TARGET SIZE GUIDELINES
FOR INTERACTIVE DISPLAYS ON THE FLIGHT DECK
Huseyin Avsar, Joel Fischer, Tom Rodden
The University of Nottingham, United Kingdom
The avionics industry is seeking to understand the
challenges and benefits of touchscreens on flight
decks. This paper presents an investigation of
interactive displays on the flight deck focusing on the
impact of target size, placement and vibration on
performance. A study was undertaken with search and
rescue (SAR) crew members in an operational setting
in helicopters. Results are essential to understand how
to design effective touchscreen interfaces for the flight
Results show that device placement, vibration
and target size have significant effects on targeting
accuracy. However, increasing target size eliminates
the negative effects of placement and vibration in most
cases. The findings suggest that 15 mm targets are
sufficiently large for non-safety critical Electronic
Flight Bag (EFB) applications. For interaction with
fixed displays where pilots have to extend their arms,
and for safety critical tasks it is recommended to use
interactive elements of about 20 mm size.
Digital devices have long since started to replace
analogue input devices on the flight deck.
Considerable changes have consolidated the number
of inputs (e.g. buttons, switches and knobs) and
outputs (e.g. displays). More recently, suppliers for
cockpit equipment have started to explore
opportunities for the integration of touchscreens in and
around the cockpit. From the manufacturer’s
perspective, the key advantage of touchscreens is that
they are adaptable to any configuration by changing
the underlying software, and they do not require
removing and reconfiguring physical input devices
. These technologies could lead to a point where
physical input devices completely disappear from the
flight deck and interactions with the aircraft system
occur exclusively through interactive displays .
Touchscreens entered the cockpit environment
through portable electronic devices (PED). The usage
was similar to an electronic flight bag (EFB). Pilots
were able to make performance calculations, create
flight plans and utilise various formats of charts and
checklists . From an air carrier’s point of view, the
benefits were reduced operational costs and crew
Leading avionics manufacturers such as
Honeywell  and Thales  have shown increasing
interest in integrating touchscreens into the cockpit.
Touchscreens for all types of aircraft are appearing,
but requirements differ for each application. Use in
safety-critical applications places a high demand on
the operator to input data accurately. For example,
SAR operations involve challenging conditions in
which the operator has to enter data while being
exposed to strong vibrations. Pilots are likely to
encounter stronger turbulences that could impede the
usability of touchscreens in helicopters, especially
when operating at lower altitudes. Two-thirds of fatal
accidents are caused by human error , which makes
designing a usable interface more important.
This work addresses the challenge how to design
these touchscreens so that they are effective and
ultimately usable by pilots. Previous studies have
found that the biggest drawbacks of soft buttons (i.e.
interactive elements) compared to their physical
counterparts are unwanted and accidental touches 
and absence of tactile feedback . The size of
interactive elements (e.g. buttons), called ‘target size’,
has a significant impact on these errors.
This paper seeks to develop design guidelines and
recommendations for integration of interactive
displays into helicopter flight decks. In a real-world
setting, this study investigates the impact of vibration
(cruise, transition and hover), placement (mobile and
fixed) and target size (5, 10, 15, 20 mm) on targeting
accuracy on touchscreens on a helicopter flight deck.
Experiments were conducted with the Spanish
Maritime Safety Agency during training flights.
Mobile device suppliers like Apple , Google
, and Microsoft  have their own
recommendations for target sizes, which are in general
a compromise between acceptable error rate and
available screen area . In academia, target sizes
have been tested in many different conditions.
Independent variables that have been studied include
activity (walking  or standing ), mobility
(mobile devices  or fixed devices ), usage (one
handed thumb  or both hands ), feedback
modality (auditory and haptic ), target population
(older adults ) and task (alphanumeric text entry
, numeric text entry  and tapping task ).
The majority of the experiments compared larger
targets versus smaller targets and investigated if
spacing between targets would have a significant
effect on the overall performance. Common results
show that larger targets result in better accuracy than
smaller targets, and that “small” spacing between the
targets does not have a significant impact.
Schedlbauer  evaluated the performance and
accuracy of data input on keypads by using a fixed
experimental apparatus, where the task was to type 10
digit GPS coordinates. His results showed that a key
size of 15 mm appears to be sufficiently large to
provide acceptable accuracy (error rate: 1.9%). This
value was confirmed by Tsang et. al  who
performed a similar experiment and defined 15 mm
targets as a cut-off point where target sizes below end
up with higher error rates. Another finding was that
there is no further improvement for key sizes beyond
20 mm. This outcome is supported by Colle and
Hiszem , who could not find a significant
difference between key size of 20 and 25 mm.
