Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
December 1999
Multi-Segmented Finger Design of an Experimental Prosthetic Hand
N. Dechev1, W.L. Cleghorn1, and S. Naumann2,3
1 Department of Mechanical & Industrial Engineering, University of Toronto, Toronto, Ontario, Canada, M5S 3G8
2 Rehabilitation Engineering Department, Bloorview MacMillan Centre, Toronto, Ontario, Canada, M4G 1R8
3 Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada, M5S 1A4
email: [email protected]
Abstract:
The design of a three segment finger that can curl
during flexion is presented. It has been applied in an
experimental, multi-fingered, child sized prosthetic
hand. Conventional prosthetic hands use rigid
members for fingers, which cannot curl around
objects, and have an artificial appearance. A review
of four other multi-segment finger designs is done.
Design considerations focused on prosthesis
requirements such as small size, low weight, high
strength and good cosmetic appearance. The
kinematics of the finger were modelled with Working
Model software, and the structure of the finger
segments were modelled with IDEAS finite element
analysis software, to ensure strength. In addition, the
finger design presented could be used with two
degrees of freedom, one to control flexion and
extension, and the other to control the radius of the
finger curl. The three segments of this finger design
cannot flex independently of one another, and as a
result, each finger can only approximately conform to
the shape of an object it touches. The advantages of
this finger design, in combination with an adaptive
grasp system (allowing different fingers to flex
independently of one another), is that multiple contact
points can be created between the fingers involved in
grasping, and the object they grasp. This allows
irregularly shaped objects to be grasped more
securely, and gives the resulting grasp a very natural
appearance.
powered anthropometric hands, such as the Otto Bock
Hands [1] or the VASI Hands [2], are among the
most advanced designs. They require minimal effort
to use, as they are powered by batteries, and are
controlled by myo-electric sensors that detect muscle
contraction signals in the remnant limb of the user.
These electric prostheses have four fingers and a
thumb, and are covered by a PVC glove to give the
appearance of a natural hand. Kinematically, they are
very simple devices, with two rigid fingers acting in
opposition to a rigid thumb to produce a pinch, as
illustrated in Figure 1. The remaining fingers are
non-functional and serve only for cosmetic
appearance.
Figure 1.
Introduction:
There are a number of different conventional,
commercially available prosthetic hands such as
passive (non-functional) wooden hands, cable
powered split hooks, and electric powered
anthropometric (geometrically similar to the human
form) hands. Of these varieties, the electrically
Conventional VASI 7-11 Hands[2]
Due to their simple design, conventional
prostheses have a number of limitations. They have
limited mechanical function since they are restricted
to a single degree of freedom, open and close action.
Cylindrical objects can be grasped well, if their
AMR 99-033-001
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
diameter is approximately the same as the inner
curvature of the finger and thumb. However, most
objects are irregularly shaped, and are grasped with
only a two or three point contact. In order to hold
objects securely, the hands require a high pinch force
to due to the limited pinch contact area. Another
limitation of this design, is that the user of the
prosthesis must carefully pre-orient the hand with an
object to be grasped, in order to ensure that the
object will be held securely by the prosthesis. This
pre-orientation is done by compensatory body
motions of the the users elbow, shoulders, and torso,
in order to make up for the lack in degrees of freedom
of the prosthesis. These compensatory body motions
are awkward, can become tiresome after many
repetitions, and make the users look unnatural as they
grasp objects. In terms of cosmetic appearance, there
has been some criticism [3] from the users of these
hands. Typical complaints are that the hands are too
boxy or that they look unnatural during a grasp, both
after the grasp is achieved, and during the grasping
motion.
A number of anthropometric experimental hands
have been developed in the last fifteen years, that
have attempted to overcome limitations of the
conventional single degree of freedom prostheses.
These experimental designs feature multi-segment
fingers that can curl as they flex (The word flex, is
the terminology used to describe the closing action of
a finger, hand or limb). The fingers are also able to
flex inwards independently of each other to better
conform around an object, which is termed as
adaptive grasp. All designs also employ a thumb that
is capable of rotating about the palm axis.
December 1999
increases significantly over that of other similar
hands. Also, the use of multiple motors to drive the
fingers increases weight, space and power
requirements. This hand employs multi-segmented
curling fingers which are restricted to a single degree
of freedom as they curl.
