2 Definition Of A Wearable Computer - The Direct Data

orangesvetElectronics - Devices

Nov 8, 2013 (3 years and 10 months ago)

130 views

Wearable Computer


TheDir
ectData.com

Page
1




1
I
ntroduction




A wearable computer is a computing device small and light enough to be worn
on one's body without causing discomfort. Unlike a laptop or a palmtop , wearable
computer is constantly turned on and interacts with the
a real
-
world task. Information
could be even very context sensitive.


Wearable computing facilitates a new form of human
--
computer interaction
comprising a small body
--
worn computer (e.g. user
--
programmable device) that is
always on and always r
eady and accessible. In this regard, the new computational
framework differs from that of hand held devices, laptop computers and personal
digital assistants (PDAs). The ``always ready'' capability leads to a new form of
synergy between human and computer,

characterized by long
-
term adaptation through
constancy of user
--
interface.


2
Definition Of A Wearable Computer




A wearable computer is a computer that is subsumed into the personal space
of the user, controlled by the user, and

has both operational and interactional
constancy, i.e. is always on and always accessible.



Generally speaking, the most salient aspect of computers (whether wearable or
not) are its reconfigurability and its generality, e.g. that their functio
n can be made to
vary widely, depending on the instructions provided for a program execution. With
the wearable computing there is no exception, e.g. the wearable computer is more
than an ordinary wristwatch or regular eyeglasses: it has the full functiona
lity of a
computer system but in addition to being a fully featured computer, it is also
inextricably intertwined with the wearer. That is why the wearable computer is apart
from other wearable devices such as wristwatches, regular eyeglasses, wearable
rad
ios, etc... Unlike these other wearable devices that are not reconfigurable, the
wearable computer is as reconfigurable as the familiar desktop or mainframe
computer. Reconfigurable means that adding new hardware or new functionality is
possible.


Wearable Computer


TheDir
ectData.com

Page
2









3

Interaction Between The User And The Computer




There are three operational modes of interaction between human and
computer:



Constancy



Augmentation



Mediation




3.1 Constancy




The computer runs continuou
sly, and is “always ready” to interact with the
user. Unlike a hand
-
held device, laptop computer, or PDA, it does not need to be
opened up and turned on prior to use. The signal flow from human to computer, and

computer to human, runs continuously to prov
ide a constant user
-
interface.









Fig 1:Constancy, a way to interact with the user.


3.2 Augmentation



Huma
n

Compute
r

Wearable Computer


TheDir
ectData.com

Page
3



Traditional computing paradigms are based on the notion that computing is
the primary

task. Wearable computing, however, is based on the notion that
computing is not the primary task. The assumption of wearable computing is that the
user will be doing something else at the same time as doing the computing. Thus the
computer should serve to

augment the intellect, or augment the senses. This is this
mode of interaction that our research is focused on.





Fig 2:Augmentation, another way to interact with the user.




3.3 Mediation




Unlike hand
held devices, laptop computers, and PDAs, the wearable
computer can encapsulate us.







Fig 3: Mediation, the third way to interact with the user.


It does not necessarily need to completely enclose us, but the concept allows for a
greater degree of encapsulation than traditional portable computers.

Thanks to encapsulation, it can function as an information filter, and allow us to block
out material we might not wish to experience, whether it is offensive advertising, or
Huma
n

Comput
er

Input

Output

Computer

Input

Output

Human

Wearable Computer


TheDir
ectData.com

Page
4


simply a des
ire to replace existing media with different media. In less severe
manifestations, it may simply allow us to alter our perception of reality in a very mild
sort of way.

It can also allow us to block or modify information leaving our encapsulated space.
Th
us wearable computing can be used to create a new level of personal privacy
because it can be made much more personal.

Because of its ability to encapsulate us, e.g. in embodiments of wearable computing
that are actually articles of clothing in direct cont
act with our flesh, it may also be
able to make measurements of various physiological quantities.














4
Attributes of the wearable computer



Wearable computer is defined by six main characteristics. These attributes
of wearable co
mputing are described from the human's point of view:



Unmonopolizing

(
of the user's attention)



Unrestrictive

(to the user)



Observable
(by the user)



Controllable

(by the user)



Attentive
(to the environment)



Communicative

(to others)








4.1 Unmonopolizing



Wearable computer does not cut you off from the outside world such as a
virtual reality game or the like. You can attend to other things while using the system.
It is built on with the assumption that computing will b
e a secondary activity, rather
than a primary focus of attention. In fact it will ideally provide enhanced sensory
Wearable Computer


TheDir
ectData.com

Page
5


capabilities. It may, however, manage (increase, alter, or deliberately decrease) the
sensory capabilities.



4.2 Unrestrictive




When using it, you can do other things, e.g. you can type in data using voice
recognition while doing some other stuff.



4.3 Observable



It can keep your attention continuously if you want it to. Wearable computer
is always observable wi
thin reasonable limitations (e.g. that you might not see the
screen while you blink). Consequently, the output medium is constantly perceptible
by the wearer.



4.4 Controllable



It is a responsive system. You can grab control of it
at any time you want.
Even in automated processes you can manually override to break open the control
loop and become part of the loop at any time you want to (for example a big Halt
button you want as an application becomes to calculate the result of huge

matrix
operations without leaving any CPU power to the other tasks would make a computer
more controllable). The constancy of user
-
interface results from observability and
controllability in the sense that there is always a potential for manual override w
hich
need not be always exercised.



4.5 Attentive



Wearable computer is environmentally aware, multimodal, and
multisensory (As a result, this ultimately gives the user increased situational
awareness). Each time it is necessary, the we
arable computer automatically adapts
these functionalities to the environment to be unmonopolizing. However, because of
its controllability, the user can ideally stop or modify the processing of adaptation to
the environment.



4.6 Communicative




It can be used as a communications medium when you want it to.
Wearable computer allows the wearer to be expressive through the medium, whether
as a direct communications medium to others, or as means of assisting the production
of expressive medi
a (artistic or otherwise).




Wearable Computer


TheDir
ectData.com

Page
6













5
The components of the wearable computer.



5.1 Human Interface System




The user interface for a wearable computer is fundamentally different to
those of the regular computers. T
he ideal human
-
computer interface for use in a
wearable environment would be one which listens for its user, understands what the
user has asked it to do as a combination of speech recognition, gestures and a bit of
machine vision. The results should be pr
esented back to the user in an intelligent
manner, when it is most appropriate and in a suitable format.


Consider an example; a quality inspector looking at car bodies going on the
assembly line. He may ask his wearable computer, “when I poin
t that the car on my
front has a fault, scan its serial number and record the error”, while pointing the
location of the serial number. This type of interaction with a wearable computer,
using spoken sentences and gestures, fall under the category of multi
-
modal and
intelligent user interfaces.


5.2 Sulawesi Architecture




A framework called Sulawesi has been designed and implemented to tackle
what has been considered to be important challenges in a wearable user interface. This
f
ramework gives the wearable computer an ability to accept input from any number of
modalities, and perform if necessary a translation to any number of modal outputs.
This system that has been designed comprises of three distinct parts.




1.