Henze and colleagues  developed a tapping
task game for smartphones. This was an unsupervised
experiment, which found that targets below 15 mm
had an increased error rate. The error rate increased to
over 40% for targets smaller than 8 mm. Leitao and
Silva , published interface design guidelines for
older people. Participants performed tapping and
swiping tasks on a handheld device. In their study, 14
mm could be considered as a break-even point since
there was no significant difference for larger targets.
Another study  with mobile devices found
that walking degrades the error rate significantly.
While standing, users performing a two-dimensional
tapping task made on average 6.77% fewer errors. The
largest tested target size was 9.5 mm (error rate 16%).
The authors claim that increasing the target size by
40% would compensate for the negative effects of
walking. Bergstrom-Lehtovirta et al.  performed
target selection while walking on a treadmill, and
conclude that all types of walking, regardless of speed,
causes a noticeable decrease in accuracy.
For applications in vehicles or with the potential
use of gloves, the Department of Defense (DOD) 
recommended target sizes between 10 mm and 25 mm.
The Federal Aviation Administration (FAA) advised
designers to demonstrate that integration of
touchscreens should not result in unacceptable levels
of workload and error rates . There was no explicit
guidance on minimum target size or acceptable error
rate under high-vibration conditions that are
particularly likely in helicopter operations.
The flight deck is an environment, in which errors
need to be minimized. However, there is little research
about the impact of dynamic (e.g. vibrating, turbulent)
environments. During a flight, pilots could face
particular difficulties operating touchscreen devices
when the display is moving or vibrating independently
from the body. Recently, Dodd et al.  published
research performed in a flight simulator, and found
that turbulence has a significant effect on error rates.
Their experimental design suggests that this research
was focused on commercial aircraft (above 8000 feet,
at an airspeed of approximately 250 knots). Since
general aviation aircraft and helicopters are smaller,
lighter and operating at lower altitudes, pilots are
likely to feel higher vibrations/turbulences. Thus,
results from a commercial aircraft setting may not be
The purpose of this research is to establish design
guidelines and recommendations for target sizes on
fixed and mobile touchscreens on a helicopter flight
Key hypotheses driving this work are:
Vibration, placement and target size have a
significant negative effect on error rates.
Increasing target size will minimize the negative
effects of vibration and placement.
Participants make fewer errors when the device
placement is mobile compared to when it is fixed.
The research was carried out in a Search and
Rescue (SAR) setting. Our site of study was the
Spanish Maritime Safety Agency, also known as
SASEMAR, between April and May 2015.
SASEMAR has eight identical Agusta Westland
AW139 Helicopters (Figure 1) distributed along the
Spanish coast. Data was collected during 12 training
flights in four different bases (Reus, Valencia, Almeria
and Jerez). The crew conducted the experiments at
their own discretion, in periods of downtime from their
Crews operate on 12-hour shift. Apart from
scheduled training and patrol flights, crews do not
know when and where they are going. Because of the
nature of rescue missions, response time is critical.
Once a distress call is received, the crew is ready to
take off within 15 minutes. In the air (1500-2000 feet
above ground level), the crew flies with maximum
cruise speed (120-130 knots) to the target location.
Targets could be small and moving objects such as a
person over board or small watercraft. Helicopters
may have to operate in challenging areas (sea or cliffs)
and weather conditions.
During training flights, the crew is simulating
possible scenarios. Variables for such operations are
search required or not required, target type, rescue
procedure, and rescue equipment used. For each
training flight, two or three possible scenarios will be
trained. This kind of training flight takes on average
2:15 hours. There are four crew members: pilot, co-
pilot, hoist operator and rescue swimmer. Each crew
member has separate responsibilities, and they are
interacting with each other continually.
Figure 1. SASEMAR AW139.
In real rescue missions, the pilot is usually the on-
scene coordinator (OSC), who coordinates all other
Detailed information about SAR operations are
available in the IAMSAR (International Aeronautical
and Maritime Search and Rescue) Manual .
We adopted a mixed methods approach. A series
of experiments (described below) were undertaken in
a lab setting prior to moving to more open-ended field
investigation in a real-world setting. Initial
experimental results showed significant differences in
targeting accuracy and movement time for using
touchscreens in a static environment compared to a
dynamic (vibrating) environment. This motivated the
transfer of experiments into a real-world setting to
achieve ecologically valid results.
