Information Technology Reference
In-Depth Information
Table 1. Overview of methods used to evaluate the user-interface for in-car computing systems
Overall
Measures
Primary
Advantages
Primary
Disadvantages
Method
Environment Task manipulations
Primary/ secondary
task performance/
behaviour, user opin-
ions, etc.
Ecological validity,
can assess behav-
ioural adaptation
Resource intensive,
ethical/liability is-
sues to consider
Real road (in ev-
eryday driving)
Multi-task (according
to driver motivation)
Field trials
Primary/ secondary
task performance/
behaviour, user opin-
ions, etc.
Multi-task (com-
monly, evaluator-
manipulated)
Balance of eco-
logical validity with
control
Resource intensive,
ethical/liability is-
sues to consider
Real road (in pre-
defined settings)
Road trials
Primary/ secondary
task performance/
behaviour, user opin-
ions, etc.
Virtual driving en-
vironment (vary-
ing in fidelity)
Multi-task (com-
monly, evaluator-
manipulated)
Control over vari-
ables, safe environ-
ment, cost-effective
Validity of driver
behaviour, simula-
tor sickness
Simulator trials
Secondary task
achieved in con-
trolled visual experi-
ence
Limited scope,
concern over valid-
ity of approach and
metrics
Standardised ap-
proach, control over
variables
Laboratory/ stati-
cally in car
Visual demand of
user-interface
Occlusion
Multi-task (although
commonly, evaluator-
manipulated)
Assesses cognitive,
as well as visual
demand
Can be resource
intensive, range of
approaches
Peripheral
detection
Road/virtual driv-
ing environment
Visual/ cognitive
workload
Specific lo-fidelity
virtual driving en-
vironment
Standardised ap-
proach, control over
variables
Difficult to relate
results to interface
characteristics
Multi-task motorway
driving scenario
Primary lateral con-
trol of vehicle
Lane change task
Secondary task
achieved without
presence of driving
task
Only relates to
certain aspects of
visual demand
Laboratory/ stati-
cally in car
Secondary task time
(whilst stationary)
15 second rule
Simple approach
No user trials take
place - models expert
performance
Only relates to
certain aspects of
visual demand
Keystroke-Level
Model (KLM)
Modelling exer-
cise
Secondary task time
(whilst stationary)
Quick/cheap, analy-
sis explains results
As for KLM, but
with additional as-
sumptions
Modelling exer-
cise
Visual demand of
user-interface
Quick/cheap, analy-
sis explains results
Requires reliability
assessments
Extended KLM
and configurations range from those with single
computer screens and game controller configu-
rations, through to real car cabins with multiple
projections and motion systems. An example of
a medium fidelity driving simulator is shown in
Figure 3.
Driving simulators have become increasingly
popular in recent years as a result of reduced
hardware and software costs, and potentially offer
an extremely cost-effective way of investigating
many different design and evaluation issues in a
safe and controlled environment (Reed and Green,
1999). Nevertheless, there are two key research
issues concerning the use of driving simulators.
Firstly, it is well known that individuals can ex-
perience symptoms of sickness in driving simula-
tors, manifested as feelings of nausea, dizziness
and headaches. There has been considerable re-
search regarding such sickness in virtual environ-
ments, and whilst there is still debate regarding
the theoretical basis for the phenomenon (see for
instance Nichols and Patel, 2002), there is practi-
cal guidance for those using driving simulators.
For instance, screening questionnaires can be used
to eliminate individuals who are most likely to
experience sickness during a trial (Kennedy et
Search WWH ::
Custom Search