-
Increasing an individual’s
quality of life via their intelligent
home
The hypothesis of this
project is: can an individual?s quality of life be
increased
by integrating “intelligent
technology” into their home environment. This
hypothesis
is
very
broad,
and
hence
the
researchers
will
investigate
it
with
regard
to
various,
potentially over-lapping, sub-sections
of the population. In particular, the project will
focus
on
sub-
sections
with
health-care
needs,
because
it
is
believed
that
these
sub-sections will
receive the greatest benefit from this enhanced
approach to housing.
Two research
questions flow from this hypothesis: what are the
health-care issues that
could
be
improved
via
“intelligent
housing”,
and
what
are
the
technological
issues
needing to be
so
lved to allow “intelligent housing”
to be constructed? While a small
number
of initiatives exist, outside Canada, which claim
to investigate this area, none
has the
global vision of this area. Work tends to be in
small areas with only a limited
idea of
how the individual pieces contribute towards a
greater goal. This project has a
very
strong sense of what it is trying to attempt, and
believes that without this global
direction the other initiatives will
fail to address the large important issues
described
within various parts of this
proposal, and that with the correct global
direction the sum
of the parts will
produce much greater rewards than the individual
components. This
new
field
has
many
parallels
with
the
field
of
business
process
engineering,
where
many
products
fail
due
to
only
considering
a
sub-set
of
the
issues,
typically
the
technology
subset.
Successful
projects
and
implementations
only
started
flow
when
people
started
to
realize
that
a
holistic
approach
was
essential.
This
holistic
requirement
also
applies to
the field of
“smart
housing”;
if we
genuinely want
it to
have
benefit to the community rather than just
technological interest. Having said this,
much of the work outlined below is
extremely important and contains a great deal of
novelty within their individual topics.
Health-Care and
Supportive housing
：
To
date,
there
has
been
little
coordinated
research
on
how
“smart
house”
technologies
can
assist
frail
seniors
in
remaining
at
home,
and/or
reduce
the
costs
experienced by their informal
caregivers. Thus, the purpose of the proposed
research
is
to
determine
the
usefulness
of
a
variety
of
residential
technologies
in
helping
seniors
maintain
their
independence
and
in
helping
caregivers
sustain
their
caring
activities.
The overall
design of the research is to focus on two groups
of seniors. The first
is seniors who
are being discharged from an acute care setting
with the potential for
reduced
ability
to
remain
independent.
An
example
is
seniors
who
have
had
hip
replacement surgery. This group may
benefit from technologies that would help them
become adapted to their reduced
mobility. The second is seniors who have a chronic
health problem such as dementia and who
are receiving assistance from an informal
caregiver
living
at
a
distance.
Informal
caregivers
living
at
a
distance
from
the
cared-for senior are at high risk of
caregiver burnout. Monitoring the cared-for senior
for health and safety is one of the
important tasks done by such caregivers. Devices
such as floor sensors (to determine
whether the senior has fallen) and access controls
to
ensure
safety
from
intruders
or
to
indicate
elopement
by
a
senior
with
dementia
could reduce
caregiver time spent commuting to monitor the
senior.
For
both
samples,
trials
would
consist
of
extended
periods
of
residence
within
the
?smart
house?.
Samples
of
seniors
being
discharged
from
acute
care
would
be
recruited from
acute care hospitals.
Samples
of
seniors being
cared for
by informal
caregivers
at
a
distance
could
be
recruited
through
dementia
diagnosis
clinics
or
through request from caregivers for
respite.
Limited
amounts
of
clinical
and
health
service
research
has
been
conducted
upon seniors (with
complex health problems) in controlled
environments such as that
represented
by the “smart house”. For ex
ample, it
is known that night vision of the
aged
is
poor
but
there
is
very
little
information
regarding
the
optimum
level
of
lighting
after
wakening
or
for
night
activities.
Falling
is
a
major
issue
for
older
persons;
and
it
results
in
injuries,
disabilities
and
additional
health
care
costs.
For
those
with
dementing
illnesses,
safety
is
the
key
issue
during
performance
of
the
activities of daily
living (ADL). It is vital for us to be able to
monitor where patients
would
fall
during
ADL.