Multimodal
-
multimedia based Input system
: It gathers raw data from
the various sensors


The system gathers real world information through a well
-
defined API.
The current implementation includes keyboard input, netw
ork input, speech
recognition input, video camera input, G.P.S. input and infrared input. This stage
Wearable Computer


TheDir
ectData.com

Page
7


helps in connecting devices on the fly, and provides a device independent abstract
layer. Any necessary pre
-
processing of the data is done in the next stag
e.



2.
Agent based core system:
It contains a natural language processing
module and service agents




The core of the system contains a basic natural language processor,
which performs se
ntence translations. This converts a sentence into a command
stream from which two pieces of information are extracted, which service to invoke
and how the output should be rendered. A service manager is responsible for the
instantiationand monitoring of t
he services. The service manager also checks and
queues commands to provide resilience against system failures.



3.

Proactive and Wearable Output system
: It decides when and how to
render the results from the serviceagents





The output stage takes a modal neutral result from a service and makes
a decision on how to render the information. The decision is made based on two
criteria, what the user has asked for, and how the system perceives the u
sers current
context/environment.


If the user has asked to be shown a piece of information, this implies a
visual rendition. If the system detects that the user is moving or busy with an activity
(through the input sensors), an assumption

can be made that theuser attention might
be distracted if results are displayed in front of him (Imagine what would happen if
the user was driving)! In this case the system will override the users request and
would redirect the results to a more suitable
renderer, such as speech.


A successful wearable user interface must combine different types of input
and output, depending on the user's context and needs.



5.3 Operating System & Applications





The operating system and the applications are specifically designed
bottom
-
up for a wearable computer, to address the humionics. These should satisfy
the below criteria,

o

Shall be constantly available to the user


always on, ready and accessible

o

Shall

not require the constant user attention or interaction

o

Shall serve to augment user’s intellect and senses

o

Shall be unobtrusive and unrestrictive to the user. The user shall be able to
walk around, ride in a crowded bus, or even hang glide while using it
. This
aspect is also true for the hardware components of the wearable computer

o

Always communicate with user within reasonable time limits

o

Shall be able to communicate to other systems & external world

o

Provide the best use of the 3D object space to scat
ter the application
windows, a big shift from the regular 2D monitors. It is important to
Wearable Computer


TheDir
ectData.com

Page
8


understand that the user has a much bigger and deeper view for work area
in wearable computers, which needs to be used


5.4


Augmented Reality





Wearable computing introduces new concepts ‘mediated reality’ and
‘augmented reality’, which are very interesting to know about.




Mediated reality refers to encapsulation of the user's senses by
incorporating
the computer with the user's perceptive mechanisms, and is used to
process the outside stimuli. For example, one can mediate their vision by applying a
computer
-
controlled camera to enhance it. The primary activity of mediated reality is
direct interaction

with the computer, which means that computer is "in charge" of
processing and presenting the reality to the user.




Augmented Reality combines real world scenes and virtual scenes,
augmenting the real world with additional inf
ormation. The computer must be able to
operate in the background, providing enough resources to enhance but not replace the
user's primary experience of reality. This can be achieved by using tracked see
-
through display units and earphones to overlay visua
l and audio material on real
objects. This technology adds value to the human knowledge, memory & intelligence.




An example of an AR application is a guidebook as above. As the tourist
walks around the library, his wearable comput
er uses sensors, for example a
combination of GPS and head tracking equipment, to detect his physical position and
orientation. Some text describing the library is shown on the display unit over the
actual building. The wearable computer assists further in

enhancing the value of the
real world experience, using augmented reality.


5.5


Display Systems




The output device of a wearable computer could be either a head
-
mounted
display (HMD) unit with an earpiece or only the earpiece fo
r some applications.
Though there could be several other display devices intended for specific applications,
HMD systems are of interest in the conversation of wearable computers.




There are two different types of HMD systems. The f
irst one, intended for
industrial or regular use will have a see
-
through lens and a small projection system.
Only on need basis, the processing system may project the output data onto the lens.
The projection usually happens only on one of the lenses and t
he other lens remains
free for clear vision.




The second type of head
-
mounted display is of blocking type and requires the
full
-
attention of the user. This is mostly for 3D modeling, used for understanding
complex mechanical design s
ystems or for personal entertainment requirements. The
Wearable Computer


TheDir
ectData.com

Page
9


HMD systems shown in these pictures have both the earpiece and the mouthpiece
built into them. The HMD systems are already well deep into the development cycle
as of today, and do support several attra
ctive features like wireless connectivity,
external connectors for audio & video, and control settings.




The attributes of Head
-
Mounted Displays are:

1.Smaller,lighter

2Focaldepth,eyerelief,convergence/accommodation

3.Cost

4.Ruggedized,temperatu
reextremes

5.Lowpower

6. Position in the field of view including moving out of the way and disabling the
display

7.Morecomfortable

8.Colordisplay

9.Wireless connection to the wearable computer

10. Resolution greater than VGA


5.6


Input Devices





By now, you would probably be having a fair understanding of the criteria
of selection for input devices for WCs. But there is no holy grail for an input device
of a wearable computer.




Speech r
ecognition may appear as the most suited input device, but may not
be preferred in all kinds of applications & environments, due to privacy and
performance issues.




Handwriting & Keyboard could be one of the most efficient input

devices,
provided the input device is not too small or awkward. Research in this wearable
domain is resulting in combination products like the SenseBoard shown in picture.
This device is just worn on the hands or wrists and senses the typing input or
hand
writing. This does not have any cables and communicates on infrared.




Getsure Input devices are simple, compact, and optimized for wearable use.
These devices receive inputs from the natural gestures. Ubi
-
Finger is such a device
,
but has not been tested for all types of applications.




But the point to be taken is that the user needs to be open
-
minded and
adaptable to the emerging input devices, in order to find the best combination.





Thumb Typing
-

Carsten Mehring, a mechanical engineer at the University
of California, Irvine, has come up with a device that turns your hands into a qwerty
-
style keyboard. Mehring’s device uses six conductive contacts on each thumb

three
on
the front and three on the back

to represent a keyboard’s three lettered rows.
Wearable Computer


TheDir
ectData.com

Page
10


Contacts on the tips of the remaining eight fingers represent its columns. Touching the
right index finger to the middle contact on the front of the right thumb, for instance,
g
enerates a j. The top contact on the thumb yields a u, while the middle contact on the
back of the thumb would produce an h. Mehring says the similarity to typing makes
his input device easier to master than others that require an entirely different set of

motions. He has applied for a patent and hopes to market a product by year
-
end of
2002.


5.7


Networks




We need to discuss two different kinds of networks in reference to a
wearable computer. One is to connect the device to the ex
ternal world and the other is
to interconnect the various components, the later one being new for wearable
computers.




The first issue of connecting to the WC to the external world has
several choices; WAP, or Cell
ular Digital packet data. This aspect of networking is
not specific for a wearable computer, and can evolve over time, from other electronic
gadgets.




The second issue of interconnecting the various parts of

the WC, may
involve both wired and wireless connections. CPU, storage unit and similar
peripherals will be connected with or without cables to the wearable motherboard,
which is a garment with (physically) flexible bus and standard expansion slots.
Periph
erals like HMD and wrist/finger worn devices may use standard wireless
connections like Bluetooth.




