Constraints & Design Problem Solving

geographertonguesΤεχνίτη Νοημοσύνη και Ρομποτική

30 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

108 εμφανίσεις




















Constraints & Design Problem Solving


Victoria Smy

SmyVA@Cardiff.ac.uk

Requires various cognitive abilities including analysis, prioritisation, decision
-
making and planning (Ball et al., 2001; Dym et al., 2005; Hacker, 1997).

Typically involves ill
-
specified, incomplete or ambiguous goals and
specifications (Ball et al., 1997).

No obvious ‘best’ solution (Purcell & Gero, 1996; Römer, et al., 2000).

Why is design difficult?

(Constraints

displayed here)

Design is relied upon in some form in most everyday situations (transport,
buildings etc.)

Design permeates all aspects of research & development.

Consequences and costs of bad & inappropriate design


Why is design important?

What is known about the cognitive
elements of design?


Smith

&

Browne

(
1993
)

highlight

5

crucial

elements

of

a

design

problem

space

that

are

intertwined

with

the

cognitive

processes

of

designers
.

1.
Goals



useful

for

evaluative

purposes
.

2.
Solutions



usually

satisfactory,

never

optimal
.

3.
Alternatives



more

alternatives

associated

with

higher

levels

of

design

creativity/innovation
.

4.
Representations



necessary

for

understanding

of

the

design

problem
.

5.
Constraints

-

???

The experimental paradigm:
Office Layout Design

Design requirement:


Design an office layout incorporating the employees specified at the
bottom of the screen. Each employee requires a separate office that must
be attached to the right
-
hand side of a corridor.


Design constraints:


When producing your design you should aim to incorporate as many of
the
rules for constraint implementation

and
explicit constraints
(employee pairings)

presented as possible.


Example rules: “An employee who is higher in status than another should
have an office that is closer to the central corridor”, and “Employees who
are incompatible should not have adjacent offices, employees who are
compatible should be adjacent”.


Example explicit constraints: “B is higher in status than E”, “G is
incompatible with D”.


Design outcomes measured:



Number of employee pairings satisfied in end design.



Time to complete design

Summary?


The

systematic

manipulation

of

explicit

design

constraints

in

a

spatial

layout

can

contribute

to

design

difficulty
.

However,

this

is

not

evidenced

in

design

completion

times,

which

vary

considerably

from

participant

to

participant
.

In

addition,

the

actual

number

of

constraints

does

not

appear

to

effect

design

efficiency,

with

designers

being

able

to

maintain

performance

despite

a

considerable

increase

in

constraints
.

Instead,

it

is

the

number

of

rules

for

constraint

implementation,

and

the

nature

of

these

rules

(i
.
e
.
,

whether

or

not

the

rules

are

can

be

easily

mapped

to

the

problem

representation)

that

has

an

effect

on

the

number

of

explicit

constraints

that

are

satisfied
.


Key

References
:



Carroll, J. M., Thomas, J. C., & Malhotra, A. (1980). Presentation and representation in
design problem
-
solving.
British Journal of Psychology, 71,

143
-
153.


Smith, G. F., & Browne, G. J. (1993). Conceptual foundations of design problem solving.
IEEE Transactions on Systems, Man, and Cybernetics, 23,

1209
-
1218.

Experiment 1


N# explicit constraints

Experiment 2


number of constraint rules

Experiment 3


nature of constraint rules

20 UGDs completed two versions of an office design task.

Tasks contained either 15 or 27 explicit constraints (employee pairings), and 3
rules for constraint implementation.

15 constraint
task

27 constraint
task

t test

Constraints
satisfied (%)

75%

(SD 16.74)

76%

(SD 8.35)

t (19) = .17,


p

= .865

Design Time
(secs)

573.40

(SD 194.8)

641.72

(SD 209.5)

t (19) = 1.21


p

= .243

Results indicate no decrease in design performance as a result of increasing
the number of explicit design constraints


may be that the increase in number
of constraints did not exceed the threshold at which constraints induce design
difficulty.

30 UGDs completed one of two versions of an office design task.

Tasks contained either 3 or 6 rules for constraint implementation, and 24
explicit constraints.

3 rule task

6 rule task

t test

Constraints
satisfied

17.80

(SD 2.76)

13.87

(SD 2.39)

t (28) = 4.18,


p

< .001

Design Time
(mins)

12.87

(SD 3.27)

12.01

(SD 3.67)

t (28) = .70


p

= .493

Results indicate that increasing the number of design rules decreases the
number of explicit constraints participants successfully implemented. There
was no effect on time taken to complete an office design .

Are rules and explicit constraints that can be directly mapped onto the design
representation (i.e., ones which mention the position of a concrete feature of
the representation such as ‘reception’, as opposed to rules that mention the
relative position of another employee) easier to implement?


44 UGDs completed one of four versions of an office design task. Tasks
contained 3 rules for constraint implementation and 27 explicit constraints.
Each version varied in the number of ‘mapped’ and ‘non
-
mapped’ rules.

2
-
3 mapped
rules

0
-
1 mapped
rules

t test

Constraints
satisfied

18.91

(SD 2.39)

12.38

(SD 3.46)

t (42) = 7.34,


p

< .001

Design Time
(mins)

19.21

(SD 5.76)

18.80

(SD 4.84)

t (42) = .25


p

= .802

Results indicate that having more rules that can be directly mapped onto a
problem representation leads to more design constraints satisfied. Once again,
time remained unaffected by manipulations to design constraints.

Future directions

Can individuals be trained to negotiate design constraints more efficiently?