARCHESE AND R VALEGGI
use away from a wheelchair, an appropriate power
supply would be required.
7 PERFORMANCE EVALUATION AND
PRELIMINARY USER TESTING
A programme of testing was performed to identify the
key safety-relate d and functional features. Quantitative
measurements of performance of the system were fol-
lowed by some simple studies of the ease of use and
range of workspace carried out by an able-bodied ‘user’.
After these preliminary tests, the system was used in the
laboratory to perform a wide variety of manoeuvres
under joystick control. At Ž rst, it was used intensively
in CPM and assistive mode to identify minor faults
which were all rectiŽ ed by simple modiŽ cations. In order
to gain experience of the system in assistive mode, it was
used, by an able-bodied subject, to simulate the perform-
ance of a number of tasks as illustrated in Fig. 6. Clearly,
testing by an able-bodied subject reveals only some
aspects of system function and, in particular, does not
deal with the aspects of control interfaces. However, the
tests have successfully demonstrated the capability of the
system to access the workspace. While preliminary test-
ing has been performed by a disabled user, extensive
studies have not yet taken place.
8 DISCUSSION
The overall aim of this project was to develop a compre-
hensive, modular powered system to provide restoration
or maintenance of motor function to all the joints of the
Fig. 6 Use of the system to perform simple tasks
Proc Instn Mech Engrs Vol 215 Part H
upper limb. Due to the economic constraints on social
services and health care, it has been essential to develop
modular equipment which can be purchased and used
without specialist prescription and supply. While the
basic suitability of the system has been demonstrated,
much remains to be done. In particular, as far as further
design and development are concerned, a large number
of users of MULOS in the assistive mode will require a
powered system to provide hand prehension. While it
was originally hoped that it might be possible to incor-
porate existing devices into the system, this now looks
unlikely because of the diYculties of interfacing and
compatibility. The design team fully accept that a pro-
gramme of research and development is required to
develop an appropriate system.
There have been particular challenges in designing the
mechanism and drive system. For instance, at the
shoulder it has been demonstrated that, using cable
drives, it is possible to achieve a low-mass compact
design which is suitable for attachment to a wheelchair
or other support. Furthermore, the use of a three-degree-
of-freedom serial linkage has been shown to provide the
necessary range of movement at the shoulder without
prejudicing its integrity.
Many challenges remain before this system is ready
for supply to people with disabilities or to other users,
the greatest being the identiŽ cation of the problems of
operation and particularly of safety. This is an aspect
which is relatively little explored but requires much work
before the supply of high-technology powered equip-
ment to disabled users can become routine. In the case
of MULOS, it will require a large amount of testing,
under closely controlled conditions, by potential users
having a range of disabilities. There are many aspects of
H06899 © IMechE 2001
 THE DESIGN OF A FIVE-DEGREE-OF-FREEDOM POWERED ORTHOSIS FOR THE UPPER LIMB
safety to be examined. Firstly, there are considerations
of reliability of operation and behaviour making it neces-
sary to incorporate a variety of fail-safe mechanisms into
all powered elements. Further safety problems are
associated with the diYculties of moving the shoulder
through a large workspace. The relative ease of shoulder
dislocation is well known and is likely to be a far greater
problem for users having weak musculature. While all
the indications are that this is not a problem with this
design, much further clinically supervised testing will be
required to conŽ rm it.
A particular safety challenge has been the need to
avoid injury or pain in the presence of uncontrolled
spasms of the arm. The approach taken in this project
has been to provide both mechanical and electrical detec-
tors which leave the aVected joint(s) completely free in
the event of an overload. The system must then be reset
by an attendant before it can be reused. Clearly, there
are disadvantage s in this approach. In particular, it
leaves the user largely helpless until the system is
restored. This is a diYcult area in which there is a lack
of documented knowledge and experience and a need
for much research. A Ž rst stage would be the develop-
ment of clinical techniques to identify the activities and
limb positions which are most likely to lead to a spasm.
In severe cases of spasticity, clinical approaches involv-
ing, for instance, the use of Botulinum toxin may be
required to eliminate the activity of a particular muscle.