There could also be a third type of communication, two wearable
computers talking to each other. This near f
ield networking could be on infrared
(IrDA or IRX) or radio based systems, to solve a need, which will invariably arise to
exchange information between two users.



5.8 Power sources




Batteries add size, wei
ght, and inconvenience to wearable computers.
There are several ways of harnessing the energy expended during the user's everyday
actions to generate power for one’s computer, thus eliminating the impediment of
batteries. However, there is no stopping to u
se to any of the miniature batteries, for
example Lithium, Li
-
MnO2, Li
-
C, that are currently being used in electronic gadgets.



5.9 Body Bus




weak electric fields can be used to create and sense tiny nano
-
amp
current
s in the user’s body. Modulating these signals creates Body Net, a personal
-
area network that communicates through the skin. Using roughly the same voltage
Wearable Computer


TheDir
ectData.com

Page
11


and frequencies as audio transmissions, this will be as safe as wearing a pair of
headphones.





Keeping data in the human body avoids the intrusion of wires, the need
for an optical path for infrared, and conventional problems such as regulation and
eavesdropping.




your shoe computer ca
n talk to a wrist display and keyboard and heads
up glasses. Activating your body means that everything you touch is potentially
digital. A handshake becomes an exchange of digital business cards, a friendly arm on
the shoulder provides helpful data, touch
ing a doorknob verifies your identity, and
picking up a phone downloads your numbers and voice signature for faithful speech
recognition.




Every emerging discipline in computing is born first as a theoretical
approach, al
most a dream in the mind of the inventor. Some of such theories have
clear practical ramifications, such as encryption algorithms. Others, on the other hand,
take some time to evolve from the concepts on paper to something that can be applied
in the real w
orld. This could be one such example.


5.10 Constraints of the wearable computer



There are many constraints with the wearable computing. Indeed, the
most important ones are the resource limitations. Due to size and weight limitat
ions,
and subsequent restrictions on battery consumption and heat emission, a wearable
computer tends to have slower CPU, a smaller space of storage (e.g. an hard disk like
others computers) and smaller main memory than today’s Personal Computers. In
addit
ion, network bandwidth and availability are similarly reduced due to the
characteristics of wireless network technology. All of these technologies and the wide
range of applications targeted result in various and specific hardware that meets these
restrict
ions. Ideally, devices (and by extension, wearable computer) are integrated in
the person’s wear. So, wearable computer has no redundancy devices.



All of these hardware constraints result in strong restrictions for the
software d
evelopment. When the computer is connected to the environment, it could
even store data on remote servers. But, in fact, these connections can break, especially
with wireless links. This kind of links are more sensitive to interferences than cables.
Design
ing should pay attention to these problems.


5.11

Objectives



Since Wcomp research team has been working on the wearable
computer, we establish a listing of the abilities and constraints of it. So, we can now
highlight objectives of my

work and the way we chose to apply them.


Wearable Computer


TheDir
ectData.com

Page
12




Resource management




First of all, due to the resource limitations and the kind of software
(like augmented
-
reality and voice recognition), the efficient resourc
e management
becomes vital for the success of the wearable computer applications.




Connectivity




Improve connectivity is a key for several applications that are very
communicative. They are in an environment and have to
be attentive to this
environment. So, one aspect of this attention consists of looking for connections with
others.




Software solution




Even if there are heterogeneous platforms and a multiplicity of
hardware, we want to
propose a solution, “a scheme” that can be reused. Indeed,
several computer domains have one and, in general, more ways to design an
application.


5.12

Privacy & Health Issues




Though wearable computing does not raise a
ny new privacy issues, it
is true that the most useful information is also most personal. Just because the
wearable computers can be used for surveillance purposes, does not mean that they
are being used.




Wearable co
mputing does not involve any privacy issues which
otherwise cannot be done. This just helps getting around with what you mostly want,
what you mostly do, what you mostly like to do anywhere and everywhere.




Health and

Safety considerations will be important when one is wearing
these things all his waking hours (and arguably sleeping hours too!). Remember the
Carpal Tunnel Syndrome? Last but not the least, the resulting outfits shall be
fashionable and provide the buyer

with a choice of fashions. After all, one doesn’t
want to go to a formal dinner looking like C3PO.

Wearable Computer


TheDir
ectData.com

Page
13




































Wearable Computer


TheDir
ectData.com

Page
14




6
System Architectures




Three different system architectural approaches were id
entified. Each
approach had both advantages and disadvantages.




The first approach building upon desktop computer technology.
Commercial off
-
the
-
shelf (COTS) hardware (e.g., Intel processors, etc.), operating
systems (DO
S, Windows, etc.), and application software could be used. As , the
desktop metaphor is poorly suited to wearable computer applications. By adopting
mainstream platforms from the desktop there is extra overhead in processing power,
storage, and power consu
mption for the wearable computer application. However,
COTS software is portable, reusable, and minimizes application development time.




The second architectural approach builds upon real
-
time embedded system
technology.
Low energy processors, real
-
time operating systems, and custom software
could produce lighter weight, longer battery life, and more responsive applications.

However, real
-
time operating systems have a limited number of software drivers and
the system deve
lopers may have to write software drivers for input/output devices that
are not normally encountered in embedded systems (e.g., global position sensing,
speech recognition, etc.). The development time for application software may be
longer than for desktop

systems since there are a limited number of software
development tools for embedded systems. However, this cost can be offset if
dedicated application metaphors are developed that have standards for
intercommunication and information exchange.





The third approach could be built upon the emerging web browser
technology. Future dedicated web browser hardware and operating systems could be
augmented by web "plug in" applications allowing the wearable computer user to
become a
network node.



The discussion in the breakout session then turned to wearable computer
subsystems. In particular, head
-
mounted displays and energy sources were
considered.









Wearable Computer


TheDir
ectData.com

Page
15




6.1 Behavioral Software Architecture Model






We propose an architecture to solve wearable computing problems. We
name this architecture the behavioral architecture. The behavioral architecture style
inspired from robotics. This architecture deals with the wearable computer’s
constraints like the adaptation to the environment or the resources management.




Presentation



Computer systems interact with the environment in the following way :





Fig 4:Computers interaction with their environment



Nowadays, component
-
based programming is used by many creators of
applications. They assemble functiona
l components to build their applications. A
functional component is a component that achieves a specific functionality. Our
approach consists of describing application not with an assembly of functional
components but with behavioral components. A behavior
al component describes a
behavior of the final application. It is independent of the rest of the application, i.e. it
deals with inputs to produce outputs without interacting with other behavioral
components.



To explain and show dif
ferences between the two approaches, we will
build an application using the two methods. For this example, we propose an
application that, starting from point A, goes to point B and avoids obstacles that can
be present or appear on the path.




On one hand, we explain one way to create this application using
functional components.



First, we split the application into four functionalities :



Obstacle detection



Positioning

Inputs

Applicati
on

Outputs

Wearable Computer


TheDir
ectData.com

Page
16




Decision making



Moving




So, we obtain the

following schema to describe interactions between functional

components:









Fig 5:Functional composition of an application


.On the other hand, we divide the application into behavioral components. We obtain
this
two behaviors :



Avoiding obstacles



Going from a position to another


The application is now built with the schema :





Fig 6:Behavioral composition of an application.





