Hand prosthesis design: Enhancing grasping capabilities through mechanical features

20.2 Hand Prostheses
The loss of the hand has a double consequence for the amputee: a drastic reduction
of the functionality (the amputee becomes unable to carry out most of the manip-
ulation and grasping tasks) and the beginning of psychological problems (the am-
putation modifies the cosmetic appearance of the upper extremity). The prosthesis
device intends to offer the possibility to reestablish the functionality and the cosmetic
appearance of the missing limb. The international norms define the prosthetic device
as an orthopedic device that replaces, even if partially, the amputated limb restoring
both functionality and also cosmetics.
From a functional point of view, hand prostheses can be classified in two main
groups: passive and active prostheses (see Figure 1).
Figure 1: Prosthetic hand classification
0.2.1 Passive Prostheses
Passive devices can be adopted in order to reestablish the cosmetic appearance of the
missing limb. A passive prosthesis hand cannot be actively controlled by the amputee.
Cosmetic hands can provide excellent aesthetics (see Figure 2). Passive prosthetic
hands are usually adopted by monolateral amputees, they can use the prosthesis as
a support for simple tasks. Passive devices are simpler and lighter compared with
the active ones but they do not restore any grasping functionality; in fact, they are
characterized by the absence of DOFs. However this type of prosthesis provides
3Figure 2: Passive prosthetic hands
some limited function, such as pushing an object or aiding the remnant limb during
grasping or manipulation tasks [23].
0.2.2 Active Prostheses
An active prosthesis (see Figure 3) can be actively controlled by the amputee, active
prostheses can be further divided in:
 Body-powered prostheses;
 Myoelectric prostheses.
Body-powered prosthetic hands are activated by the harnessing of relative motions
of other bodily parts, for example: scapula adduction pulling on an harness that is
linked to the terminal device, which opens. The harness acts in balance with an elas-
tic element. Body-powered prosthetic hands are simpler and more affordable than
myoelectric prosthetic hands, but they can cause a discomfort due to the presence of
the control cable [78]. The functionality is quite limited; in fact, they are character-
ized by one DOF for the opening/closing movement.
Myoelectric prostheses are controlled using signals derived from muscle on the fore-
arm. The muscle contraction generates small voltage that can be detected, amplified
4(a) Body-powered (b) Myoelectric
Figure 3: Active prosthetic hands
and used as control signals. These are known as Electromyogram (EMG). Conven-
tional systems use a simple ON/OFF control: when the tension exceed a threshold
an ON command is achieved, a second muscle is then used to control the movement
of the hand in the other direction. There is no feedback within the hand, so the user
has to judge by sight when to stop moving the hand [56].
Commercial myoelectric prostheses are characterized by a maximum of two DOFs,
the first one for the opening/closing movement and the second one for the wrist ro-
tation (pronation-supination of the hand) [39].
0.3 Motivations
As reported in Section 0.2.2, commercial active prostheses are basically simple grip-
pers with one or two DOFs, which barely restore the capability of the thumb-index
pinch. Although this performance can be considered as acceptable for usual tasks,
commercial devices have the potential to be improved significantly [9].
Despite of several research efforts aimed at innovating artificial hands technology,
surveys on user’s satisfaction in using prosthetic hands revealed that 30 to 50% of the
upper extremity amputees do not use their prosthetic hand regularly [29]. The main
5factors that cause this loss of interest for active hand prostheses can be synthesized
in three points [14]:
 Low functionality;
 Low cosmetics;
 Low controllability.
It is important to point out that prosthetic hands are designed primarily for grasping
tasks and not for manipulation tasks; manipulation requires advanced mechanical
features and sensors, complex control strategies and natural interfaces between the
peripheral nervous system and the artificial device [21], [58]. Although the research
in the field of robotic hands has produced several advanced mechanical hands (see
[6] for a complete analysis of the state of the art), some requirements are still not
matched for two main reasons: the lack of lightweight and compact actuators with
high output torque [44] and the lack of a natural interface.
0.3.1 Low Functionality
The adoption of a prosthetic device reduces the prehensile movements to two main
categories:
 Precision grasp;
 Power grasp.
The main feature of the second is the ability to handle heavier weight as compared
with the first one that allows the accurate grasping of little objects [67]. In the
precision grip the contact points are located at the fingertip of the thumb, index
and middle finger; the thumb works in opposition to the other fingers. On the other
hand, power grasp is characterized by a wider interested contact surface with the
object; the grasped object is held in the hollow formed by the thumb, the palm and
the finger, thus the fundamental role played by the inner parts of the hand (palm
6and proximal phalanxes) to enhance both the stability of the grip can be frequently
observed [6]. Little and ring fingers are utilized for augmenting the grasp stability
during the prehensile movements [4] (see Figure 4). While precision grasp can be
(a) Precision grasp (b) Power grasp
Figure 4: Precision and power grasp
easily performed by a commercial prosthetic hand; on the contrary, power grasp is
prevented by the low flexibility of the device.