The target population are pilots. However, for
safety reasons pilots could not directly participate in
field trials. Participants were hoist operators and
rescue swimmers. 14 male crew members conducted
the experiment. Their age ranged from 27 to 52 years
old (M=35.6, SD=11.8). Two of the participants were
left-handed. The number of years on duty ranged from
3 to 25 years (M=9.6, SD=8.6). 13 Participants used a
touch-enabled device (smartphone or tablet) and rated
their touchscreen skills on a 10-point scale. (10 means
very good) (M=7.9, SD=0.9).
In initial research aimed at learning about the
features, content and functionality that pilots would
like to see in an electronic flight bag (EFB), we asked
what kind of tablet device they would prefer to use
within the cockpit.
Results from pilot trials showed that an 8-inch
tablet would be sufficiently large to display flight
related information. Three pilots already used an iPad
Mini as an EFB. Thus, an Apple iPad Mini (7.9” with
capacitive touchscreen) was used for the entire
During the flight, vibrations were recorded with a
Samsung Galaxy S4 (GT-I9505). The onboard
accelerometer sensor is a K330 3-axis from
STMicroelectronics. The resolution is 0.001m/s2 and
the range is 19.613m/s2. Minimum delay is 0.01
Experiments were performed with two different
device placements (mobile and fixed). In the mobile
condition, participants hold the device while
performing the experiment. In the fixed condition
(Figure 2), the tablet is attached to a suction cup holder
mounted on the window. The distance from the seating
position is 65 cm, which is approximately the same
distance as that between pilots and the main
instrument panel. Some double-sided tape was affixed
to the window in order to stabilize the tablet in its
position and to absorb its vibrations.
A 2x3x4 within-subjects design with repeated
measures was used for the experiment.
Independent variables in this experiment were
placement (2 levels - fixed and mobile), vibration (3
levels – cruise, transition and hover) and target size (4
levels – 5 mm, 10 mm, 15 mm and 20 mm). The
minimum target size (5 mm) was determined using
Google’s Design Guidelines. The largest target size
(20 mm) was adopted from previous work, in which
authors achieved almost 100% accuracy. The target
was displayed randomly, and the position and size of
the target was recorded.
Recorded dependent variables were movement
time, touch position, distance and error rate. There was
no minimum quantity of data that participants had to
generate during a flight.
An application called “Physics Toolbox
Accelerometer”  was used to record vibrations
within the aircraft. Measurements were taken in three
different locations. The first measurements were
collected at the point where the experiment was
conducted with fixed device placement. These
measurements were compared with another
measurement on the dashboard (Figure 3). The
smartphone was attached between the Multi-Function
Display and Central Display Unit. When the
placement was mobile, participants held the device in
their hand with the aim to see whether the human body
is able to compensate a certain amount of vibration. 50
measurements were recorded per second.
Another research objective aimed at
understanding how pilots interact with the cockpit
system; thus, video recordings were made. The camera
was positioned at an angle from which it was able to
capture the pedestal, dashboard and the outside view
from the pilot’s side (see Figure 4).
Figure 2. Experimental Setting (fixed placement).
Figure 3. Vibration Measurement.
Figure 4. Flight Recording.
These recordings were used to double-check in which
flight mode (cruise, transition, or hover) the aircraft
During the flights, the investigator was on board,
controlling the order of the experiment and recording
events that occurred at specific times.
Events include briefing, initiated checklist,
engine start, taxi, take-off, landing, transition to cruise
and hover, cruise, approach target, type of target, type
of training, rescue swimmer preparation start, hover,
door open, relative position to the target, dummy and
rescue man down and up events; used equipment,
changes in speed and altitude, how participants held
the device when it was mobile, and change from
mobile to fixed placement were all recorded.
The ISO 9241-9  recommended task design
for input devices evaluation is illustrated in
Figure 5. In this multi-directional tapping task
targets are arranged around a circle. The task is to tap
all targets in a consecutive order. Taps outside of the
circle are recorded as an error. The distance (D) and
the width (W) changes after the trial is completed.
This task design was tried out in the lab. Initial
results showed that participants tended to hover their
finger over the next target before clicking the current
target with the other hand. This kind of predictability
would bias the movement time measurements
compared to realistic operational use.
Restricting participants to use only one hand
would have conflicted with the goal of seeing how
participants use the device in a real world situation. It
was not intended to compare results with prior work
that applied the ISO task design. It was decided to
create a task in which the size and the distance of the
targets changed dynamically after each target.
A tapping task (land-on touch strategy) was
tap targets (displayed as red circles) sequentially. The
app recorded performance data in a .csv file.
Data recording occurs as follows: the first target
is displayed and the user taps the target. The position
of the target and the actual touch position are recorded.