Patients
and
caregivers
activities
would
be
monitored
and
data will be collected in the following
conditions.
Projects
would
concentrate
on
sub-populations,
with
a
view
to
collecting
scientific
data
about
their
conditions
and
the
impact
of
technology
upon
their
life
styles. For example:
Persons with stable chronic disability
following a stroke and their caregivers: to
research
optimum
models,
types
and
location
of
various
sensors
for
such
patients
(these
patients
may
have
neglect,
hemiplegia,
aphasia
and
judgment
problems);
to
research pattern of movements during
the ambulation, use of wheel chairs or canes on
various
type
of
floor
material;
to
research
caregivers
support
through
e-health
technology; to monitor frequencies and
location of the falls; to evaluate the value of
smart
appliances
for
stroke
patients
and
caregivers;
to
evaluate
information
and
communication technology
set up for Tele-homecare; to evaluate technology
interface
for Tele-homecare staff and
clients; to evaluate the most effective way of
lighting the
various part of the house;
to modify or develop new technology to enhance
comfort
and
convenience
of
stroke
patients
and
caregivers;
to
evaluate
the
value
of
surveillance systems in assisting
caregivers.
Persons with Alzheimer?s
disease and their caregivers: t
o
evaluate the effect of
smart
house
(unfamiliar
environment)
on
their
ability
to
conduct
self-care
with
and
without
prompting; to evaluate their ability to use
unfamiliar equipment in the smart
house; to evaluate and monitor persons
with Alzheimer?s disea
se movement
pattern;
to evaluate and monitor falls
or wandering; to evaluate the type and model of
sensors
to monitor patients; to
evaluate the effect of wall color for patients and
care givers; to
evaluate the value of
proper lighting.
Technology -
Ubiquitous Computing
：
The
ubiquitous
computing
infrastructure
is
viewed
as
the
backbone
of
the
“intelligence” within
the house.
In common with
all ubiquitous computing systems,
the
primary
components
with
this
system
will
be:
the
array
of
sensors,
the
communication infrastructure and the
software control (based upon software agents)
infrastructure.
Again,
it
is
considered
essential
that
this
topic
is
investigated
holistically.
Sensor
design:
The
focus
of
research
here
will
be
development
of
(micro)-sensors and sensor arrays using
smart materials, e.g. piezoelectric materials,
magneto
strictive
materials
and
shape
memory
alloys
(SMAs).
In
particular,
SMAs
are a class of smart materials that are
attractive candidates for sensing and actuating
applications primarily because of their
extraordinarily high work output/volume ratio
compared to other smart materials. SMAs
undergo a solid-solid phase transformation
when subjected to an appropriate regime
of mechanical and thermal load, resulting in
a
macroscopic
change
in
dimensions
and
shape;
this
change
is
recoverable
by
reversing the thermo mechanical loading
and is known as a one-way shape memory
effect. Due to this material feature,
SMAs can be used as both a sensor and an actuator.
A
very
recent
development
is
an
effort
to
incorporate
SMAs
in
micro-electromechanical
systems
(MEMS)
so
that
these
materials
can
be
used
as
integral
parts of micro-sensors and actuators.
MEMS are an area of activity where some
of the technology is mature enough
for
possible commercial applications to emerge. Some
examples are micro-chemical
analyzers,
humidity
and
pressure
sensors,
MEMS
for
flow
control,
synthetic
jet
actuators and optical
MEMS (for the next generation internet).
Incorporating SMAs in
MEMS
is
a
relatively
new
effort
in
the
research
community;
to
the
best
of
our
knowledge, only one group (Prof. Greg
Carman, Mechanical Engineering, University
of California, Los Angeles) has
successfully demonstrated the dynamic properties
of
SMA-based
MEMS.
Here,
the
focus
will
be
to
harness
the
sensing
and
actuation
capabilities of
smart materials to design and fabricate useful and
economically viable
micro-sensors and
actuators.
Communications: Construction
and use of an “intelligent house” offers extensive
opportunities to analyze and verify the
operation of wireless and wired home-based
communication services.
While some of these are already widely
explored, many of
the
issues
have
received
little
or
no
attention.