Power grasps are characterized by multiple points of contact between the grasped
object and the surfaces of the finger and the palm. They maximize the load carrying
capabilities and present a high stability due to a large number of distributed contact
points on the grasped object. Furthermore, the location of the contact points on the
grasped object becomes less critical for stability due to the redundant numbers of
contact.
Due to the lack of DOFs prosthetic hands are characterized by a low functionality
particularly related to power grasping. This characteristic allows for only two or three
contact points with objects that are grasped, as a result high grip forces are required
to succeed in the grasping task. The prosthetic device must rely on friction forces to
maintain the object within the hand, thereby requiring precise and conscious effort
on the part of the wearer to ensure optimum grip.
70.3.2 Low Cosmetics
Two main problems are related to the cosmetic appearance of the prosthetic device:
active prostheses turn out to be bulky and heavy due to the presence of the actuator
and transmission system within the hand structure; moreover, the lack of DOFs
affects the finger movements and the grasping capabilities that result unnatural.
0.3.3 Low Controllability
Although advanced prostheses have been developed, the control of such devices re-
quires a considerable training and a great attention during grasping activities from
the amputee. Commercial prostheses are equipped with a limited set of sensors, con-
sequently, there is no feedback within the hand and the user has to judge by sight
when to stop moving hand. Moreover using EMGs derived from opposing muscles on
the forearm allows the active control of maximum two DOFs and the execution of
simple movements unless sophisticated control strategies are introduced.
0.4 Contributions
In order to overcome the limitations above mentioned generally associated with com-
mercial prostheses, the following criteria must be addressed [74]:
Functionality: the prosthetic device should perform a stable power grasp with a
wide variety of object with complex shapes;
Cosmetic: the prosthesis should have the same weight and dimensions of the hu-
man hand. Furthermore, the grasping movements should results as natural as
possible;
Controllability: the prosthesis should be easy to operate.
This thesis presents two different approach aiming at solving the listed weaknesses:
the first one is based on an integrated design aimed at embedding different functions
8(actuation, transmission, sensing and control) within a housing closely replicating the
shape, size, and appearance of the human hand. This approach can be synthesized
with the term: “biomechatronic” design that leads to an active adaptive grasp [11].
Adaptive grasp is the ability of the finger and thumb to adapt to the shape of the
grasped object in order to increase the number of contact points between hand and
grasped object during a power grasp [25].
The second approach is based on the exploitation of underactuated mechanisms
for the design of an innovative prosthetic hand capable of passive adaptive grasp. In
summary, this work illustrates the difference in design approaches between active and
passive adaptive grasp.
The thesis is organized as follows:
In Chapter 1 the biomechatronic design approach is presented and the develop-
ment of a prosthetic hand (RTR I hand) based on this approach is described.
In Chapter 2 the concept of underactuation is discussed and the state of the art
concerning underactuated robotic and prosthetic hands will be presented.
In Chapter 3 two examples of underactuated prosthetic hand are reported: the
RTR II hand, based on an underactuated mechanism previously developed by Hirose
et al. [46] and the SPRING hand, based on an innovative underactuated mechanism.
In Chapter 4 the adoption of compliant mechanisms in the design of underactuated
mechanical grippers is discussed in order to demonstrate that the use of compliant
mechanisms can represent a further reduction of mechanical complexity.
Chapter 1
Biomechatronic Design
Commercially hand prostheses have limited object stability during prehension. In this
Chapter the active adaptability is addressed, an increase in number of active degrees of
freedom will improve the adaptability and the grip stability of these devices; on the other
hand, the necessary grip force will be minimized [62]. Limited space to install actuators is
one of the most difficult problem in the design of prosthetic hands with multi-DOFs.
9
10
1.1 Introduction
As reported in 0.3 commercial hand prostheses have one or two DOFs providing fin-
ger movement and thumb opposition; such device are characterized by a low grasping
functionality, in fact, they do not allow adequate encirclement of the grasped object;
as a result, contact areas between fingers and grasped object are small, and high grip
forces are required to succeed in the grasping task. In fact, the prosthetic device must
rely on friction forces to maintain the object within the hand. Therefore object must
be grasped accurately to be held securely. In order to fit the myoelectric prostheses
for different amputation levels, all the actuators have to be embedded in the hand
structure (intrinsic actuation). Due to this, complex mechanisms and a high number
of actuators can not be embedded within the device. This design approach leads to
an extreme reduction of available DOFs. The final consequence is that commercial
prostheses do not allow shape adaptation while the motion of the phalanxes is deter-
mined at the design stage (see Figure 1.1).
Figure 1.1: The motion of the phalanxes is determined at the design stage, no shape
adaptation is possible
The objective described in this Chapter is to develop a myoelectric prosthetic hand
aimed at increasing the grasping capabilities through a multi-DOFs design approach.