Overall Model


INPUTS

OUTPUTS

Obstacle

detection

Positioning

Decision

Making

Moving

INPUTS

OUTPUTS

Avoid
ing

Going

Wearable Computer


TheDir
ectData.com

Page
17



At the ve
ry beginning of behavioral architecture is the concept of a set
or collection of objects, in particular, sets of data on which we base our quantitative
work.







Defination:


Let
I

be the finite set of physical input d
ata. We can write that




1
,
0
,



I
k
n
k
i
I

where
k
i

is an input piece of data and
I
n

the size of
I

(i.e. the
number of physical input data).


Defination:



Let
O

be the finite set of physical output data. This is the same definition as
below but for output data




1
,
0
,



O
k
n
k
o
O

where
k
o

is an output piece of data
and
O
n

the size of
O
.



Defi
nation:


Now, we can define the environment (the entire set of input/output data)
E

by :

O
I
E




by using this set of data, in the future, we will separate systems in different class (for
example


O
I


i.e. some data can be inputs and outputs).




Defination:


We consider the set of behavioral components




1
,
0
,



n
j
c
C
j

where
j
c

is
a component and
n

is the number of components loaded int
o the environment.



Defination:


We define the mapping function
I
j
m

:








1
,
0
1
,
0
:
,
1
,
0





I
I
j
I
j
n
n
m
n
j


so :







)
(
,
1
,
0
,
1
,
0
,
1
,
0
l
m
k
I
I
j
I
j
i
i
n
k
n
l
n
j












by using this mapping function, we can retrieve a physical input data from a given
virtual input and an
index.


5
)
0
(
9

I
m

and
18
)
1
(
9

I
m
.

Wearable Computer


TheDir
ectData.com

Page
18


This example shows how to retrieve input data of behavioral component
9
c
, i.e. input
data 0 of
9
c

is
5
i

and input data 1 is
18
i
.

In the same manner, we define the outputs.


Defination:


We define the mapping function
O
j
m

:








1
,
0
1
,
0
:
,
1
,
0





O
O
j
O
j
n
n
m
n
j


so :







)
(
,
1
,
0
,
1
,
0
,
1
,
0
l
m
k
O
O
j
O
j
o
o
n
k
n
l
n
j












by using this mapping function, we can retrieve a physical output d
ata from a given
virtual output .



We can now define the following behavioral architecture :



I

0
i

1
i

1

I
n
i

k
i

Set of physical
input data

O

0
o

1
o

1

O
n
o

k
o

Set of phy
sical
output data

C

0
c

1
c

j
c

1

n
c

Set of

behavioral components

I
m
0

I
m
1

I
j
m

I
n
m
1


O
m
0

O
m
1

O
j
m

O
n
m
1


Wearable Computer


TheDir
ectData.com

Page
19





Fig 7: The behavioral architecture



Application


Let’s use the model to represent the examp
le described in presentation.


Input data of the system are the following :



0
i

: longitude of the position



1
i

: latitude of the position



2
i

: free distance in front of the system



The output
data of the program are :



0
o

: instruction for left wheel motor



1
o

: instruction for right wheel motor


The two behavioral components are those listed in introduction :



0
c

: Going



1
c

: Avoiding


In order to get equivalence between the input data of the components and those of the
system, here are the two input mapping functions :



I
m
0

:




1
,
0
1
,
0


and
2
0

I
n

(i.e. behavioral c
omponent
0
c

has two inputs)



I
m
1

:




0
2

,
1
1

I
n


Mapping functions exist also for the output data :



O
m
0

:




1
,
0
1
,
0

,
2
0

O
n



O
m
1

:




1
,
0
1
,
0

,
2
1

O
n



So, we design the behavioral architecture that corresponds to this example :







Wearable Computer


TheDir
ectData.com

Page
20



Fig 8:Example of a behavioral architecture.


By using diagram, we
point out a problem with behavioral architecture. The next
paragraph explains our approach to solve them.




Conflicts


When a component is put into the environment, he has to access input and output
data. As a consequence, when a behavioral component needs
to read/write data, if
another behavioral component wants to do the same action, conflicts may appear.




Detection of conflicts


If the application is formalized, detection of conflicts is made by :


Defination:




For the inputs,


1
,
0
'
,



n
j
j

/


1
,
0



I
j
n
l

and


1
,
0
'
'



I
j
n
l

and




1
,
0



I
n
k

and
k
l
m
l
m
I
j
I
j


)
'
(
)
(
'
.


Defination:



In the same way, here is the formulae for the outputs,


1
,
0
'
,



n
j
j

/




1
,
0



O
j
n
l

and


1
,
0
'
'



O
j
n
l

and


1
,
0



O
n
k

and
k
l
m
l
m
O
j
O
j


)
'
(
)
(
'
.


Translation in plain text would be : there are at least two input data (
)
(
l
m
I
j
i

and
)
'
(
'
l
m
I
j
i
)
of two distinct components (
j
c

and
'
j
c
) that correspond to the same input data (
k
i
) of
the application.


O

0
o

1
o

C

0
c

1
c

I

0
i

1
i

2
i

I
m
0

I
m
1

O
m
0

O
m
1

Wearable Computer


TheDir
ectData.com

Page
21




Input data


Even if there are conflicts in input, this is not a problem because the value of the input
data will not be alte
r by multiple reading at the same time. The operating system of
the system has to deal with this.




Output data


The output data conflicts are real problems. In fact, an output data takes only one
value at a time. So, to solve conflicts between the behavio
ral components, we
introduce a combination mechanism. It produces the value for each output data
conflict.




Combination mechanism


Defination:


Combination mechanism works on a part of the output data. Indeed, the conflicts
are only a part of the set o
f outputs. If we define


as the set of conflicts, we can say
that
O


.





1
,
0
,




O
k
n
k
o

with


the number of conflicts.



Defination:


Then, we define the combination mechanism as a se
t of functions :





k
O
f
n
k
F
,
1
,
0






Where
k
f

is the identity function if


k
o

and a combination function when


k
o
.



Combination function


Defination:



For each output, we define the set of

values of the output data. For example, for a
monochrome paint, this set is


black
white
,
, i.e. a list of the possible values.

k
k
k
k
o
valueOf
O
o






)
(
,
,


Now, a combination function
k
f

works only for an output data
k
o

and takes only
values in
k

. As we can see, k is always the same number. So, we will write
f
,
o

and


instead of
k
f
,
k
o

and
k

.


We obtain the following set of values for a given output :





l
n
l

,
1
,
0








Wearable Computer


TheDir
ectData.com

Page
22


A combination function
f

is defined as following :




p
f
:

where
p

= Number

of behavioral components that use this output


Consequently, a combination function uses the following pattern :

)
,
'
,
(



f

with the number of parameters equals to the number of conflicts.




Application


Let’s keep the example used in 3. Both

behavioral components
0
c

and
1
c

have the
same output data
0
o

and
1
o
. We define the set of output
0


:


1
,
0
,
1


(0 means that
the motor
will not turn,
-
1 moves backward and 1 moves forward). Consequently, the
set of output
1


is


1
,
0
,
1

.