It
is
proposed
to
investigate
the
following issues:
Measurement of
channel statistics in a residential environment:
knowledge of
the
indoor
wireless
channel
statistics
is
critical
for
enabling
the
design
of
efficient
transmitters and
receivers, as well as determining appropriate
levels of signal power,
data
transfer
rates,
modulation
techniques,
and
error
control
codes
for
the
wireless
links.
Interference, channel distortion, and
spectral limitations that arises as a result
of equipment for the disabled
(wheelchairs, IV stands, monitoring equipment,
etc.) is
of particular interest.
Design,
analysis,
and
verification
of
enhanced
antennas
for
indoor
wireless
communications. Indoor wireless
communications present the need for compact and
rugged antennas.
New antenna designs, optimized for
desired data rates, frequency
of
operation, and spatial requirements, could be
considered.
Verification
and
analysis
of
operation
of
indoor
wireless
networks:
wireless
networking
standards
for
home
automation
have
recently
been
commercialized.
Integration of one or more
of these systems into the smart house would
provide the
opportunity
to
verify
the
operation
of
these
systems,
examine
their
limitations,
and
determine whether the standards are
over-designed to meet typical requirements.
Determination
of
effective
communications
wiring
plans
for
“smart
homes.”:
there
exist
performance/cost
tradeoffs
regarding
wired
and
wireless
infrastructure.
Measurement and analysis of various
wireless network configurations will allow for
determination of appropriate network
designs.
Consideration
of
coordinating
indoor
communication
systems
with
larger-scale
communication
systems:
indoor
wireless
networks
are
local
to
the
vicinity
of
the
residence.
There
exist
broader-scale
networks,
such
as
the
cellular
telephone
network, fixed
wireless networks, and satellite-based
communication networks.
The
viability and usefulness
of
compatibility between these services
for the
purposes of
health-care monitoring, the tracking of
dementia patients, etc needs to be considered.
Software Agents and their Engineering:
An embedded-agent can be considered
the
equivalent
of
supplying
a
friendly
expert
with
a
product.
Embedded-agents
for
Intelligent
Buildings
pose
a
number
of
challenges
both
at
the
level
of
the
design
methodology
as
well
as
the
resulting
detailed
implementation.
Projects
in
this
area
will
include
：
Architectures
for
large-scale
agent
systems
for
human
inhabited
environment:
successful deployment of agent
technology in residential/extended care
environments
requires
the
design
of
new
architectures
for
these
systems.
A
suitable
architecture
should be simple and flexible to
provide efficient agent operation in real time. At
the
same
time,
it
should
be
hierarchical
and
rigid
to
allow
enforcement
of
rules
and
restrictions
ensuring
safety
of
the
inhabitants
of
the
building
system.
These
contradictory requirements have to be
resolved by designing a new architecture that
will be shared by all agents in the
system.
Robust Decision and
Control Structures for Learning Agents: to achieve
life-long
learning
abilities,
the
agents
need
to
be
equipped
with
powerful
mechanisms
for
learning
and
adaptation.
Isolated
use
of
some
traditional
learning
systems
is
not
possible due to high-expected lifespan
of these agents. We intend to develop hybrid
learning
systems
combining
several
learning
and
representation
techniques
in
an
emergent
fashion. Such systems will apply different
approaches based on their own
maturity
and on the amount of change necessary to adapt to
a new situation or learn
new behaviors.
To cope with
high levels
of
non-determinism
(from such sources as
interaction
with
unpredictable
human
users),
robust
behaviors
will
be
designed
and
implemented capable of dealing with
different types of uncertainty (e.g. probabilistic
and
fuzzy
uncertainty)
using
advanced
techniques
for
sensory
and
data
fusion,
and
inference mechanisms based on
techniques of computational intelligence.
Automatic
modeling
of
real-
world
objects,
including
individual
householders:
The
problems
here
are:
“the
locating
and
extracting”
of
information
essential
fo
r
representation of personality and
habits of an individual; development of systems
that
“follow and adopt to” individual?s
mood and behavior. The solutions, based on data
mining and evolutionary techniques,
will utilize: (1) clustering methods,
classification
tress
and
association
discovery
techniques
for
the
classification
and
partition
of
important relationships among different
attributes for various features belonging to an
individual, this is an essential
element in finding behavioral patterns of an
individual;
and (2) neuro-fuzzy and
rule-based systems with learning and adaptation
capabilities
used
to
develop
models
of
an
individual?s
characteristics,
this
is
essential
for
estimation and
prediction of potential activities and forward
planning.
