1.0   Introduction

Throughout history, mankind has long since dreamed of developing
automated systems to take over tasks completed by man. Since the dawn of the
machine age, they have aided in the creation and destruction of jobs formally
completed through manual labour, leading to an overall improvement in
efficiency and performance 1.
Robotics have brought benefits to a number of industries such as commercialised
agriculture, manufacturing and space exploration 2. Today, robots are
used in most professional disciplines to execute menial to highly complex
tasks.

The hand after the brain is the most sophisticated organ of
the Human body 3.
There is a theory that the hand, along with other parts of the body, and their
ability to manipulate technology, has played a role in our evolution 4 and thus, hands have
essentially ‘Shaped the brain’ 5. As the early use of
sophisticated technology by humans increased, so did the demand for cognitive
ability 6. One justifying the
existence of the other. Thomas Johannesson, former dean at Lund Technical
College, describes how development of the hand has been characterized by the
interaction between man’s inner activities (thoughts) and his outer activities
(Handicrafts) 7. It is the foremost
means in which human’s transfer thought into physical output. Our hands are
centrally involved in many of our daily activities from reaching, grasping and
manipulating objects in an almost effortless manner. To arrive at a
comprehensive picture of what it requires to make hands a versatile and central
tool it is essential to consider the connection between cognition and physical
output 8.

In 1959 ‘the father of robotics’ Joseph Engelberger, invented
the first robotic arm (and gripper); the Unimate #001. This model was installed
in a General Motors factory to execute dangerous tasks in place of humans. It’s
presence in the factory completely transformed the industry of manufacturing 9. Since then there
has been 100’s of engineering attempts in both industry and academics to
emulate the human hands dexterity and manipulation for purposes from factory
line assembly to prosthetics. The field of grippers, in this sense, has become
extremely diverse. Many researchers have gone down the path aiming for general
anthropomorphism, in the belief that the grasping environment is defined by the
human morphology the best performing gripper would the closest replication of a
human hand. These grippers normally stipulate a high degree of freedom, five
fingered hand. the and others executing designs with a specialized task
intended

In general, all the hands that are produced tend to be
benchmarked quite ambiguously in a variation of existing Hand and grasp
taxonomies. A proposed reason for this is that the development of each hand,
due to the limitation in the field, is aimed at a relatively small area of
purpose. Thus, engineers choose the metrics for their hand based on their goals.

The focus of this study will be developing a highly capable
yet simplistic, anthropomorphic gripper to be attached to a robotic arm
belonging to a service robot. A service robot is, as defined by the
International federation of Robotics, ‘A
robot that performs useful tasks for humans or equipment excluding industrial
automation application’ 10.  A service robot, ‘Stevie’, is currently
being developed by Trinity college Dublin and is purposed to provide additive
health care provisions to elderly people within a home-based environment. The
gripper should achieve adequate levels of grasping and manipulation, it will
possess a wrist and hand to arm connection method. The capabilities of the
gripper and the design process will be defined by the Design Goal Question
Metric Method (dGQM) created by Conor Mcginn 11

1.1   Motivation and Challenges

With these
previous publications it is hard to select an area where there is room for
improvement or reason to design a new solution. The selection of a project must
be justified through the effect of its contribution to an area and the need in
the discipline. With this in regard, I will outline the motivations to produce
a new gripper and identify the challenges associated.

At the
moment, there is no functioning gripper associated with department which in my
research is being facilitated. The service robot, ‘Stevie’, whilst currently
not requiring a purpose-built gripper, will do so in the future. This project
is the second generation of grippers being developed by the department and the
first generation with ‘Stevie’ in mind.

Within the
department there is also a PhD student, Patrick Lynch, who’s thesis will be
concentrating on the global Vs. local ……. Need to know formal …… This will
require the gripper to be a platform which can adapted to switch modes of
visual sensing easily. The final product and the decisions made in the dGQM
methodologies design process should reflect these motivations accurately.