A vital aspect of research and development in this Ž eld
is the active involvement of users. However, so far, there
is relatively little experience in the rehabilitation com-
munity of the necessary research techniques which can
identify these problems and maximize the input of users
to the Ž nalizing of design. A few research centres are
actively developing the appropriate research method-
ology but there is little published data at this time.
However, probably the greatest challenge of designing
equipment of this kind, is that of providing the means
by which the user with severe disability can control the
technology suYciently easily for it to provide a real ben-
eŽ t. Many of the potential users of the system in assistive
mode are likely to have bilateral involvement ruling out
the use of a joystick. This general problem of user con-
trol is being addressed by a number of research groups.
For instance, the designers of the MANUS system have
demonstrated the power of speech-operated controls.
The Delaware group has also shown the potential for
the use of a sophisticated speech-based system allowing
the identiŽ cation of objects and tasks using a simple
grammar [19]. Even when the basic problems of interfac-
ing have been overcome, much work remains to be done
to increase the speed of controlled operation, so that
tasks can be performed at speeds which are available
from the technology (Schuyler et al. [20]). Clearly such
developments, which must always take into account con-
siderations of safety, will be a major emphasis for future
research.
H06899 © IMechE 2001
283
In this project, it was decided to use an approach in
which an articulated structure was attached intimately
to the user’s arm. However, in the currently avail-
able MANUS system, the user performs tasks with a
wheelchair-mounted robot. There are undoubtedly some
advantages to this approach: for instance, it becomes
possible to access a workspace outside that of the arm.
Furthermore, the safety problems are very considerably
reduced when the user is not physically connected to the
robot. This approach was rejected for the MULOS
system because of the wish to provide therapeutic as well
as assistive movement of the limb. While this has led to
a number of diYcult design problems, it is believed that
there will be a long-term advantage in terms of the
number of users who may beneŽ t from a more versa-
tile system.
Finally, it is important to repeat the need for assistive
technology to be developed and tested in close collabor-
ation with users. While the Consortium has been in close
collaboration with clinical personnel throughout the
project, it is acknowledged that much remains to be done
in this general area. In particular, there is a need to
develop test protocols which allow objective evaluation
of the beneŽ ts of the system in assistive mode. While
some work has been performed by the Delaware group
[20 ], much remains to be done to allow benchmarking
of these assistive technology devices.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the help and
support of many colleagues and potential users of the
MULOS system. Particular thanks are due to the staV
of TVI ( Europe) Limited for all their eVort in the manu-
facturing and development of the prototype described
here. Thanks are also due to Professor Georg NeV of
the Free University of Berlin for the clinical support
provided. Finally, the authors pay particular tribute to
the very great input of the late Dr Robin Platts who, in
many respects, was the inspiration behind the project.
The research has been supported by the European
Commission TIDE Programme (Project 1057).
REFERENCES
1 Snelson, R., Karchak, A. and Nickel, V. L. Application of
external power in upper extremity orthotics. Orthotics
Prosthetics Appl. J., 1961, 15, 345–348.
2 Engen, T. J. and Spencer, W. A. Development of externally
powered upper extremity orthotics. Texas Institute for
Rehabilitation and Research, Houston, Texas, 1969.
3 Rosier, J. C., van Woerden, J. A., van der Wolk, L. W.,
Driessen, B. J. F., Kwee, H. H., Duimel, J. J., Smits, J. J.,
Tuinhof de Mode, A. A., Honderd, G. and Bruyn, P. M.
Rehabilitation robotics: the MANUS concept. In
Proc Instn Mech Engrs Vol 215 Part H
284 G R JOHNSON, D A CARUS, G PARRINI, S SCATTAREGGIA MARCHESE AND R VALEGGI
Proceedings of the Fifth ICAR International Conference
on Advanced Robotics, 1991, Vol. 1, pp. 893–898.
4 Salter, R. B. The biologic concept of continuous passive
motion of synovial joints. The Ž rst 18 years of basic
research and its clinical application. Clin. Orthop., 1989,
242, 12–25.