Investigation
of
framework
characteristics
for
ubiquitous
computing:
Consider
distributed and internet-based systems,
which perhaps have the most in common with
ubiquitous
computing,
here
again,
the
largest
impact
is
not
from
specific
software
engineering
process
es, but is from available
software frameworks or ?toolkits?, which
allow the rapid construction and
deployment of many of the systems in these areas.
Hence, it is proposed that the
construction of the ubiquitous computing
infrastructure
for
the
“smart
house”
should
also
be
utilized
as
a
software
engineering
study.
Researchers would
start by visiting the few genuine ubiquitous
computing systems in
existence
today,
to
try
to
build
up
an
initial
picture
of
the
functionality
of
the
framework.
(This
approach
has
obviously
parallels
with
the
approach
of
Gamma,
Helm,
Johnson
and
Vlissides
deployed
for
their
groundbreaking
work
on
“design
patterns”.
Unfortunately,
in
comparison
to
their
work,
the
sample
size
here
will
be
extremely
small,
and
hence,
additional
work
will
be
required
to
produce
reliable
answers.) This
initial framework will subsequently be used as the
basis of the smart
house?s software
system. Undoubtedly, this initial framework will
substantially evolve
during
the
construction
of
the
system,
as
the
requirements
of
ubiquitous
computing
environment unfold. It is believed that
such close involvement in the construction of a
system is a necessary component in
producing a truly useful and reliable artifact. By
the end of the construction phase, it
is expected to produce a stable framework, which
can
demonstrate
that
a
large
number
of
essential
characteristics
(or
patterns)
have
been found for
ubiquitous computing.
Validation and
Verification (V&V) issues for ubiquitous
computing: it is hoped
that
the
house
will
provide
a
test-
bed
for
investigating
validation
and
verification
(V&V)
issues
for
ubiquitous
computing.
The
house
will
be
used
as
an
assessment
vehicle to determine which, if any, V&V
techniques, tools or approaches are useful
within this environment. Further, it is
planned to make this trial facility available to
researchers
worldwide
to
increase
the
use
of
this
vehicle.
In
the
long-term,
it
is
expected
that
the
facilities
offered
by
this
infrastructure
will
evolve
into
an
internationally
recognized
“benchmarking”
site
for
V&V
activities
in
ubiquitous
computing.
Other technological
areas
：
The
project
also
plans
to
investigate
a
number
of
additional
areas,
such
as
lighting
systems, security systems, heating, ventilation
and air conditioning, etc. For
example,
with
regard
to
energy
efficiency,
the
project
currently
anticipates
undertaking two studies:
The Determination of the
effectiveness of insulating shutters: Exterior
insulating
shutters over time are not
effective because of sealing problems.
Interior shutters are
superior and could be used to help
reduce heat losses. However, their movement and
positioning
needs
appropriate
control
to
prevent
window
breakage
due
to
thermal
shock.
The initiation of an opening or
closing
cycle would
be based on measured
exterior light
levels;
current
internal
heating
levels; current
and expected use of the
house by the current inhabitants, etc.
A
comparison
of
energy
generation
alternatives:
The
energy
use
patterns
can
easily be monitored by instrumenting
each appliance.
Natural gas
and electricity are
natural choices for
the main energy supply.
The
conversion of the chemical energy
in
the fuel to heat space and warm water can be done
by conventional means or by use
of a
total energy system such as a V
olvo
Penta system.
With this
system, the fuel is
used to power a
small internal combustion engine, which in turn
drives a generator for
electrical
energy production.
Waste
heat from the coolant and the exhaust are used
to heat water for domestic use and
space heating.
Excess
electricity is fed back into
the power
grid or stored in batteries.
At a future date, it is planned to
substitute a
fuel
cell
for
the
total
energy
system
allowing
for
a
direct
comparison
of
the
performance of two advanced systems.