Most of the
challenges experienced in the project will arise during its process. As the
project will be one step in a series towards larger goals, the main challenge
will be in defining the objectives and their analysis once complete. The dGQM
methodology and benchmarking should aide in this regard. Also, as this gripper
is being built there must be careful regard for its’ ability to be a platform
for further development once complete.

Compromise
when building the hand will also be challenging. As the hand is being formed
being able to gauge the degree to which characteristics can be taxed on order
to concentrate on another will be difficult. In previous work, the ability to
be able to produce a successful gripper while balancing features such as
weight, grasping & manipulation abilities, cost and sensing is both
essential and challenging.

1.2   Research objectives

The aims and
objectives of this proposal derive from the method of dGQM, and from that, the
previous hand built by Noel Frisby is considered throughout.

The overall
purpose of this project is to answer the proposed question; “Is it
possible to create the design and control of a manipulator and wrist for
compliant applications in service robotics?”. The aims and objectives
derive from this.

The aim of
this project is to create an anthropomorphic gripper that works seamlessly with
the larger system of a service robot. The robot requires a gripper that can
successfully achieve a reasonable level of grasping and be a source point for
valuable sensing data such as grip force, temperature, imaging, positioning and
so on.

In the
process of building this hand I will pursue, with a slightly altered scope, to
improve the version finished by Noel Frisby in 2015 12. Noels hand,
although a success, had clear weaknesses. The project I am presenting will
attempt to improve on the areas thought to be most critical. The proposed
research will undertake the following main goals;

Sensing; The hand will possess several
sensors that produce valuable data fed back to the larger control system that
is beneficial to both the tasks of the hand and the larger robot system. A
closed loop system will be executed.

Wrist with at least 1 degree of
freedom (DOF); The
addition of an actuated wrist with at least 1 DOF is essential to a practical
gripper and will improve the grasping ability

Improved overall actuation; In Noels hand the method of
compliant actuation was, in practice, quite poor. I will attempt to produce a
cleaner actuation system with the better overall performance.

The aims and
goals of this project are not limited to the above. There are other smaller
goals that will contribute to the overall success. Some of these are; An
efficient control system, a final size close to a human, better overall design
appearance and quality, clean wiring, efficient cabling, improved motion,
gesturing, weight, grasp force, compliance etc.

 

1.3   Overview

The research
on robotic manipulators has been quite extensive since the conception of
robotics in 1959. In recent years the research and industrial field has become
more popular due to general advancements but more specifically due to the push
for a viable ‘Disaster Robot’ made shortly after the Fukashima disaster 13 14. There has been a
vast amount of publications on robotics, and with its popularity increasing
dramatically 15 this trend can be
expected to continue. Although these factors have contributed greatly, the
field remains quite primitive and the goal for realistic anthropomorphism in
manipulators is still quite some distance from being achieved.

In the past
decades a lot research has been dedicated to the development of highly
articulated, general-purpose robotic hands and prosthetics 6-10, but in spite
of continuing advances, most of the manipulators have not been successful in
achieving a large user base. Those hands that do perform well in their intended
fields tend to either come at a great cost or have a limited end functionality.

Sensing in
hands and closed loop control of grippers enables optimization of grasps.
Sensor heavy hands on the market mostly feature three key sensing technologies;
tactile, joint position measurement and actuation force measurement. Hands like
the DLR/HIT III hand 7 are some of the best sensing hands on the market.

Energy
consumption in hand hardware comes down to two factors: weight and method of
actuation. Actuation in hands is as important a field in robotics as grasping.
The research in robotic actuation concentrates principally on a combination of
three principles; power, weight and size. The number of actuators to joints
defining whether the hand is under or over-actuated. Previous work has been
done in various actuation methods, the most common of those being motors.
Recent research has been in various methods, namely pneumatic actuators, the
abilities of which have shown to be suitable for robotics 11. In this report,
an analysis will be produced to choose the most suitable actuator for an under-actuated,
tendon operated gripper.