The current target disappears and the next target is
displayed, the user taps the next target. Again, the
actual target and touch position are recorded. Using
time stamps the duration between two targets
(movement time in milliseconds) is calculated and
stored. In addition, the distance between targets is
recorded. Touching outside the target is recorded as an
error. The target remains until the user touches the
target. The number of errors per task are recorded. The
mean errors are calculated by dividing the number of
errors by the number of tasks. This paper covers error
rate and vibration analyses.
The aims and objectives were explained to
participants. Each participant was notified that the aim
was to investigate the impact of vibration and
turbulence to targeting accuracy and movement time
on touch-enabled devices. Participants were asked to
be as accurate as possible, while performing the task
at a normal pace.
Figure 5. ISO-9241 Input Device Evaluation Task.
Figure 6. Tapping Task and Recorded Variables.
The experiment started with a baseline
determination, replicating previous work. Participants
conducted some trials on the ground to practice.
Figure 7 illustrates the default positions of each
crew member during take-off. The investigator sat on
the seat from which the experiment would be
conducted in the fixed placement condition.
In the following sections, possible time frames
are described, in which crew members were able to
perform the experiment. To avoid fatigue effects, the
investigator asked participants to stop after 5 minutes.
Participants took their gloves off during the
experiment. Some hoist operators had gloves without
index finger, thus they were able to conduct the
experiments while wearing gloves.
Before take-off, the screen of the tablet was
cleaned. The experiment started in the mobile
placement condition. After take off the rescue
swimmer started with the tapping targets activity.
After approximately 5 minutes, the rescue swimmer
handed over the tablet to the hoist operator and he
continued the experiment. The pilot notified the
persons in the rear cabin approximately 10 minutes
before reaching the target. The rescue swimmer started
with preparations. The investigator gave the hoist
operator a signal when the transition to hover was
attempted (around 80 knots). Once the aircraft was in
hover, pilots required on average 3 minutes to position
the aircraft close to the target. The hoist operator
handed over the tablet to the rescue swimmer. The
rescue swimmer continued with the experiments. The
hoist operator opened the door and spoke with the pilot
to make fine adjustments for the position of the
aircraft. It was also possible for the hoist operator to
take full control over the aircraft and position the
aircraft by using his controller. At this stage, the
experiment was done in the mobile condition for all
flight modes (cruise, transition and hover).
After the first training was completed and the
door was closed, the investigator attached the tablet
device to the fixture. From that point, the experiments
were conducted using the fixed placement.
Participants were requested not to fasten seatbelts to
save time. However, participants were asked not to
lean towards the display. The helicopter flew away
from the target and circled. The investigator swapped
his seat with the hoist operator. Once the helicopter
approached the target (when transitioning occurred),
the hoist operator started with the taps. The hoist
operator finished the task once the helicopter was
ready for opening doors. He swapped his seat with the
rescue swimmer who continued with the task. The
rescue swimmer stopped once his duty started.
Once the second training was completed, the
hoist operator closed the door and the helicopter took
off and turned for the third scenario if there was one,
otherwise, the crew returned to base. During this
transit flight, the crew would perform the experiment
again. Approximately 10 minutes before landing, the
investigator gave the hoist operator a signal to start the
experiments; after 5 minutes, he swapped with the
rescue swimmer who performed the experiments until
Data was recorded in nine flights as mentioned
above. At this point, it was noticed that more data had
been collected in the mobile condition than with the
fixed placement. Thus, during the last three flights the
experiment was conducted only in the fixed condition.
Figure 7. Aircraft Layout illustrating the Experimental Setup.
The application recorded the acceleration in x, y,
and z directions with a timestamp. The magnitude of
the vibration was calculated by using Equation 1.
𝑀 = √𝑥2+ 𝑦2+ 𝑧2
At least 15 measurements are recorded per
second. The flight protocol and recordings were used
to determine the timeframes for specific flight modes.
The data was annotated with a key value describing
the flight mode. The key value is the same as described
in the next section. Timelines are added to visualize
flight modes. (Note: transition phases are the
timeframes between cruise and hover)
Figure 8 shows vibrations recorded during a
flight in Valencia. The smartphone was attached to
another suction cup holder, which is mounted behind
the fixed device placement (see Figure 7). For this
particular flight, the mean vibration for cruise was
around 5 m/s2, for transition 12 m/s2 and for hover 7
Figure 8. Vibration Measurement in Fix Position
Figure 9. Vibration Measurement on the Dashboard
Figure 10. Mobile Vibration Measurement
However, this does not mean that vibrations
always lead to the same values. The airspeed is a
significant factor during cruise that can cause high
vibrations. During this flight, the cruise speed was
always below 120 knots. During a different flight in
Reus, the cruise speed was sometimes over 130 knots
and the smartphone measured a mean vibration of 6
Depending on the weather and location,
vibrations during hover could be as small as 4 m/s2.