Intelligent architecture:
user interface design to elicit knowledge models
Much
of
the
difficulty
in
architectural
design
is
in
integrating
and
making
explicit
the
knowledge
of
the
many
converging
disciplines
(engineering,
sociology,
ergonomic
sand
psychology,
to
name
a
few),
the
building
requirements
from
many
view
points,
and
to
model
the
complex
system
interactions.
The
many
roles
of
the
architect
simply
compound
this.
This
paper
describes
a
system
currently
under
development
—
a
3Ddesign
medium
and
intelligent
analysis
tool,
to
help
elicit
and
make
explicit
these
requirements.
The
building
model
is
used
to
encapsulate
information throughout the building
lifecycle, from inception and master planning to
construction
and
?lived
-
in?
use.
From
the
tight
relationship
between
material
behaviour of the model, function
analysis and visual feedback, the aim is to help
in
the resolution of functional needs,
so that the building meets not only the aims of
the
architect, but the needs of the
inhabitants, users and environment.
The
Problem of Designing the Built
Environment
：
It
is often said that architecture is the mother of
the arts since it embodies all the
techniques
of
painting:
line,
colour,
texture
and
tone,
as
well
as
those
of
sculpture:
shape, volume, light and shadow, and
the changing relative position of the viewer, and
adds to these the way that
people
inhabit
and
move
through
its
space
to
produce
—
at its
best
—
a spectacle reminiscent
of choreography or theatre. As with all
the arts, architecture is subject to
personal critical taste and yet architecture is
also a
public art, in that people are
constrained to use it. In this it goes beyond the
other arts
and is called on to
function, to modify the climate, provide shelter,
and to subdivide
and
structure
space
into
a
pattern
that
somehow
fits
the
needs
of
social
groups
or
organizations
and
cultures.
Whilst
architecture
may
be
commissioned
in
part
as
a
cultural or aesthetic
expression, it is almost always required to
fulfill a comprehensive
programme of
social and environmental needs.
This
requirement to function gives rise to three
related problems that characterize
the
design
and
use
of
the
built
environment.
The
first
depends
on
the
difference
between explicit
knowledge
—
that of which we
are at least conscious and may even
have
a
scientific
or
principled
understanding
—
and
implicit
knowledge,
which,
like
knowing
your
mother
tongue,
can
be
applied
without
thinking.
The
functional
programmes
buildings
are
required
to
fulfill
are
largely
social,
and
are
based on
implicit rather than explicit bodies of knowledge.
The knowledge we exploit
when we use
the built environment is almost entirely applied
unconsciously. We don?t
have to think
about buildings or cities to use them; in fact,
when we become aware of
it the built
environment is often held to have failed. Think of
the need for yellow lines
to help
people find their way around the Barbican complex
in the City of London, or
the calls
from tenants to ?string up the architects? when
housing estates turn out to be
social
disasters.
The
second
is
a
problem
of
complexity.
The
problem
is
that
buildings
need
to
function
in
so
many
different
ways.
They
are
spatial
and
social,
they
function
in
terms
of
thermal
environment,
light
and
acoustics,
they
use
energy
and
affect
people?s health, they need to be constructed and
are made
of
physical
components
that
can
degrade
and
need
to
be
maintained.
On
top
of
all
this
they
have
an
aesthetic and cultural role, as well as being
financial investments and playing an
important role in the
economy. Almost all of
these factors
are interactive
—
decisions
taken for structural reasons have impacts on
environment or
cost
—
but are often
relatively independent in terms of the domains of
knowledge that need to
be
applied. This
gives
rise to
a complex design
problem in
which everything
knocks
on
to
everything else, and in which no single
person
has a grasp of all the domains
of
knowledge required for
its resolution. Even when
the knowledge
that needs to be applied
is
relatively
explicit
—
as
for
instance
in
structural
calculations,
or
those
concerning
thermal
performance
—
the
complex
interactive
nature
of
buildings
creates a situation in which it is only
through a team approach that design
can
be
carried
out,
with
all
that
this
entails
for
problems
of
information
transfer
and
breakdowns in
understanding.