We consider that
0
c

avoids obstacles (it works to clear the pathway) by turning on the
left. An
d
1
c

goes to the destination by sending the adequate instructions to
0
o

and
1
o
.


In the two following cases, we use the same algorithm for the two outputs. It is an
average, this method works i
nternally with float but needs integers and outputs
integers. In order to round the floats (when necessary), it emits 1 if the average is over
0 and

1 if the average is less than 0.









Simple case



Fig 9:Spatial representation wit
hout an obstacle.


If no obstacle is present in the front of the system and the destination is straight
forward,
0
c

outputs 0 for
0
o

and 0 for
1
o

because it has not to avoid obstacle.
1
c

outputs 1 for
0
o

and 1 for
1
o

because it has to go to the destination.

Following these two combination functions
0
f

and
1
f

produces 1 :





1
1
,
0
0

f





1
1
,
0
1

f


We draw an arrow to show the direction of the system resulting of these output data :


Syste
m

Destinati
on

front

back

Wearable Computer


TheDir
ectData.com

Page
23




Fig 10:Movement of the system without an obstacle.




Simple case with obstacle


We now add an obstacle on the pathway.



Fig 11:Spatial representation with one obstacle on the pathway.


0
c

avoids it by turning on the left to his output data are

1 for
0
o

and 1 for
1
o
.

But,
1
c

goes over this obstacle, it outputs 1 for
0
o

and 1 for
1
o
.


In this case, the “avoiding” behavioral component alters the “going” component and
the outputs of the system are 0 for
0
o

and

1 for
1
o

:





0
1
,
1
0


f





1
1
,
1
1

f


The direction is now the following :



Fig 12:Movement of the system with one obstacle.



6.2 Coordination Mechanisms state of the art




In

the behavioral architecture resources of the wearable computer are shared out
by a part or all of the behavioral components. Thus, combine the output data of each
behavioral component in order to obtain a rational and coherent global behavior is
necessary
. I based this state of the art several references but the most important one is
the behavior coordination mechanisms from Paolo Pirjinian[4]. Indeed, behaviors
-
based robotics are used for years. Consequently, we think that behaviors are
particularly adapt
ed for the wearable computing. In addition to the resource
limitations and reactive constraints like the robotics, wearable computers have user
-
interactive problems.


Coordination mechanisms are shared in two families :

Syste
m

Destinati
on

front

back

Syste
m

Destinati
on

front

back

Syste
m

Destinati
on

front

ba
ck

Wearable Computer


TheDir
ectData.com

Page
24




Arbitration



Fusion


Arbitration algo
rithms give full access to a resource, i.e. one behavioral component
outputs it own data on the resource at a time.

In contrary, fusion mechanisms produce a data for a resource, i.e. the data on a merge
of all behavioral components that want this resource
as an output.




Arbitration




Subsumption architecture



In this architecture, layers of competence are defined, i.e. each behavioral component
is in layer of competence. A layer is an informal specification of a desired class of
behavior at different abstra
ction levels. Higher level layers can subsume the role of
lower layers by suppressing their outputs. However, lower levels continue to function
as higher levels are added. This organization (subsumption architecture) organizes
behaviors in a network of har
dwired finite state machines.


The main contribution of this architecture is to see the system not as a sequential
process. Indeed, the decomposition separates a system into behaviors, exactly what
we need to combine with behavioral architecture. Brooks co
nsidered that complex
(and useful) behaviors need not necessary to be the product of an extremely complex
system.


In fact, what will happen ? The outputs of the wearable computer will be shared in the
time to all the behavioral components according to the

layers of competence. This
system is similar to the computer’s CPU sharing. Indeed, in most of the operating
systems, a scheduler based on priorities shares CPU power in the time by giving time
to each task (according to its priority).







Fig 13:The subsumption architecture.


Level

1

Level

0

Level

2

Level

3

Inputs

Outputs

Wearable Computer


TheDir
ectData.com

Page
25


The system presented above can be partitioned at any level, and the layers below form
a complete combination mechanism. Layers of competence definition permits to
observe, inject and suppress the normal data flo
w, e.g. a task running at level one can
inject its own output data and suppresses the data produced by tasks at the level 0
competence.


The behavioral architecture does not accept communications between the behavioral
components. Thus, using the subsumpti
on architecture as the coordination mechanism
requires many constraints. The data flow of the behavioral components that are in
lower layer can not be observed. But, in addition to the output value, behavioral
components can produce an information (or a sp
ecial output value) that shows if it
requires the output resource, e.g. a behavioral component that avoids obstacles does
not always need access to the outputs (probably only when an obstacle is detected).
The coordination mechanism that knows all the beha
vioral components (and thus, the
layers) decides even if a component asks for the output if it will output his data.
Finally, for the wearable computer, the controllable of the layers will reflect the
controllable of the system. This aspect is indeed very
easily modified by changing
behavioral components from one level to another.


The main limitation of the subsumption architecture is the incremental construction of
the layers. Indeed, Brooks claimed that a new layer is added only when the previous
layer i
s completely achieved and functional. Even if the subsumption architecture is
quite old and restrictive, a full running example is shown in [5] and so, it is an
effective solution. Moreover, mechanisms based on this architecture would be
probably very usef
ul with the kind of problems similar to the robotics, e.g. moving
assistance for a blink.




Discrete Event Systems


Jana Košekcá and Ruzena Bajcsy present the “Discrete Event Systems based
approach” to run cooperative behaviors in [7]. They present modeling

analysis and
synthesis of visual behaviors for navigational tasks. By using Discrete Event Systems
theory, behavioral components could be combined.


In this approach, behavioral components selection is done using state
-
transition,
where upon detection of
a certain event shift is made to a new state and thus a new
behavioral component. Using this formalism, combination mechanisms are modeled
in terms of finite state automata, where states correspond to execution of behavioral
components and events which cor
respond to inputs modification, cause transitions
between the states.


This team develops a Task Description Language in [9]. Thus the tasks (i.e.
behavioral components) are specified as networks of process which are themselves
described with finite state
machine. Thus, in the behavioral architecture, the last
Wearable Computer


TheDir
ectData.com

Page
26


aspect is the most important because we have to combine the output data of the
behavioral components.


Let’s use the simple example described at paragraph
0
. The finite state automata is the
following :



Fig 14:Example of finite state automata for discrete event systems.


With this automata, we see a method to arbitrate the “owner” (temporary) of the
physical output. At each step of this

FSA, a behavioral outputs his value without
regarding if this value is opposite to another component’s value.


With this combination mechanism, the output values are easy to compute. The inputs
are necessary for this kind of cooperation. Indeed, even with

complex goals, we
define the most appropriate behavior for each situation the system has to deal. For the
wearable computer, this mechanism permits to be very attentive to the environment.
When the system detects changes, it adapts the behavioral componen
ts selection to
comply with the situation.