The Belgrade Hand [9-10] is an anthropometric
robot hand. It has been designed for robotic
applications, and as such is not suitable for a
prosthesis, however it possesses some interesting
features. Like the Southampton Hand, it also
employs single degree of freedom, multi-segmented
fingers. The rationale for single degree of freedom
fingers is reduced complexity and reduced
computation requirements, in contrast to designs
where each finger segment (phalanx) is actively
controlled. The adaptive grasp of this design is
achieved by the combination of a passive mechanical
solution, and the use of multiple motors.
One of the more sophisticated anthropometric
robot hands, is the MIT/Utah Hand [11-12]. This
four fingered robot hand has 16 independently
controllable degrees of freedom. The complexity of
the hand is high, and the system of actuators used to
drive the hand, make it unsuitable for use as a
prosthesis. Each phalanx of each finger is controlled
by an antagonistic pair of tendons. Further, each
phalanx is equipped with a sensor to detect contact
with an object. The resulting design is able to
achieve a great variety of configurations and is
capable of computer controlled adaptive grasp.
However, the control system is computationally
intensive and complex.
Design of the Prototype Hand:
The Montreal Hand [4-6] is an experimental,
adult sized prosthesis featuring multi-segmented
curling fingers and adaptive grasp. The adaptive
grasp of this hand is passive in that the fingers
automatically conform around the object being
grasped, without the use of sensors. The mechanism
used to achieve adaptive grasp is based on a clutch,
spring loaded pulley and cable system. During trials
of the Montreal Hand, it was found that there were
reliability problems[3], due to the cable design.
A prototype prosthetic hand was designed in an
attempt to increase mechanical function and cosmetic
appearance over that of conventional prostheses, by
implementing various suggestions from rehabilitation
professionals [3], and examining the experimental
hands listed. In addition, a survey on individuals
with upper limb loss, to quantify and rate preferences
for improvements to conventional prosthesis [13],
was used.
The Southampton Hand [7-8] is an experimental
prosthesis that is also capable of five fingered
adaptive grasp. However, it has been achieved by the
use of four motors, sensors and computer control for
finger coordination. With the addition of computer
control and sensors, the complexity of the hand
A combination of three features give the
prototype hand the ability of passive adaptive grasp.
Firstly, the fingers are able to curl while they flex.
Secondly, the fingers are able to flex independently
of each other during closing, due to a parallel spring
mechanism within the palm. Finally, the thumb is
AMR 99-033-002
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
able to adduct and abduct (rotational motion) as well
as flex and extend.
A passive adaptive grasp system refers to the
fact that all five fingers conform around the contours
of the object they grasp, without any active control.
There are no sensors, or computers within this design
to actively co-ordinate finger motion. Instead, the
adaptive mechanism relies on the physical contact
force of the fingers with an object, to adjust the
positions of the fingers relative to each other. This
allows for a simple design that can fit within a small
space and is low in weight. The adaptive mechanism
is based on a parallel spring mechanism within the
palm of the hand, and is described in detail in [1415].
FINGER DESIGN
December 1999
and distal phalanges respectively, within a natural
finger. Links 4, 5 and 6 are connecting links that
drive the motion of the first three links. All four
fingers are exactly the same, which allows for a
savings on manufacturing costs and easy
serviceability due to interchangeable parts.
The assembled finger is shown in Figure 2(b).
It is connected to the palm knuckle via a revolute
joint on the top right end of link 1. The finger is
actuated by driving link 6 left for extension and right
for flexion, via a pin which passes through the hole
located on its right end. The pin travels in a slot, as
illustrated in Figure 3, which constrains the pin
motion to the x-axis. This finger has a single degree
of freedom, allowing it to move through a specific
path in space. The finger tip trajectory is illustrated
by the dashed line.
A number of requirements dictate the finger
design. Firstly, the fingers are to appear as natural as
possible, when still or when in motion. They must be
sized like that of a child in the 7 to 11 year range,
both in length and cross section. They are to be as
lightweight as possible, yet be strong enough to
withstand the operating conditions. Finally, the
finger design must be able to transmit a certain force
for pinch.
RIGID LINK FINGER
A three segment finger design was chosen since
this would look most like a human finger. The
natural finger has three segments, known as
phalanges. To approximate the finger, a rigid
linkage system was developed, as illustrated in
Figure 2.
Figure 2.
Diagram of Finger Links and Assembly
Each finger on the prototype hand is comprised
of six links, as shown in Figure 2(a). Links 1, 2 and
3 correspond to the proximal, medial
Figure 3. Single Degree of Freedom Finger Motion
The trajectory shown has been designed to
closely approximate the flexion path of a natural
finger during typical grasping. The trajectory of
natural fingers will vary depending on the size of an
object and the task [16], so it would not be possible
to mimic all of these trajectories with a single degree
of freedom finger.