Kinematics
and kinematic chains are essential in defining the movement in a multi-jointed
gripper. Kinematics are used by creating a model using the individual joints in
the hand.12. The hand model could be used to evaluate realistic hand
functionality.

This is an
overview of the area. The literature review section in this dissertation will
more provide a deeper analysis and knowledge of the area.

 needs reviewal

 

 

1.4   Dissertation Contribution

Assuming that the objectives that have been laid out are
achieved, it is the hope of the author that this dissertation will have made
the following contributions to the field:

       
I.           
The first use of the dGQM methodology in a practical
manner since publication.

      II.           
Production of a novel gripper that TBCTBCTBCTBCTBCTBC

    III.           
Design of a modular design that facilitates researcher
value in future projects.

 

 

2.0   Literature
Review

The
following is a literature review that has been carried out on the current state
of the art of robotic gripper devices. These have been collected and compared
in chosen areas that are deemed relevant to the area of research. Some topics,
that are less directly relevant are reviewed in a lesser extent to reflect
that.

The
different sections of the literature review reflect the areas that were
researched by the author to increase their knowledge in those areas and to be
able to accurately make decisions during both the experimental and design
process. The first part of the review considers gripper morphology with a concentration
on the human hand as the project is anthropomorphic. An overview on the different
types of grippers is shown to exemplify the current state of the art of
grippers. Kinematics are then reviewed as a starting point to the
considerations associated with gripper design. This leads to an in-depth
analysis in the existing literature around the various grasp taxonomies, making
sure to review the reasoning of their creation. As we begin to look at the
grippers motion; actuation, under actuation, compliance and passive dynamics
are overviewed with careful consideration to the options available,
characteristics of incorporation and examples of current implementation.
Sensing technology that has been used in robotic grippers is discussed in
detail with a focus on the justification of use and effectiveness in practical
application. Finally, some existing hand designs will be overviewed and
sectioned in a table format detailing existing features for benchmarking.

2.1  Anthropomorphism

Anthropomorphism
is ‘the attribution of human
characteristics to a god, animal or object’. In the field of robotics this
refers to the incorporation of human characteristics in the design of robotics.
The most common results are robots featuring a head, two arms and two legs,
such robots and the reasons featuring these designs are explained 16. Anthropomorphism is
one of the most important topics in the field of robotics and there is
extensive documentation for the reasoning behind such 17.

2.1.1       
The Human Hand

The shape of the human hand was a result of evolution. It’s
development and form is a due to thousands of years of adherence to functional
requirement. The hand distinguishes the human from most animals and grants
humans the ability to create in and manipulate the environment around them. As
mentioned earlier, the combination of the hand and brain is considered what
allowed humans to advance so dramatically in comparison with other species 7.

The anatomical hand is connected to the
wrist through the palm. It has 27 major bones (eight carpals, five metacarpals
and fourteen phalanges), it has 18 joint articulations with 26 degrees of
freedom (DoF). This is all driven by 40 muscles 18. Each of the fingers
consists of a proximal middle and distal phalanx. The thumb has a proximal and
distal phalanx. The finger joints all exhibit felxion and extionsion and the
proxiamal phalange joints exhibit abduction and adduction, as well. The thumb
joints exhibit flexion/extension, abduction/adduction and rotation about the
axis of the meta carpal phalange 19. The muscoskeletal
structure is shown in Figure 1. Not al lof these features will be exhibited in
the final product but as the hand develops cetrain features and their choice
justified. There are a number of papers that focus on anlysis of range of
motion, biomechanics and tendon kinematics in the human hand 20 21 22.

 

2.1.2       
The Uncanny
Valley

With the increasing incorporation of anthropomorphism in
robotics there has been a reflective amount of research on their effects. One
of the main areas of concern is on the effect of robotics ‘looking human’ and
whether they should do so. This research has led to a descriptive
characteristic of certain robotics falling into the category of the ‘Uncanny
Valley’.