There are more algorithms that use finite state automata to arbitrate between
behavioral components during the time of execution. The “sequenced coordination”
error reference source not found developed by Arkin a
nd MacKenzie is another
solution. Indeed, they also use an automata but it is implanted using the hardware
definition language BHDL (for the triggers) and the motor schemas (for the states).
With behavioral architecture, we promote the Rapid Application De
velopment
Wcomp 3.0 to design behavioral components. So, the events of the finite state
automata only has to be implemented.








Bayesian Decision Analysis


Going to
the
destinati
on

Avoiding

Obstacle
s

Obstacle
detected

Clear path

Wearable Computer


TheDir
ectData.com

Page
27


Steen Kristensen proposes in error reference source not found Bayesian Decision
Analysis to decide

what sensing actions (behavioral components) should be
performed.


Bayesian Decision Analysis allows for reasoning about uncertainties, or rather, it
provides a framework for systematically dealing with uncertainties in decision
problems. This is because
in a Bayesian framework knowledge is not represented by,
e.g., production rules but rather as beliefs and conditional probabilities. An
introduction to this theory is presented in error reference source not found.


The fundamental idea behind BDA is to sel
ect the action that in the long run will
optimize the expected utility. The expected utility is the gain that we could wait for an
action (i.e. giving output to a specific behavioral component), e.g. S. Kristensen
shows the example of winning at the lotter
y 100$ with one ticket (probability 0.5)
and, another lottery 500$ (probability 0.01). With these parameters, we choose the
first case that presents the best expected utility. But the expected utility is a little more
complex process:






j
i
M
j
i
j
i
z
C
U
C
z
P
C
ility
ExpectedUt
,
)
(
4




with
i
C

the behavioral component
i
, the
probability


i
j
C
z
P

of the event
j
z

occurs after outputs data of
i
C

and


j
i
z
C
U
,

the
utility of
this combination.


After these calculations, choosing the right component results in the maximum of all
the expected utilities. In addition to the probabilities on the events (the state of the
sensors), there are probabilities on the information that produ
ces events, i.e. the
reliability of information given by the sensors.

Now to make the right choice, a tree is built with probabilities on the links and
utilities on the node. Starting from leafs to root, the best expected utility appears.


With the followi
ng example, we want to decide if the system has to launch the
avoiding/going behavioral component and if the control task is necessary due to the
reliability of sensors. Because we have no preference between avoiding/going
behaviors, the probabilities of t
he first level are 0.5. In the same way, the second level
presents 0.5 probabilities to know if the control is necessary. The last level presents
probabilities on reliability that depend on controlling or not. The utilities on the leafs
are tied to the suc
cess and the cost of each behavioral component. Thus, even if the
designer puts some rules to compute utility values, they evolved during the executing
time, e.g. the turn on utility of the avoiding component will increase at the proximity
of obstacles.


Wearable Computer


TheDir
ectData.com

Page
28




Fig 15:A Bayesian Decision’s tree.


For the behavioral architecture, Bayesian Decision Analysis is a good solution of
combination mechanism. Indeed with the probabilities, network failures, sensors
problems and uncertainties of the sen
sor’s values are simply introduced in the process
of combination.




Learning Algorithms


Another kind of mechanisms uses learning to arbitrate between the behavioral
components. Hierarchical Q
-
Learning in error reference source not found and W
-
learning in e
rror reference source not found . They are both based on the same
principle that is the learning during the execution.


To simplify, Hierarchical Q
-
Learning separates the task (global behavior) to achieve
into sub
-
tasks (behavioral components). In this alg
orithm, the learning process, i.e. the
memory, is placed into the switch (combination mechanism).


Another approach consists of letting the behaviors learning themselves error reference
source not found. Even if this aspect could appear profitable because
of an auto
-
organization of the behavioral components, it is not. Indeed our architecture does not
allow communications between components.


The idea of learning functions in the combination mechanism is very useful to
increase the attentive aspect of the w
earable computer. In fact, the system could
automatically choose the behavioral components preferred in some kind of
environment based on the historic of the user’s modifications. Learning can be
adapted for all the combination mechanisms that can be tuned
, e.g. modify some
probabilities of a Bayesian Decision process in order to comply with characteristics of
sensors between the day and the night. Contrarily, the finite state automata are
difficult to extend with learning functions by the presence of defin
ed events that could
occur.





Activation networks


90

-
10

0.
9

0.1

0

280

-
20

0.
9

0.1

10

180

-
120

0.
5

0.5

30

60

-
40

0.
1

0.9

50

25

20

22.5

0.
5

0.
5

0.
5

0.
5

0.
5

0.
5

Avoiding or
Going

?

Controlling or
not

?

Reliabilit
y

Wearable Computer


TheDir
ectData.com

Page
29


Pattie Maes presents the activation networks to solve action selection problem
(combination function for us) in error reference source not found. She splits the
problem to solve in competence modules (beha
vioral components) that separately
achieve a specific goal. To decide which module to execute an activation network is
used.


Each node on this network is a module (behavior) and is represented by a tuple :


i
i
i
i
d
a
c

,
,
;
.
i
c

is a list of preconditions which have to be fulfilled before that the
competence module can become active.
i
a

and
i
d

represent the expected effects of
the competence module's action in terms of an added list and a
deleted list. In
addition, each competence module has a level of activation
i

. A behavioral
component’s output is selected at time t when all of its preconditions are observed to
be true at time t. A behavioral component may be select
ed, which means that it
performs some real world actions.


Nodes representing behavioral components are linked in a network through three
types of links:



successor links



predecessor links



conflicter links


Predecessor and successor links are the same links
, if x is the predecessor of y so y
has x as successor. x is a predecessor of y when an action in
i
a

is also included in
j
c

of y. The conflicter links are between a node that requires an action to be activated
and
another node that removes the benefit of this action.

Thanks to these kinds of links, an internal activation (of the states) is performed by :



spreading activation through successor links



activating predecessors in order to be selected when they will succe
ed



decreasing the activation level of predecessors that are on conflicter
links


Another way to activate the node is the external activation. The sources of this
activation are :



states, the level of activation is decrease until at least one state matches
the required amount.



goals, the activation energy is injected by the goal and goes up via
predecessors links.



protected goals, the activation energy is decreased on the states that can
undo an achieved goal.


Using the activation networks to combine the o
utput of behavioral components is time
consuming. In fact, this mechanism is more complex than the algorithms based on
events. While activation networks have to compute great amount of data at each event
in order to decide which behavioral component is sel
ected. In contrary, the finite state
Wearable Computer


TheDir
ectData.com

Page
30


automata just makes transitions between states when events are triggered. Only a very
high frequency of event’s arrival will cause the mechanism to be time consuming.


The key of activation networks is the level of act
ivation. Effectively, the tuning of this
permits controlling the number of behavioral components that are selected to output
their value. To avoid conflicts, this parameter must activate only one output at a time
in the behavioral architecture. This mechan
ism extends the event based algorithm
because each state continues its evolution even if it is not actually selected. This kind
of algorithm is more flexible than those based on finite state automata.




DAMN


One of the most popular action selection mechani
sm (equivalent to the combination
mechanism for the behavioral architecture) is DAMN. DAMN is a Distributed
Architecture for Mobile Navigation. Julio Rosenblatt presents this solution to solve
Mobile Navigation in his thesis error reference source not foun
d.