The magnitude of vibrations during transition phases
depend on how fast the pilot transitions through the
critical speed where the vibrations are highest. Thus,
the measurements reflect when the pilot decreased
speed during a transition down phase more slowly. In
this transition phase, vibrations of more than 15 m/s2
The data shown in Figure 9 was recorded on the
main instrument panel during a night flight in Almeria.
Vibrations for cruise were around 3 m/s2, hover were
2.5 m/s2 and transitions were 5 m/s2. The second
recording in this setting had similar values.
The last Figure 10 is a collection of different
vibration measurements, which were taken on the
hand of participants, to see whether the human body is
able to compensate vibrations. Results show that the
majority of measurement for cruise and hover were
below 2 m/s2 where the average was around 1.5 m/s2.
During transition phases, vibrations increased to 3
m/s2. There are fluctuations in the measurement,
which are likely caused by hand movement.
All measurements were imported to IBM SPSS to
test the groups for statistical significance. ANOVA
revealed for all cases that the levels of vibration
(cruise, hover and transition) are significantly different
from each other. An ANOVA for mobile measurement
was not performed because of few and intermittent
296 data sets (comprised of 14,504 data points)
were imported from the app. Each task received a key
value describing the placement, vibration and target
size. The key value consists of four digits (see Figure
11). The first digit describes the placement (1-fixed, 2-
mobile), the second digit describes the vibration (1-
cruise, 2-transition, 3-hover) and the last two digits
describe the target size. For example, 1115 means that
the task was performed with a fixed placement, during
cruise and the target size was 15 mm.
Data received their key value by using the flight
protocol. These values were double-checked with
vibration measurements and video recordings. Tables
1 through 5 present the mean and standard deviation
on task error rate in percent versus several different
conditioning factors. A probability value (p) of 0.05
was chosen as a cut-off level for statistical
Statistical analyses were performed using IBM
SPSS. The present analyses starts at the top level
where all independent variables were considered
separately. In the next level, the data was examined for
significant interaction (multiple effects) between
independent variables. The last step evaluated
significant differences between each condition.
Figure 11. Independent Variables. I-III
correspond to different levels of analysis.
With the aim of establishing a baseline and
familiarizing participants with the task, data for the
mobile condition was collected on the ground in the
briefing room. Data for the fixed placement was
generated afterwards in the lab by fixing the tablet at
the same distance as it was in the aircraft. An
independent t-test applied to the baseline data revealed
that both conditions had the same mean error and
standard deviation (M=0.07; SD=0.30), thus no
significant difference was found. However, the same
method was applied to the data generated in the air
revealed significant differences. Levene’s test rejected
the assumption of equality of variances. The scores for
the fixed placement were significantly higher than for
the mobile condition (see Table 1).
There was a significant effect of vibration on
error rates at the p<.05 level. Least significant error
(LSD) and Bonferroni post-hoc test compared effects
pairwise. Results showed that all combinations are
significantly different from each other (see Table 2).
There was a significant effect of target size on
error rates at the p<0.05 level. LSD and Bonferroni
found a significant difference for pairwise
combinations apart from the combination of target
sizes 15 mm and 20 mm (see Table 3).
Table 1. T-Test for Placements.
t(13407)=6.74; p = <0.01 (two tailed)
Table 2. ANOVA for Vibrations.
Table 3. ANOVA for Target Sizes.
Univariate Analysis of Variance (Level II)
A univariate analysis of variance revealed
significant interaction effects between placement and
target size and also vibration and target size. There was
no significant interaction between placement and
vibration (Table 4). This suggests that the impact of
placement and vibration depends on the size of the
Figure 12 shows the error rates by vibration and
placement. It is noticeable that participants made
fewer errors when the device was mobile.
Table 4. Uni. ANOVA for Independent Variables.
Placement & Target Size
Vibration & Target Size
Placement & Vibration
Figure 12. Mean Errors for Fixed vs. Mobile
Placement by Vibration (including the Baseline).
All Conditions ANOVA (Level III)
In the following Figure 13 and Figure 14, error
rates for each placement condition are plotted by target
size. Mean Errors and their standard deviations for all
conditions are shown in Table 5.
The largest difference in error rates occurred in
the mobile condition for a 5 mm target size. The
difference between cruise and transition was 20% (for
the fixed placement this value is 19%). This margin
decreases for all vibrations with increasing target size.