The proposal of this ‘Valley’ by A. Mori “hypothesized that a person’s response to a humanlike robot would
abruptly shift from empathy to revulsion as it approached, but failed to
attain, a lifelike appearance. This descent into eeriness is known as the
uncanny valley.” 23 24. A Mori
described valley as the place where the efforts to appear human by a robot
begin to produce the opposite of the desired effect.

In his work, A. Mori discusses the effect of affinity in
humans to robotics. Explaining how an industrial robot, with no attempt to look
human, doesn’t evoke affinity whereas the skin and flesh covered prosthetic arm
does so as its appearance is required to fit seamlessly with humans. The
argument begins to reflect A. Bicchi’s statement that “whether artificial hands should look like those of humans, is not
quite settled, and answers depend much on what exactly is expected from the
hand.” 25.

A. Mori in his paper introduces a graphical representation of
the theory. This is seen in FIGURE X. This
features discussion within the paper as the effects of movement on the
appearance.

 

2.2  End – Effector Morphology

In recent years, interest in robotic hands has been
increasing as their value in various fields, such as pick and place actions,
upper limb prosthesis and automation, has become more apparent. An overview of
robotic hands and their evolution are illustrated in

Development of hands is justified by the end purpose. Purpose
built industrial and automation grippers lead to a closed set of abilities for
the robot – such as welding car parts or microscopic surgery described in 26
and . This leaves the robot insufficient when required to deal with
unexpected tasks.

Inconsistent environments are mostly experienced in
situations where humans have options to manipulate it – such as at home, in a
shop etc. This requires a gripper to have an ability to adapt to the existing
human environment. Since the environment is already streamlined to be easily
manipulated by the human hand it can be said that end effectors with high
levels of anthropomorphism are desirable.

An index to measure robotic hand capability is referred to as
‘dexterity’ 27; A robotic hand
which is dextrous retains certain characteristics that are deemed desirable. There
are a number of definitions of what dexterity is:

–         
 “(The) capability of changing the position and
orientation of the manipulated object from a given reference configuration to a
different one, arbitrarily chosen within the hand workspace.” 25

–         
“(When)
multiple manipulators, or fingers, cooperate to grasp and manipulate objects.” 28

–         
“(The)
process of manipulating an object from one grasp configuration to another.” 29

–         
“(The)
kinematic extent over which a manipulator can reach all orientation.”  30

–         
“Skill
in use of hands.” 31

2.2.1       
Anthropomorphic

Anthropomorphism
in end effectors are normally included in the aims of achieving certain design characteristics
that are expertly executed in the human hand. Considering these characteristics,
L. Biagiotti suggests, as a method of defining the level of
anthropomorphism in a hand, the Anthropomorphism index 32. The three main characteristics
are displayed in Figure X.

The
aspects which define the anthropomorphic level in a robotic hand are: (a) presence
of an opposing thumb and the main morphologies associated as well as the number
of phalanges in the palm. (b) The way contact is made with objects on the
surface are of the hand. (c) The size of the end effector in comparison to the
human hand including correct size ratios between joints 32.

2.2.2       
Non-anthropomorphic

Non-anthropomorphic end effectors feature devices that tend
to be purpose built and do not resemble the characteristics outlined in 32. Grippers like the
CartMAN that was produced to compete, and subsequently won, the Amazon Picking
challenge 33. Featuring a suction
and pincer gripper, it shows how anthropomorphic grippers are not always
required when dealing with human environments.

There are many more types of non- anthropomorphic grippers
other than application built, such as soft grippers explained in hybrid
grippers and many more. These options can supply advantages such as compliance,
low
cost, simple controls enabled by nonlinear mechanical properties of materials,
light weight, low loading of weight-bearing surfaces, and low centre of gravity