It has been demonstrated that methodologies and knowledge developed for robotic
hands can be applied to the domain of rehabilitation to augment final performances.
Multi-DOFs robotic hands, developed for research and space application, represent
the state of the art concerning the design and development of end-effectors; for this
reason, the design approach of these devices have to be taken into consideration for the
11
development of an innovative prosthetic hand. In Section 1.2 a comparative analysis
of the state of the art concerning robotic hands will be reported.
1.2 Robotic Hands
Impressive dexterous hands have been built in the past [79], [59]. The popularity of
designing and building robotic hands is demonstrated by the large number of uni-
versities and research organization that have hands named after them. In the past,
dexterous hands have been developed to perform laboratory research on grasping and
finger manipulation. With this objective J. K. Salisbury designed the Stanford/JPL
hand [82]. The hand has three fingers, each of them has three DOFs and four control
cables; the hand is controlled by an actuator pack of 12 DC servo motors with 25:1
speed reducers (see Figure 1.2). The majority of Salisbury control work has been in
the area of fingertip prehension.
Figure 1.2: The Stanford/JPL hand
The same field has been investigated with the Utah/MIT hand that closely copies
the outward appearance of the human hand [54]. The Utah/MIT dexterous hand has
four DOFs in each of the three fingers, and a four DOFs thumb (see Figure 1.3). The
geometry of the hand is roughly anthropomorphic. The thumb is, however, perma-
nently in opposition and the phalanx lengths and joint positions have been altered to
facilitate the routing of tendons. The 16 DOFs hand is actuated using an antagonistic
tendon approach, which requires a system of 32 independent polymeric tendons and
pneumatic actuators. The pneumatic actuators are fast, low friction, and can gen-
12
erate relatively high forces. The lowest level of control for the Utah/MIT dexterous
hand includes an analog controller for each of the 16 DOFs which executes position
control and tendon management.
Figure 1.3: The Utah/MIT hand
Bekey et al. [3] developed a four DOFs, five-fingered end-effector called the Bel-
grade/USC hand (see Figure 1.4). The hand has four fingers with three joints each,
each pair of finger is driven by one motor. The motion of the three joints is not
independent but is modelled on observation on human hand during grasping. The
articulated thumb moves in an arc into opposition to one or more fingers, another
motor flexes and extents it at its second joint. The volar surfaces of the fingers and
the palm are covered with pressure sensors.
The Belgrade/USC hand can be useful as a model for an innovative prosthetic
hand. It executes the grasp phase adapting the finger positions to the shape of the
object. In contrast with the approach taken in the development of the Utah/MIT
hand, Bekey et al. believe that the grasp control should reside within the hand [52].
An alternative approach was represented by the Hitachi hand with its shape mem-
ory alloy (SMA) actuation technology. The hand was characterized by a high power-
to-weight ratio and a high compactness. The Hitachi hand used a large number of
thin SMA wires; each finger had 0.02mm diameter SMA wires that were set around
the tube housing of the spring actuators. The SMA wire, when heated by passing
electric current through it, reacted by contracting against the force of the spring (see
Figure 1.5).
More recently, the DLR (Deutches zentrum fur Luft-und Raumfahrt) has devel-
13
Figure 1.4: The Belgrade/USC hand
Figure 1.5: The Hitachi hand
oped a multisensory four fingered hand with in total twelve DOFs with the declared
goal to integrate all the actuators in the hand palm or directly in the fingers [8]. Force
transmission in the fingers is realized by special tendon, which are optimal in terms
of low weight and backlash despite of fairly linear behavior. Each finger shows up a
two DOFs base joint realized by artificial muscles and a third actuator of this type
integrated into the bottom finger link. The aim of this project is to develop a robotic
hand for space operations e.g., handling drawers, doors, and bayonet closures in an
internal lab environment. In 2001, DLR presented the second prototype of the DLR
hand; the main target, developing the second prototype, has been the improvement
of the grasping performance in case of precision and power-grasp [7] (see Figure 1.6).
The development of a robotic hand for space operations is currently ongoing also
in the Robotic Systems Technology Branch at the NASA Johnson Space Centre [63],
14
(a) DLR I (b) DLR II
Figure 1.6: The DLR’s robotic hands
[32]. The goal of the Robonaut project is to reduce the extra-vehicular activity (EVA)
burden on space station crew and also to serve in a rapid response capacity. The
Robonaut hand has a total of fourteen degrees of freedom. It consists of a forearm
which houses the motors and drive electronics, a two DOFs wrist, and a five finger,
twelve DOFs hand. The forearm, which measures four inches in diameter at its base
and is approximately eight inches long, houses all fourteen motors, twelve separate
circuit boards, and all the wiring for the hand. The hand itself is broken down into
two sections: a dexterous work set which is used for manipulation and a grasping set
which allows the hand to maintain stable grasp while manipulating or actuating a
given object (see Figure 1.7).