The organization of the DAMN architecture is shown in the following figure-2; it
consists in a group of distributed reasoning modules (behavioral components)
communicating with a centralized command arbiter (combination mechanism),
sending votes in fav
or of actions that satisfy its objectives and against those actions
which do not. The arbiter is then responsible for combining the modules' votes and
generating actions which reflects their objectives and priorities. Within the framework
of DAMN, behavior
s must be defined to provide the task-specific knowledge for
modifying the wearable computer. Each behavior in the system is responsible for a
particular aspect of wearable computer or for achieving some particular task,
receiving as input only that sensor
y information pertinent to the task, and processing
that input to generate as output votes for and against possible actions. Voting takes
place in a commonly defined action or state space, thus providing a uniform interface
to the arbiter that is independe
nt of the task the behavior.



Fig 16:The DAMN architecture


Implementing the DAMN architecture as the combination for the behavioral
architecture requires some adaptations:



Votes are needed

Avoiding

Going

DAMN Arbitrer

votes

inputs

outpu
ts

Wearable Computer


TheDir
ectData.com

Page
31




Lower CPU usage is required


First of all, b
ehavioral components must be modified or designed to produce votes.
Effectively, instead they work on the outputs of the system, they now vote for their
preferred values and mark the value they deny. As a result, the DAMN arbiter can
select the appropriate
d value (that has won the elections) after eliminating denying
values.

Secondly, due to the limited CPU power, the implementation of DAMN should be the
most simple. Effectively, instead vote for all the possible values that can be held by
the outputs, the
behavioral component may only vote for a sample of significant
values (at the point of view) and deny some values. So, the combination mechanism
will work on a reduced set of values for the outputs and work faster than a full
implementation of DAMN.


As ot
her approaches, DAMN has been tested with success on several applications for
mobile navigation. Now, it can be a good solution for wearable software too. The use
of votes also permits reducing the set of values for outputs. It can be useful, e.g. when
the

batteries of the computer fall down under a critical level, it can decide to stop
using values that make high electrical consumption. Wearable computer will be easily
attentive to the environment and controllable by the user.




Fusion




Fuzzy command fusion


The fuzzy command fusion is already used in behaviors coordination in error
reference source not found. In this paper, Alessandro Saffiotti shows how to use the
fuzzy logic in order to coordinate behavioral components.

The two main advantages in using fu
zzy logic to combine behavioral components are:



the ab ility to express partial and concurrent activations of
behaviors



the smooth transitions between behaviors


Generally speaking, these advantages are typical of the fusion algorithms.




Fig 17:The Fuzzy/Defuzzification process


Validity

TRUE

FALSE

Power

Ma
x

0

Wearable Computer


TheDir
ectData.com

Page
32


The Fuzzy logic permits more flexible arbitration policies than a fixed arbitration, it
can be obtained using fuzzy meta-rules, or context rules, of the form IF context THEN
behavior, meaning that behavior shou
ld be activated with a strength given by the truth
value of context, a formula in fuzzy logic. In the figure above, the output power of the
behavior is linked to the belonging to the space of validity.


A perceptual condition can be used to decide between
two alternative behaviors. For
instance, the following rules can be used to navigate to a target while reactively
avoiding obstacles on the way:

IF (obstacle is close) THEN Avoiding

IF (obstacle is NOT close) THEN Going to destination


When the obstacle i
s only partially closed, both behaviors are both partially activated;
thus, the commands issued by the “Going to destination” behavior can be taken into
account during the obstacle avoidance moves. To obtain this result, each behavioral
component as a modu
le expressing preferences as to which command to apply;
degrees of preference are represented by a possibility distribution (or fuzzy set) over
the command space. Then, we use a fuzzy operator to combine the preferences of
different behaviors into a collec
tive preference. Finally, we chose a command from
this collective preference.


According to this view, command fusion is decomposed into two steps:



preference combination (defuzzification, combination in particular order)



decision


As we see in the followi
ng figure, the order of defuzzification and combination
process modifies the result of the fuzzy command fusion.



Fig 18:Combination and defuzzification process


0

10

5



10

6.
5

0

10

8

0

5


7

9

8


10

8

0

Defuzzificati
on

Defuzzifica
ti
on

Combination

Combination

0

10

7

9

Wearable Computer


TheDir
ectData.com

Page
33


With the fuzzy command fusion, behavioral components have to manipulate
structures more complex than simple physical output (like, e.g., vote based command
combination). But, thanks to this restriction, it enables smooth transitions between
behavioral components. In a wearable computer, this point must serve to be
unmonopolize
d of the user’s attention. In fact, a behavior will not make the user loose
concentration. Another problem is the CPU power necessary to compute this
combination. Developers have to be very attentive of the way they implement the
fuzzy set in order to prod
uce fast computing for the wearable computer.




Motor schemas



The last but not the least combination mechanism in this state of the art is the
motor schemas based command fusion. Many researchers have proposed coordination
schemes where severa
l behaviors can be concurrently activated, and their outputs are
fused by some form weighted combination. The most popular approaches of this type
are based on a vector summation scheme: each command is represented by a force
vector, and commands from diff
erent behaviors are combined by vector summation.
R. C. Arkin presents the motor schemas in error reference source not found and uses
this approach to implement robot navigation.



This algorithm is less complex than fuzzy command fusion because

it consists
uniquely of a weighted sum of the behavioral components output. It does not require
any specific protocol for the outputs. The following scheme shows a very simple
motor schemas combination.



fig 19:Weighted sum example.


In this algorithm, the weights used to compute the sum have not to be set once, they
can evolve during the execution time.



With wearable computer, the main advantage of this mechanism is the simple
and fast way to compute physical output valu
es. But, if this combination could appear
too simple in particular case, the weights can be tuned. Thanks to the dynamic
modification of this parameters, more complex global behavior can be obtained. In the
behavioral architecture, this algorithm is a solu
tion for most platforms that have very
limited resources for the CPU.


Behavioral
component
1

Behavioral
component
2

5

10

6

BC1_weight =
4

BC2_weight =
1

5*BC1_weight

+

10* BC2_weight

Wearable Computer


TheDir
ectData.com

Page
34



















7
Applications of Wearable Computer




Augmented Memory
:

The Remembrance AgentRememberance Agents
(RAs) are wearables that








continuously reminds the wearer of potentially relevent information based on
the wearer’s current physical and virtual context. R
As generally have features
available that are not present in current laptops or Personal Digital Assistants
(PDAs, such as Palm
-
top computer), one of them being that it is always on and
always active and working, instead of being 'woken up' when needed.

It is the
combination of the real and the virtual to assist the user in his/her
environment. There are lots of applications for this field: devices for the
disabled, architecture construction and telemedicine.




Finger Tracking

This is one of the simplest applications of camera
-
based
wearable computing. The computer would be able to visually track the user's
finger, such an interface can therefore replace conventional components like
the mouse with his/her own finger. The user wo
uld then be able to control the
operating system in this manner, or even digitise an image and virtually
annotate it.




Face Recognition


Working in conjunction with an appropriate face
-
finding
software, face recognition system can be adapted for use in we
arable
computing. Names would be overlaid on faces as the user moves about the
world in his/her wearable computer. Markets include the police, reporters,
politicians, the visually disabled (with an audio interface), and those with bad
memories for faces.