The largest difference for placement was also
found at 5 mm target. The difference for all vibrations
were around 12-13%. Like before, increasing the
target size reduces the effect of the placement.
LSD and Bonferonni post-hoc analyses compared
all conditions pairwise for significant difference. The
results are visualized in a 24x24 matrix on Figure 15.
Figure 13. Errors by Target Size for the Fixed
Figure 14. Errors by Target Size for the Mobile
Table 5. M and SD for all conditions
As shown in Figure 15, 5 mm target sizes were
significantly different to all other target sizes.
However, there were a few pairs which were not
significantly different (1305/2205, 1305/2305 and
2105/2305); amounting to 2% of the comparisons in
which 5 mm targets were involved.
Comparing 10 mm targets with the same level
and larger target sizes reveal more cases that are not
significantly different. 24% of the pairwise
comparisons in which 10 mm targets were involved
showed no significant difference.
The first level of analysis with all factors
considered independently showed no significant
difference for 15 mm and 20 mm targets. Considering
all conditions separately as shown in Figure 15
showed that the error rate for 15 mm targets during the
transition phase with a fixed placement (1215) differed
significantly from 15 and 20 mm targets during cruise
for both conditions (1115, 1120, 2115 and 2120). 58%
of the comparisons in which 15 mm targets were
involved showed no significant difference.
Comparing conditions that have 20 mm targets
involved did not show any significant difference.
Usage and Handling
Interaction in the fixed placement condition was
performed with one hand. Participants always used
their preferred hand. They were encouraged to take a
break when feeling fatigue in their arms. Eight
participants were observed to tend to hold on to the
device from the side or above. To avoid bias
participants were asked not to hold on to the device.
However, the observation suggests that people tend to
hold on to the screen to stabilize their hands. This
could be factored in when designing the hardware as
well as the software interface. For example, the
display could be designed in such a way that it enables
pilots to stabilize their hands from all directions (from
behind included) and interactive elements should be
placed along the sides.
In the mobile placement condition, six
participants initially used both of their hands to hold
the device, and used their thumb to tap the task (see
Figure 16b). Eight participants held the device with
Figure 15. ANOVA for All Conditions
Figure 16. Tablet Hold Strategies used in the
Figure 17. Recommended Interactions Areas for
Two Hands Holding, Thumbs Interaction .
their non-dominant hand and performed the
experiments with their preferred hand’s index finger
(see Figure 16a). In two cases, participants switched
from two-handed thumb to one handed index finger
It was observed that participants that used both
hands had difficulties touching the target at the centre
of the tablet. Participants had to readjust their grip
frequently. This is a known drawback of this hold
strategy. Figure 17 shows recommended interaction
areas for two-handed holding. Post interviews
revealed that participants prefer to use the tablet
device in the mobile condition. In contrast, the fixed
placement was described as more fatiguing.
It was expected that vibrations measured in the
fixed condition would be more intense than those on
the main instrument panel, which is installed on a
system, which absorbs a certain amount of vibrations.
By contrast, in the fixed placement condition the
smartphone and tablet were attached to the window via
a suction cup fixture, which transferred the entire
airframe vibration to the devices without absorption.
Interviews with pilots showed that there are
times, especially during winter months, in which they
have to operate in challenging weather conditions. In
these times, pilots are exposed to higher vibrations and
turbulences. Thus, experiments conducted with higher
vibrations resulting from the fixed placement may be
considered to emulate a certain amount of realism.
The analysis of vibration measurements gathered
in the mobile condition showed that the human body
is able to absorb a certain amount of vibration. The
peak value was measured as expected during transition
phases. In other flight modes, which cover the
majority of the flight, vibrations did not increase
beyond 3 m/s2.
Observations showed that pilots performed more
‘manual’ actions during hover compared to cruise.
During hover, the wind is pushing the aircraft away
from its position and the pilot has to steer manually to
keep the aircraft at the desired position. This causes
additional unexpected movements in the aircraft.
Another factor, which could impede the accuracy, is
the downwash wind that blows into the door during
New cockpit designs have fixed as well as mobile
touchscreens integrated. Pilots have to extend their
arms towards the dashboard to interact with the
aircraft systems. The study presented here confirms
that without support this increases the likelihood to
make more errors in a vibrating environment.