The third is the problem
of ?briefing?. It is a characteristic of building
projects
that
buildings
tend
not
to
be
something
that
people
buy
?off
-the-
shelf?.
Often
the
functional
programme
is
not
even
explicit
at
the
outset.
One might
characterise the process that actually takes place
by saying that the design
and
the
brief
?co
-
evolve?.
As
a
project
moves
from
inception
to
full
s
pecification
both
the
requirements
and
the
design
become
more
and
more
concrete
through
an
iterative
process
in
which
design
of
the
physical
form
and
the
requirements that it is
expected to fulfill both develop at once. Feasible
designs are
evaluated
according
to
what
they
provide,
and
designers
try
to
develop
a
design
that
matches
the
client?s
requirements.
Eventually,
it
is
to
be
hoped,
the
two
meet
with
the
textual
description
of
what
is
required
and
the
physical
description of the building that will
provide it more or less tying together as the
brief
becomes a part of the contractual
documentation that the
client signs up
to.
These three problems compound
themselves in a number of ways. Since many of
the core objectives of a client
organization rest on implicit
knowledge
—
the need for a
building
to
foster
communication
and
innovation
amongst
its
workers
for
instance
—
it is all too easy
for them to be lost to sight against the more
explicitly
stated requirements such as
those concerned with cost, environmental
performance or
statutory
regulations.
The
result
is
that
some
of
the
more
important
aspects
of
the
functional programme can lose out to
less important but better understood issues. This
can
be
compounded
by
the
approach
that
designers
take
in
order
to
control
them
complexity of projects. All too often
the temptation is to wait until the general layout
of
a
building
is
?fixed?
before
calling
in
the
domain
experts.
The result
is
that functional design has
to resort to retrofitting to resolve problems
caused by the strategic plan.
The
Intelligent
Architecture
project
is
investigating
the
use
of
a
single
unified
digital
model
of
the
building
to
help
resolve
these
problems
by
bringing
greater
intelligence to bear at
the
earliest ?form generating? phase of the design
process when
the client?s requirements
are still being specified and when both physical
design and
client
expectations
are
most
easily
modified.
The
aim
is
to
help
narrow
the
gap
between
what
clients
hope
to
obtain
and
what
they
eventually
receive
from
a building project.
The strategy is simple. By capturing
representations of the building as a physical
and spatial system, and using these to
bring domain knowledge to bear on a design at
its earliest stages, it is hoped that
some of the main conflicts that lead to sub-
optimal
designs
can
be
avoided.
By
linking
between
textual
schedules
of
requirements and the
physical/spatial model it is intended to ease the
reconciliation of
the brief and the
design, and help the two to co-evolve. By making
available some of
the
latest
?intelligent?
techniques
for
modelling
spatial
systems
in
the
built
environment, it is hoped to help put more of the
implicit knowledge on an equal
footing
with
explicit
knowledge,
and
by
using
graphical
feedback
about
functional
outcomes where explicit knowledge
exists, to bring these within the realm of
intuitive
application by designers.
The Workbench
：
In
order
to
do
this,
Intelligent
Architecture
has
developed
Pangea.
Pangea
has
been designed as a
general-purpose environment for intelligent 3D
modelling
—
it does
not pre-suppose a particular way of
working, a particular design solution, or even a
particular application domain. Several
features make this possible.
Worlds can
be constructed from 3D and 2D primitives
(including blocks, spheres,
irregular
prisms
and
deformable
surfaces),
which
can
represent
real-world
physical
objects,
or
encapsulate
some
kind
of
abstract
behaviour.
The
3D
editor
provides
a
direct
and simple interface for manipulating
objects
—
to position,
reshape, rotate and
rework.
All
objects,
both
physical
and
abstract,
have
an
internal
state
(defined
by
attributes),
and
behaviour,
rules
and
constraints
(in
terms
of
a
high-level-language
?script?).
Attributes
can
be
added
dynamically,
making
it
possible
for
objects
to
change
in
nature,
in
response
to
new
knowledge
about
them,
or
to
a
changing
environment.
Scripts
are
triggered
by
events,
so
that
objects
can
respond
and
interact, as in the built environment, molecular
systems, or fabric
falling into folds
on an irregular surface.