After many iterations, and with the aid of
Working Model 2D software [17], satisfactory
kinematics and dynamics of the finger were achieved.
The software helped to create a good finger tip
trajectory, to minimise finger link dimensions, and to
maximise the mechanical efficiency of the finger.
This was not an easy task due to the number of
variables, and the interrelations between the three
parameters. For example, a compromise had to be
reached between finger thickness (measured from the
topside of a finger to the bottom side) and finger
AMR 99-033-003
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
mechanical efficiency (defined as the input force at
link 6 of Figure 3, over the tip pinch force at link 3).
The thicker the finger could be made, the better the
mechanical efficiency. However, the thicker the
finger became, the less useful it was for grasping
tasks and the less cosmetic it was. Priority was
placed on finger thickness over mechanical
efficiency, for cosmetic and functional reasons.
Another compromise had to be made between the
finger tip trajectory and the finger link sizes. Since
trajectory was deemed most important, the medial
phalanx (link 2) of the finger was actually made
shorter than anthropometrically normal. However,
this shortening of the medial phalanx helped
mechanical efficiency slightly.
Hundreds of
iterations were performed, leading to the final result.
The analysis with Working Model 2D software
determined the link lengths required for the finger
and also revealed the amplitude and direction of the
forces that each link would be subjected to during
grasping. In order to test whether the individual link
designs could withstand these forces, a finite element
software package called Ideas 5.1 [18] was used to
simulate the stresses within the finger links. The
amplitudes and directions of the forces found during
Working Model 2D simulations were reprogrammed
into the Ideas 5.1 finite element link models.
Simulation results showed that often, the most
stressed area of the models was in the material around
the pin joints. Ideas 5.1 was useful in determining the
thickness of the links to minimise their weight and
size, yet to keep them adequately strong. Generally,
the links were designed such that the principal
stresses and the shear stresses experienced, were 5
times lower than the ultimate tensile strength of the
material.
December 1999
continue to flex inward and wrap around the object.
This is referred to as finger wrap and is not possible
with a single degree of freedom finger. The
Southampton Hand and the Belgrade Hand described
previously, also have this finger limitation.
Figure 4.
Limits of Single D.O.F. Fingers
CABLE FINGER DESIGN
Since the finger wrap limitation, when using rigid
links, was well known during the start of the
prototype hand design, a finger wrap design using a
cable was initially attempted. The idea was to keep
the finger actuation limited to a single degree of
freedom, yet to add compliant finger wrap by making
use of the flexible properties of a cable. Figure 5
illustrates the cable finger design.
LIMITATIONS OF RIGID LINK FINGERS
The single degree of freedom finger design
needs one input to completely define the shape and
trajectory of the finger, as illustrated in Figure 3.
However, there is a grasping limitation when using
this type of finger. When any of the phalanges of the
finger encounter an object, finger flexion comes to a
stop. For example, if the prototype hand grasps an
object irregularly in such a way that the proximal
phalanx of a finger makes contact first, that fingers
motion will stop. The medial and distal phalanges of
that finger will be stopped in space, as shown in
Figure 4. This type of grasp looks unnatural. If the
proximal phalanx of a natural hand made contact with
the object first, the medial and distal phalanges would
Figure 5.
Cable Finger Design
The cable finger uses three links, one representing
each phalanx, and makes use of a flexible cable. The
cable is shown as a dotted grey line and runs through
metal tubes rigidly fixed onto the palm, the proximal
AMR 99-033-004
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
December 1999
phalanx link and the medial phalanx link. It is
attached at a point on the distal phalanx link. To
operate the finger, the cable is drawn back via an
actuator, through the tube rigidly linked to the palm,
causing the finger to flex closed. When grasping a
large object in a power grip, this design works
reasonably well.
However, there are some major problems with
this design. The first is that this finger design is not
appropriate for precision grasps involving finger tip
and thumb tip contact (bi-digital pinch), because the
finger becomes unstable. Working Model 2D
simulations of the bi-digital pinch confirmed the
instability. Upon formation of the bi-digital pinch,
the medial interphalangeal joint would buckle
inward, as shown in Figure 6, and cause the pinch to
collapse. In the process of this collapse, the object
within the grasp would be ejected outwards.
Figure 7.