In the mobile setting the user was able to pull the
device inside his “zone of convenient reach ”,
causing the device to vibrate similarly to the human
body, ‘absorbing’ a certain amount of vibration, which
is not the case in the fixed condition. Results
confirmed the hypotheses that participant were likely
to make more errors in the fixed condition than in the
Independent variables were tested systematically,
starting broadly at the top level and gradually going
into more detail. In the first set of analysis significant
difference for all variables were found. Only target
sizes between 15 mm and 20 mm were not
significantly different. Detailed analyses showed that
there are few cases where significant difference
between 15 and 20 mm exist.
In the second level of analysis, interactions
between independent variables were calculated, which
showed that, two of three possible combinations have
significant interaction effects.
The last level of analysis considered each
possible case (24) separately and in pairwise
comparisons. The provided matrix shows that the
effects of placement and vibration disappear with
increasing target size.
Target sizes beyond 20 mm were not tested,
however helicopters are able to absorb higher
vibrations. Keeping previous works in mind it is
unlikely that targets bigger than 20 mm would lead to
significant improvement. Therefore, it is
recommended to use 20 mm targets for fixed devices
for which pilots have to extend their arms to reach, and
for safety critical tasks. The expected error rate for 20
mm targets during transition phase with a fixed
placement (worst case) is 3 %.
Airlines are increasingly interested in the
integration of portable touchscreen devices into the
cockpit. In 2011, FAA has authorized use of the Apple
iPad as EFB . Currently, many Airlines are in the
transition phase to a paperless cockpit. American
Airlines (AA) was the first major commercial carrier
that completed their EFB program. The software, used
by AA, has the following features :
Enroute charts and airport diagrams
(Displays own-ship position)
Arrival, departure and approach procedures
Change notifications (terminal and enroute)
As seen above, mobile devices are (currently) not
used for safety critical task. Thus, 15 mm targets for
mobile devices may be sufficiently large. The
expected error rate for 15 mm targets during transition
when the device is held rather than fixed is 3%.
As mentioned in the literature review an
acceptable error rate for this application area has not
been established. However, it is expected that
authorities will establish guidance for acceptable error
rates for different tasks (safety critical and non-safety
critical tasks). If designers require a higher accuracy,
it is not recommended to increase the target size
beyond the recommended values. Instead, adding an
additional safety layer with message box saying: “Do
you want to proceed?” would make the interface more
error proof (redundant).
To give another example, “shutting down
engines” may be classified as a safety critical task,
accidental shutting down must be avoided. The
interaction may be designed to minimise the error
probability in the following way. To shut the engines
off, the pilot would need to navigate to a menu item,
select and touch the ‘off’ button, upon which the
system would prompt the pilot to confirm if they want
to shut down the engines. In total, the pilot would have
to take three steps within the system to shut down the
engine. If we assume all interactive elements have the
recommended size, the error rate is at worst 3% per
layer. Adding three layers will reduce the probability
of shutting down the engines by accident to 0.0027%
(0.03x0.03x0.03=0.000027). However, alternatively
certain safety-critical actions may only be supported
by traditional physical switches.
The scope of this paper covered error rates,
vibration analyses and usage. During the experiments
additional data was recorded, which will enable
further analyses. The approach differed significantly
from the recommended ISO standard, however
movement analyses and throughput calculations could
give us a better understanding of the impact of various
It was expected that there is a significant
difference between the mobile and the fixed placement
conditions. One question for future work is how does
the distance between user and display impact the
performance? The ISO standard could be used to
determine optimal display position within the cockpit.
As mentioned during the introduction, each
application area has its own special requirements.
Another effect, which could degrade the accuracy, is
the G-Force that occurs during steep turns. This is
another issue, which particularly fighter pilots may
have to face. An initial lab trial could show whether
additional G-Force has a significant effect.
This study investigated the effects of vibrations
on accuracy of task performance using touchscreen
devices on the flight deck. It was confirmed
statistically that all flight modes are different in
character. The potential impact of vibration, touch
target size and placement was evaluated. All factors
were found to have a significant impact. As shown in
previous work the target size is the most significant
factor, which may be utilised to minimise other
degrading factors by selecting an appropriate target
size. It was demonstrated that using touch-enabled
devices that are fixed in place in vibrating
environments produces significantly higher error rates
than when the device can be held by the user.
S. Dodd, J. Lancaster, A. Miranda, S. Grothe,
B. DeMers and B. Rogers, “Touch screens on
the flight deck: the impact of touch target size,
touch technology and turbulance on pilot
performance,” in Proceedings of the Human
Factors and Ergonomics Society 58th Annual
Meeting, Chicago, 2014.
S. Bonelli and L. Napoletano, “Flying with
complexity; bringing touch screens into the
cockpit,” Alicia, Rom, 2013.