Dynamic
linking
allows
Pangea?s
functionality
to
be
extended
to
include
standard
?off
-the-
peg?
software
tools
—
spreadsheets,
statistical
analysis
applications, graphing
packages
and domain-specific analysis software,
such
as
finite
element analysis for air-
flow
modelling. The
?intelligent
toolkit?
includes
neural
networks
[Koho89]
[Wass89],
genetic
algorithms
[Gold89]
[Holl75]
and
other
stochastic
search
techniques
[KiDe95],
together
with
a
rule-
based
and
fuzzy
logic
system
[Zade84].
The
intelligent
tools
are
objects,
just like the
normal
3D primitives:
they
have 3D presence and
can interact
with other 3D objects. A
natural consequence of this design is
easy ?hybridisability? of
techniques,
widely
considered
as
vital
to
the
success
of
intelligent
techniques
in
solving
realistically
complex
problems
[GoKh95].
This
infrastructure
of
primitive
forms,
intelligent
techniques
and
high-level
language
makes
it
possible
to
build
applications
to
deal
with
a
broad
range
of
problems,
from
the
generation
of architectural
form,
spatial
optimisation,
object
recognition
and
clustering,
and
inducing
rules
and
patterns
from
raw
data.
Embedding
Intelligence
：
Many consider that there is an
inevitable trade-off between computers as a pure
design
medium,
and
computers
with
intelligence,
?as
a
thinking
machine?
[Rich94].
We propose here that it is possible to
provide both these types of support, and allow
the user to choose how best to use
each, or not, according to the situation.
It is essential that the creative role
of the architect is preserved as he or she uses
the
work
bench,
that
the
architect
as
artist
may
draw
manipulate
the
world
as
seen
through the workbench as freely as they
would when using a sheet of paper. Much of
the
knowledge
entered
into
the
workbench
in
this
way
is
unexpressed:
an architect may draw a block, but
intend that the block to be a stair or
a
door
or
a
room.
By
using
a
clustering
algorithm
we
have
tried
to
capture
some
of
the
knowledge
that
the
architect
as
artist
does
not
express.
In
this
manner,
the
architect
as
engineer
may
then
pick
up
the
sketch
and
continue
to
work
with a technical drawing. Once we have identified
the components, we can also
apply
rule-based
systems
to
make
recommendations
for
the
design.
Thus,
by
using
a
clustering algorithm, we are crossing
the bridge from implicit knowledge
of
the architect as artist to the explicit knowledge
of architect as engineer.
The
object-oriented
nature
of
the
workbench
allows
a
common
interface
to
the
clusterer,
while
implementing
the
clustering
engine
itself
by
a
choice
of
simple
linkage
clustering
algorithm
or
neural
network
(either a back-propagation multi-layer
network or Kohonen). The architect specifies
the attributes of an object in the
world that are considered important to the
definition
of a set of objects, and we
make a normalised vector representation of those
attributes.
Attributes
chosen
might
be
volume
or
the
greenness
or
an
arbitrary
combination
of
attributes
such
as
these.
The
clustering
algorithm
is
applied
to
these
vectors,
and
places each cluster of objects into a
collection. After clustering, the architect can
name
the
set
of
objects
retrieved
(for
example,
?doors?
or
?stairs?);
the
workbench
can
make
sensible guesses as to
which new objects belong to those
clusters, and the workbench or
architect can use the information to reason about
the
clusters.
Since each
object
can
contain inside
itself the rules relating to
it, these
can be
triggered
into
operation
when
a
particular
event
occurs.
For
example,
when
a building
core (the area containing lifts, stairwells,
ventilation shafts, and so
on) is
moved, an event is sent to it, and it can
automatically go through its set of rules
to determine if it still meets
regulations for access and fire safety. This is
particularly
valuable when many rules
need to be processed simultaneously, making it
possible to
investigate
various
designs,
with
rapid
feedback.
This
can
be
taken
further,
by
encapsulating rules in ?expert? or
?critic? objects [FNOS93], whose job it is to
oversee
-
-
-
-
-
-
-
-
-
上一篇：中国伊斯兰教协会章程
下一篇：翻译句子(中译英)