Cable Finger Tip Trajectory
Observe that the cable is attached to the distal
phalanx, but only slides through the medial and
proximal phalanges via the tubes. When tension is
applied to the cable, a torque is developed around
joint A and the distal phalanx rotates, however, the
cable does not transmit any force to the medial or
proximal phalanges. Only when the distal phalanx
fully flexes, can the cable create a torque around joint
B, and later still around joint C.
Finally, a minor disadvantage is that the cable
design needs springs to return the fingers to the open
position when the cable tension is relieved.
Therefore, the cable finger design was abandoned
due to its disadvantages.
Figure 6.
Cable Finger Collapse During Pinch
The second problem with this design is that the
fingers would not close in a trajectory suitable for
grasping objects. This trajectory was also very noncosmetic in dynamic appearance. During finger
flexion, first the distal phalanx link would flex fully,
then the medial phalanx link would flex somewhat,
finally followed by the proximal phalanx link. This
incorrect motion is shown in Figure 7. During the
close of a natural hand, all of the phalanges flex
inward at the same time, with the proximal phalanx
flexing through a greater angle during the process. In
this way, the natural hand can either pinch an object,
or grasp an object and continue to wrap the
remaining phalanges around the object. The reason
for the incorrect trajectory of the cable finger design
can best be explained with the use of Figure 7.
The cable design was originally pursued after
investigation of a plastic toy robot hand with flexing
fingers. It is interesting to note that the cable finger
design will work under the special condition where
the cable is semi-flexible, as was the case for this toy.
The cable itself could be a thick nylon or
polypropylene tendon, such that when tension is
applied to it, the flexing of the tendon defines the
finger shape, not the phalanges surrounding the cable.
Because the tendon has memory it returns to its
original shape after the tension is released,
eliminating the need for springs. The only drawback
of this design is the great amount of energy that is
needed to constantly flex this semi-rigid tendon.
Such designs are currently being pursued by a group
in California[19].
TWO D.O.F. RIGID LINK FINGERS
The single degree of freedom fingers that have
been designed for the prototype hand, as shown in
AMR 99-033-005
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
Figures 3 and 4, can be converted into two degree of
freedom fingers quite easily. This was not done for
the prototype hand due to space requirements and the
need for a second actuator, but it is shown here to
demonstrate the versatility of the rigid link design.
One degree of freedom can control the flexion and
extension of the finger as before, while the additional
degree of freedom can control the curl or trajectory
of the finger. This is achieved by greater control of
link 6 in the finger as depicted in Figure 8.
December 1999
THUMB DESIGN
The distal portion of the thumb is similar to that
of the fingers, however the thumb uses only four
links. The thumb is able to flex and extend only the
two outer segments (distal phalanx, and medial
phalanx). The base segment (carpometacarpal
phalanx) cannot flex or extend, but is able to adduct
and abduct (rotate). Rotation of the thumb is
achieved by having the operator of the prosthesis use
his opposite hand to position the thumb, or by
pushing the thumb against a fixed object, such as a
table, to move it to the desired position.
The thumb is illustrated in Figure 9. Link 4T is
the drive link that actuates the thumb. The entire
assembly is able to rotate about the dashed line
labelled as the thumb rotational axis. This is
achieved by rotation of the thumb base about the
palm base shaft.
Figure 8.
Two Degree of Freedom Design
Link 6 in the prototype hand is connected at
three points. One of these points is the slot pin which
slides through the straight slot in the x-axis. If the
link 6 slot pin is pulled in the positive x-direction, as
indicated in Figure 8, the finger will flex closed. If it
is driven in the negative x-direction, the finger will
extend open. In order to add a second degree of
freedom to this finger design, a mechanism could be
created that could drive the link 6 slot pin in the ydirection, in addition to the x-direction. By driving
the slot pin in the positive y-direction, the finger
would curl inwards more tightly during flexion.
Alternatively, by driving the slot pin in the negative
y-direction, the finger would curl less tightly
(straighten out more) during flexion. This curling
inward or straightening out motion would be
independent of the flexion or extension of the finger.
Together, if controlled properly, these two degrees of
freedom could add increased functionality and
dynamic cosmesis to a hand design.
To implement the two D.O.F. design, an
additional actuator would be needed, more space
would be required, and a proper control system for
finger coordination would need to be developed.
Figure 9.
Diagram of Thumb Assembly
The drive cable provides the mechanical actuation
needed to flex and extend the two distal segments
(phalanges) of the thumb. Flexion is achieved by
applying tension to the drive cable, which is looped
over the thumb pulley, and connected to link 4T.