Wearable Computer


TheDir
ectData.com

Page
35




Visual filter


This is particularly useful for the visually disabled, for example
to map around 'blind spots'. The wearable computer can digitally magnify an
image or a prose through the use of a 'virtual fish eye lens' for help in reading.
This can be do
ne through the use of a digital visual filter. The basic concept is
to process video images digitally in real time to assist the user in everyday
tasks.




Navigation 8 Connecting

a
Glo
bal Positioning System (GPS)

to a
wearable and certain mapping software allows the user to track himself while
exploring a city. A visually impaired person might be able to receive warnings
of approaching objects and hence promote safety in their daily li
ves.




Repair instruction

By putting as little as 3 distinctive marks at known
distances from each other, a wearable camera with known focal length can
recover the 3
-
Dimensional (3
-
D) location of an object defined by these three
marks. With the help of an
on
-
line manual on the technician's wearable, one
can extrapolate and derive the rest of the object's 3D location. The repair
technician can then easily trans transmit the diagnostics of a broken machine
to the wearable. The wearable automatically determine
s the problem, locates
the 3
-
D position of the object, and overlays specific 3
-
D real
-
time step
-
by
-
step
guidelines on the object for the technician to follow.




Wearable Audio Computing (WAC)

So far we have seen wearables
derive their interfaces rom
concept in desktop computing such as
keyboards,pointing devices and graphical user interface .To make the the
wearable computer more 'usable' or more user
-
friendly, one can choose to rely
on audio as a primary medium of the interface, such that the wearab
le can
become as natural as clothing.



Continuous Audio Capture and Retrieval

Again this can be
considered as an advance in note
-
taking applications which the user might
wish to use while engaged in a conversation. While typing is slow, clumpsy
and dist
racts the user, audio can well serve as the input modality allowing
speech from the conversation to be recorded transparently without distracting
the user. Audio retrieval techniques could then be used to access useful
information at a later time. Such an
application would be useful while
attending a conference where the flow of spoken information (for example
seminars and hallway conversations) is typically overwhelming. The system
could also be useful for journalists, students and criminal investigations.





Communications Management


A WAC can be used to manage personal
communications naturally with much mobility. Mobile phones and email can
be intergrated into a single interface. Synthetic speech can be used to read
email and voice messages to the user;
speech recognition

can be used to
convert the user's responses (with constraints on vocabulary and grammar)
Wearable Computer


TheDir
ectData.com

Page
36


into text for email responses. This, in turn, combines speech recognition,
synthesis and digital

audio recording to act as a virtual secretary which
manages mobile communications, thus saving manpower for other more
important things.




Remote sensing and maintenance
::
In maintenance, repair, construction
and manufacturing

there is always a need for effective communication and
collaboration with the use of wearables, especially in providing hands
-
on
expert advice and information for the field or repair workers. To have a
wearable means that field workers would be able to ge
t remote assistance and
expertise through digital data, audio and image. With these help, even non
-
expert maintenance personnel can accomplish simple epairing tasks with the
aid of remote experts at the help desk.


















8
Future of Wearable Com
puters



As you can see, wearable computer application has spread into many


fields e.g. medical, military and entertainment. For example,




Medical field
:: the advances in developing an artificial pancreas is only
hampered by
the development of a suitable and efficient glucose
-
monitoring
sensor. The problems associated with the implantable insulin pump and its
control system has more or less been solved and any life
-
endangering risks
Wearable Computer


TheDir
ectData.com

Page
37


have almost been reduced to zero. Examples o
f problems related to glucose
monitoring are biocompatibility (if the sensor is implantable), poor sensitivity
and differentiating the glucose signal from other irrelevant sources due to
events changing in the body of the user. It must also be reliable, if

it is
implantable, since taking out the device again will be costly and tedious. Once
these problems have been overcome, a wearable of implantable glucose level
monitoring device (artificial pancreas) can be produced with ease. Other
issues to take into c
onsideration are the portability and cost of this device.
Another future application for wearable computers is to show information
about the patient instead of looking up the patient’s file every time he/she
comes in for a check up. Or even for training su
rgeons and any other medical
personnel or showing patient condition when the surgeon is operating on
him/her. Another use could be a portable device which monitors the patient's
heart rate or blood pressure and will automatically contact the doctor when th
e
patient's readings are not satisfactory. This will allow patient's would need to
be constantly monitored to be allowed out of the hospital.




Military field::

Wearable computers are already in use to provide more
information to soldiers on the battlefiel
d. Some of the issues which remain to
be solved are portability, easy user interface and telecommunications
capability. Some of the currently DARPA (Defense Advanced Research
Projects Agency) funded projects are wearable tactical information assistant
(wea
rable which gives the user maintenance, immigration and naturalization
information), wrist interactive device (wristwatch that can communicate and
display information) and combat management system (a fully configured
wearable PC into a soldier's clothing).

One of the cool devices that I think
deserves to be mentioned is a hand held MTI which is a moving target
indicator that can detect movement even behind walls. Law enforcement
agencies have been using computers to manage criminal records and court
cases.
But they are also using new technologies to capture criminals.
Examples include monitoring devices for automated finger print identification
and exchanging information among law enforcement organizations.
Combining all these components into one single wear
able device will make
the law enforcer's work much more simple and less time consuming. Imagine
in the future when a policeman could call up information regarding a certain
suspect with just one touch of a button or even better voice
-
activated. And
then us
e the same wearable to scan the suspect's fingerprints, sending it back
to the headquarters' database through a mobile phone and calling up the
suspect's records.




Entertainment field

:
:wearable computer as devices to immerse the player
more fully in the c
omputer games or experiences him/she is currently
interacting with. Currently there are already VR headsets which allows the
user to change his view just by moving his head but the player still needs to
use the keyboard, mouse or joystick as inputs. This d
oes not allow the player
Wearable Computer


TheDir
ectData.com

Page
38


to fully immerse himself in the game environment. In the future there will be
wearable gloves (or probably shoes?!?) which will allow the user to fully
move and manipulate objects in the game world. Maybe in the future the
gloves w
ill have tactile feedback so that when a player touches a game object
he/she will feel it. At the same time these devices will have to be low cost so
that the normal household can afford to purchase these devices.











9
Conclusion



Whatever area wearable computer technology is applied to you can
see that it willl improve the quality of life and make day
-
to
-
day life less complicated.
It is only our imagination which will limit the number of applications for this new
emerging techn
ology. Wearable computer is a platform for the rapid application
development, it promotes behavioral architecture and Java for the design of
applications on wearable computers.


In addition of the prototypes that has been released
to test the viability
of the architecture .There are even some reports that wearables will be the fashion of
tomorrow. It may take some time for wearables to be commonly accepted. After all, it
was once unusual to see people using cell phones or wireless m
icrophones, but they
have been embraced.

















Wearable Computer


TheDir
ectData.com

Page
39
















10
BIBLOGRAPHY:




Magazines like Electronics for you



Some websites like



Google.com



Yahoo.com



Howstuffwork.com



Friends and Teachers