D. Barstow, “The aviation iPad revolution,”
Aviation Management Association,
American Airlines, “American airlines
completes electronic flight bag
implementation,” 24 6 2013. [Online].
implementation. [Accessed 13 12 2013].
Honeywell, “Honeywell brings modern touch
to gulfstream cockpit,” Honeywell, 31 10
2014. [Online]. Available:
cockpit. [Accessed 1 6 2015].
Thales, “Thales unveils avionics 2020 for
helicopters,” Thales, 25 02 2014. [Online].
avionics-2020-helicopters. [Accessed 6 6
Civil Aviation Authority (CAA), “CAP 708 -
Aviation safety review,” TSO, West Sussex,
A. Degani, E. Palmer and K. Bauersfeld, “Soft
controls for hard displays: still a challenge,” in
36th Annual meeting of the human factors
society, Santa Monica, 1992.
S. Kaminani, “Human computer interaction
issues with touch screen interfaces in the flight
deck,” in Digital Avionics Systems Conference
(DASC), Seattle, 2011.
Apple Inc., “iOS Human Interaface Guidlines
(iOS 8),” Cupertino, 2014.
Google, “Android Interface Design Guidlines
(Android 5.0),” 2014.
Microsoft, “Touch interactions for Windows,”
N. Henze, E. Rukzio and S. Boll, “100,000,000
Taps: Analyses and inprovement of touch
performance in the large,” in MobileHCI,
B. Schildbach and E. Rukzio, “Investigating
selection and reading performance on a mobile
phone while walking,” in HCI Mobile, Lisbon,
H. A. Colle and K. J. Hiszem, “Standing at a
kiosk: Effects of key size and spacing on touch
screen numeric keypad performance and user
preference,” Ergonomics, vol. 47, no. 13, pp.
B. K. Perry and J. P. Hourcade, “Evaluating
one hand thumb tapping on mobile touchscreen
devices,” in Graphics Interface Conference,
A. Sears and B. Shneiderman, “Improving
touchscreen keyboards: Design issues and a
comparison with other devices,” Interacting
with Computers, vol. 3, no. 3, pp. 252-269,
S. C. Lee and S. Zhai, “The perofrmance of
touch screen soft buttons,” in CHI, Boston,
R. Leitao and P. A. Silva, “Target and spacing
sizes for smartphone user interfaces for older
adults: Design patterns based on an evaluation
with users,” in Conference on Pattern
Languages of Programs, Arizona, 2012.
M. Schedlbauer, “Effects of key size and
spacing on the completion time and accuracy
of input tasks on soft keypads using trackball
and touch input,” in Human Factors &
Ergonomics Society 51st Annual Meeting,
S. N. Tsang, A. H. Chan and K. Chen, “A study
on touch screen numeric keypads: effects of
key size and key layout.,” in International
MultiConference of Engineers and Computer
Scientists , Hong Kong, 2013.
J. Bergstrom-Lehtovirta, A. Oulasvirta and S.
Brewster, “The effects of walking speed on
target acquisition on a touchscreen interface,”
in Proceedings of the International Conference
on Human Computer Interaction with Mobile
Devices and Services. ACM, 2011.
Department of Defense, “Design Criteria
Standard (MIL-STD-1472G),” 2012.
Federal Aviaton Administrations (FAA),
“Controls for flight deck systems AC(20-
IAMSAR, International Aeronautical and
Maritime Search and Rescue (IAMSAR)
C. Vieyra and R. Vieyra, “Physics Toolbox
Apps,” Vieyra Software, 1 1 2015. [Online].
[Accessed 17 6 2015].
International Standard Organisation, “ISO
9241-9 Ergonomic requirements for office
work with visual display terminals,” ISO,
S. Pheasant, Bodyspace - anthropometry,
ergonomics and the design of the work, 2nd
ed., London: Taylor & Francis, 1996.
K. Murphy, “The Paperless Cockpit,” The New
York Times, p. B6, 4 7 2011.
C. Pschierer, T. Thompson, R. Ellerbrock and
S. Haffner, “From Captain Jeppesen's little
black book to the iPad and beyond,” in Digital
Avionics Systems Conference (DASC),
IEEE/AIAA 31st, Williamsburg, 2012.
I would like to thank the Spanish Maritime Safety
Agency and INAER for opening their facilities.
Especially, Nestor Perales Gomez who organized my
visits, flights and approved required permissions. I
would like to thank GE Aviation Systems Ltd., which
is the industrial partner of my EPSRC ICase
(EP/K504506/1) PhD Program.
Huseyin Avsar - firstname.lastname@example.org
34th Digital Avionics Systems Conference
September 13-17, 2015