When the tension in the drive cable is relieved, the
thumb return spring exerts a force on link 4T to
return the thumb to the extended position. Figure 10
illustrates the adduction and abduction (rotational)
range of motion of the thumb. The use of a cable
provides for a very compact thumb design. The key
to the design is to keep the drive cable coaxial with
the thumb rotational axis. In this way, no matter
which angle the thumb assembly is rotated
AMR 99-033-006
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
to, the drive cable will always be able to flex and
extend the thumb without slipping off the thumb
pulley. The drive cable is connected to the adaptive
grasp mechanism at the other end.
Figure 10.
Thumb Range of Motion (Front View)
The thumb can be rotated, within the angles
shown, giving the hand the ability to grasp flat
objects such as a credit card or key with the thumb
out to the side, or to carry objects with only the
fingers, such as a suitcase.
Results and Discussion:
One prototype hand was built according to the
designs described. Figure 11 shows the prototype
hand from the dorsal side, with fingers in the
extended position and the thumb abducted (rotated
out to the side position).
Figure 11.
Prototype Hand, Dorsal View
December 1999
The cosmetic appearance of the hand is very
good. It is proportioned more or less like a childs
hand. The fingers are 66 mm long (from knuckle to
tip), 9 mm wide and 11 mm thick. The palm is 80
mm long, 65 mm wide and 25 mm thick. The palm is
longer than normal, but future design changes can
shorten it by up to 12 mm. To improve cosmetic
appearance, a conventional prosthesis glove was
heated and stretched to fit over top the prototype
hand. With the glove on, the dynamic cosmesis of
the hand, that is, the appearance of the hand while it
is in motion, is excellent.
The mass of the prototype hand alone is 217
grams without the glove. The mass of the hand
including the motor, gearbox and wrist structure is
280 grams. The motor used with the hand is a 6 Volt
MicroMo 1724 with a 22:1 gearbox. Aluminium
(7075-T6) was used for the finger links. The palm
and thumb base were made of Delrin plastic. The
cable is made of Kevlar and the pulleys on which it
rotates over are made of aluminium. In comparison,
the Montreal Hand, which is capable of similar
mechanical function to the prototype hand, has a
mass of 540 grams [6].
The hand is able to achieve a maximum pinch
force of 14 N. This is approximately one-third the
pinch force of a conventional prosthesis, such as the
VASI 7-11 hands. Although the adaptive grasp may
secure some objects better, precision pinches of items
such as a fork would still require a high pinch force.
Therefore, the pinch force must be improved upon.
Also, the time to fully close the hand and achieve a
firm pinch is 4 to 5 seconds, which is too slow to be
practical. It should be in the range of 1 to 1.5
seconds. Both of these deficiencies are interrelated
and can be overcome with the implementation of a
two speed automatic transmission. Such a device has
been designed for prosthetic hands [20]. The Otto
Bock 2000 prosthesis [1], is a conventional, mass
produced prosthesis that also uses a two speed
automatic transmission. The transmission would
serve to increase the pinch force (when geared low)
while decreasing the closing and opening times
(when geared high).
Pull-out tests were done to determine the amount
of force required to pull objects out of the hands
grasp. The grasp stability was found to be greater for
small objects, when the fingers were able to partially
or completely enclose objects by curling and adapting
around them. Even with a pinch force of only 14 N,
AMR 99-033-007
Proceedings of the Sixth National Applied Mechanisms & Robotics Conference
these objects could not be easily removed from the
hand in any direction. This enclosure of small objects
by the fingers is not possible with a conventional
prosthesis.
The size and weight of the prototype hand are
comparable to a conventional child prosthesis, but
currently the energy use, the closing/opening times
and the pinch force are not able to match a
conventional prosthesis. It is expected that with the
addition of a transmission, the pinch force and time
deficiencies will be solved, but this will be at the cost
of increased weight and energy use.
Conclusion:
A child sized prosthesis has been designed
which incorporates a passive adaptive grasp
mechanism. The adaptive grasp of this hand is
achieved in three ways. Firstly, the fingers are able to
curl as they flex, and secondly an internal spring
mechanism allows the fingers to flex inwards
independently of one another. The finger design has
been achieved with the use of multi-segmented
fingers that are restricted to a single degree of
freedom. Although limited in some respects, this
finger design greatly increases hand adaptability, with
relatively little increase in complexity. Thirdly, the
thumb can be passively rotated, providing more
grasping configurations. These features have been
built into a working prototype that is significantly
smaller than, and almost half the weight of any other
experimental hand in its class.
References:
[1] Otto Bock Orthopaedic Industry GmbH, MYOBOCKArm Components 1997/98, Otto Bock(1997).