5 Carus, D. A., Thorpe, J. R., Logan, G. M. and Doyle, M. J.
Linkage design for the ‘machine-to-patient’ interface of a
servo-controlled motorised hand therapy machine. IEE
Colloquium (Dig.), 1995, 107, 13/1.
6 Cozens, A. Robotic assistance of an active limb exercise in
neurologically impaired patients. IEEE Trans., Rehabil.
Engng, 1999, 7, 254–256.
7 Krebs H. I., Hogan, N., Aisen, M. L. and Volpe, B. T. Robot
aided neurorehabilitation. IEEE Trans., Rehabil. Engng,
1998, 6, 75–87.
8 Riekensmeyer, D. J. Guidance based quantiŽ cation of arm
impairment following brain injury. IEEE Trans., Rehabil.
Engng, 1999, 7, 1–11.
9 Lum, P. S., Burgar, C. G., Kenney, D. E. and Van der Loos,
H. F. M. QuantiŽ cation of force abnormalities during pass-
ive and active-assisted upper-limb reaching movements in
post-stroke hemiparesis. IEEE Trans., Biomed. Engng,
1999, 46, 652–662.
10 Jacobsen, S. C., Knutti, D. F, Johnson, R. T. and Sears,
H. H. Development of the Utah artiŽ cial arm. IEEE Trans.
Biomed. Engng, 1982, BME-29, 249–269.
11 Davalli, A., Sacchetti, R. and Schmidl, H. Multi-functional
prosthetic orthotic systems. When? In Proceedings of the
IEEE International Conference on Systems, Man and
Cybernetics, 1993, Vol. 3, pp. 531–533.
12 Romilly, D. P., Anglin, C., Gosine, R. G., Hershler, C. and
Raschke, S. U. A functional task analysis and motion simu-
lation for the development of a powered upper limb
orthosis. IEEE Trans., Rehabil. Engng, 1994, 2, 119–128.
Proc Instn Mech Engrs Vol 215 Part H
13 Buckley, M., Yardley, A, Johnson, G. R. and Carus, D. A.
Kinematic and kinetic parameters of the upper limb during
performance of the tasks of everyday living—a review of
the current knowledge base. Proc. Instn Mech. Engrs, Part
H, Journal of Engineering in Medicine, 1996, 210 (H4),
241–247.
14 Barnett, N. D., Duncan, R. D. D. and Johnson, G. R. The
measurement of three dimensional scapulohumeral
kinematics—a study of reliability. Clin. Biomechanics,
1999, 14, 73–76.
15 Franklin, P., Price, C. I. M., Curless, R., Rodgers, H. and
Johnson, G. R. Active and passive scalpulohumeral move-
ment in healthy persons: a comparison. Arch. Phys. Med.
Rehabil., 2000, 81, 28–31.
16 Buckley, M. A. and Johnson, G. R. Computer simulation
of the dynamics of a human arm and orthosis linkage mech-
anism. Proc. Instn Mech. Engrs, Part H, Journal of
Engineering in Medicine, 1997, 211 (H5), 349–357.
17 Bergamasco, M., Allota, B., Bosio, L., Ferretti, L.,
Parrini, G., Prosco, G. M., Salsedo, F. and Sartini, G. An
arm exoskeleton system for teleoperation and virtual
environment applications. In Proceedings of the IEEE
International Conference on Robotics and Automation,
1994, Vol. 2, pp. 1449–1454.
18 Scattareggia Marchese, S., Parrini, G., Buckley, M. A.,
NeV, G., Platts, R. and Johnson, G. R. A system for assess-
ment of user capabilities and interfaces (SAUCI ). In
Proceedings of the International Conference on
Rehabilitation Robotics, Bath, 1997, pp. 143–146.
19 Kazi Z. and Foulds R. Knowledge driven control of an
intelligent assistive robot. In Proceedings of the RESNA
Annual Conference, Pittsburgh, 1997, pp. 393–395.
20 Schuyler, J. L., Mahoney, R. M. and Hove, P. A.
Standardised assessment tests for the human robot system.
In Proceedings of the International Conference on
Rehabilitation Robotics, Bath, 1997, pp. 99–102.
H06899 © IMechE 2001