[2] Variety Ability Systems Inc., Small and Lightweight
Electric Hands for Children, VASI, Toronto(1996).
[3] Ramdial, S.;Wierzba, S., Personal Communication,
Myoelectrics Service, Bloorview MacMillan Centre,
Toronto.
[4] Lozach, Y.; Drouin, G.; Vinet, R.; Beaudry, N., A
new multifunctional hand prosthesis, Proceedings of the
International Conference of the Association for the
Advancement of Rehabilitation Technology, Montreal, pp
86-87, 1988
[5] Lozach, Y.; Drouin, G.; Vinet, R.; Chagnon, M.;
Pelletier, M., Specification for a New Multifunctional
Hand Prosthesis, Proceedings of RESNA 1986,
Minneapolis, Minnesota, pp 117-119, 1986
[6] Lozach, Y.; Madon, S., Clinical Evaluation of a
Multifunctional Hand Prosthesis, Proceedings of RESNA
International 1992, Toronto, 1992
December 1999
[7] Kyberd, P.J.; Chappell, P.H., The Southampton Hand:
An intelligent myoelectric prosthesis, Journal of
Rehabilitation Research and Development, Vol.31, No. 4,
pp 326-334, 1994.
[8] Chappell, P.H.; Kyberd, P.J., Prehensile control of a
hand prosthesis by a microcontroller, Journal of
Biomedical Engineering, Vol. 13, pp363 -369, 1991
[9] Bekey, G.A.; Tomovic, R.; Zeljkovic, I., Control
Architecture for the Belgrade/USC Hand, pp 136-149. In:
Venkataraman, S.T.; Iberall, T, Dextrous Robot Hands,
Springer-Verlag New York Inc, New York, 1990
[10] Rakic, Miodrag, Multifingered Robot Hand With
Selfadaptability, Robotics & Computer-Integrated
Manufacturing, Vol.5, No. 2/3, pp.269-276, 1989
[11] Jacobsen, S.C.; Wood, J.E.; Knutti, D.F.; Biggers,
K.B., The Utah/M.I.T. Dextrous Hand: Work in
Progress, The International Journal of Robotics Research,
Vol. 3, No. 4, pp 21-50, 1984
[12] McCammon, I.D.; Jacobsen, S.C., Tactile Sensing
and Control for the Utah/MIT Hand, pp 239-266. In:
Venkataraman, S.T.; Iberall, T, Dextrous Robot Hands,
Springer-Verlag New York Inc, New York, 1990
[13] Atkins, D.J.; Heard, D.C.Y.; Donovan, W.H.,
Epidemiologic Overview of Individuals with Upper-Limb
Loss and Their Reported Research Priorities, Journal of
Prosthetics and Orthotics, Volume 8, Number 1, pp 2-11,
1996
[14] Dechev, N, Design of a Multi-Fingered, Passive
Adaptive Grasp Prosthetic Hand: Better Function and
Cosmesis, M.A.Sc. Thesis, Department of Mechanical and
Industrial Engineering, University of Toronto, 1998
[15] Dechev, N.; Cleghorn, W. L.; Naumann, S., Multiple
Finger, Passive Adaptive Grasp Prosthetic Hand,
Mechanism and Machine Theory, Submitted for
Publication, June 99.
[16] Taylor, C.L., Patterns of Hand Prehension in
Common Activities, Engineering Prosthesis Research
No.3, Department of Engineering, University of California
Los Angeles, 1948
[17] Knowledge Revolution, Working Model 2D version
4.0, 1997, Knowledge Revolution, 66 Bovet Road, Suite
200, San Mateo, CA.
[18] Structural Dynamics Research Corporation, I-DEAS
5.1, 1997, Structural Dynamics Research Corporation,
2000 Eastman Dr., Milford, Ohio.
[19] LeBlanc, M.; Setoguchi, Y.; Bowen, W.D.; Milner,
C.; Burkholder, F.; Chen, S., Design Concepts for an
Endoskeletal Child-Sized Hand, Journal of Prosthetics
and Orthotics, Vol. 9, Num. 3, pp. 123-126, 1997
[20] Guo, G.; Quian, X; Gruver, W.A., A Single-DOF
Multi-Functional Prosthetic Hand Mechanism with an
Automatically Variable Speed Transmission, pp 149 -154.
In: Robotics, Spatial Mechanisms, and Mechanical
Systems, ASME, New York, DE-Vol.45, 1992
AMR 99-033-008
