THE INTERNATIONAL JOURNAL OF AVIATION PSYCHOLOGY, 17(1), 91–108

Copyright © 2007, Lawrence Erlbaum Associates, Inc.

PRACTIONER ARTICLE

HUD With a Velocity (Flight-Path)

Vector Reduces Lateral Error During
Landing in Restricted Visibility
Örjan Goteman
Scandinavian Airlines
Stockholm, Sweden

Kip Smith and Sidney Dekker

Department of Industrial Ergonomics
Linköping Institute of Technology

The operational community has assumed that using a head-up display (HUD) instead
of conventional head-down displays will increase accuracy and safety during ap-
proach and landing. The putative mechanism for this increase in safety is that previ-
ously demonstrated improvements in lateral and vertical control of the aircraft in
flight should carry over to the landing situation. Alternatively, it is possible that, dur-
ing approach and landing, the HUD might affect the pilot’s ability to assimilate out-
side cues at the decision height, thereby reducing the success ratio for landings using
an HUD. This article reports a pair of experiments that test these competing hypothe-
ses. Taking advantage of the opportunity when an air transport operator introduced
HUD in an existing aircraft fleet, we were able to use a Boeing 737–700 full-motion
simulator flown by commercial airline pilots. We explored the effects of (a) HUD use,
(b) ambient visibility, and (c) length of approach lighting on the size and location of
the touchdown footprint. We also explored the effects of HUD use on approach suc-
cess ratio. HUD use reduced the width of the touchdown footprint in all tested visibil-
ity and lighting conditions, including visibility below the minimum allowed. HUD

Correspondence should be sent to Örjan Goteman, Flight Operations Standards, STOOS, Scandina-
vian Airlines, S–195 87 Stockholm, Sweden. Email: Orjan.Goteman@sas.se
92 GOTEMAN, SMITH, DEKKER

use had no effect on the length of the touchdown footprint. We could not detect any
decrease in approach success rate for HUD approaches. Based on these empirical
data, the minimum visibility for approaches using HUDs could be set lower than for
approaches without an HUD.

This article reports two experiments designed to assess the impact of commer-
cial airline pilots’ use of a head-up display (HUD) on approach and landing per-
formance under reduced runway visual range (RVR) conditions. Our goal is to
report experimental work designed to inform the development or revision of reg-
ulations on HUD use in low-RVR conditions. We begin by reviewing previous
research on HUD use and by describing how ambient RVR affects approach and
landing operations. The review motivates a pair of hypotheses regarding the im-
pact of HUD use on the size of the landing footprint and the probability of ap-
proach success. We discuss the experiments and their findings in turn with par-
ticular emphasis on the mechanisms behind the effects and the practical
implications for approach and landing in restricted visibility.

HUD USE IN COMMERCIAL FLIGHT OPERATIONS

The HUD projects computer-generated aircraft flight-path and energy symbols

onto a transparent screen in the pilot’s primary view. Pilots can concurrently
view the HUD symbology and the surrounding world. Commercially available
HUDs replicate the information on the pilot’s conventional flight instruments,
showing aircraft attitude, speed, altitude, and heading, and containing a
flight-path symbol showing the aircraft velocity. From the pilot’s perspective,
these data project onto the outside world. HUDs with this 1:1 relation between
the displayed symbols and their real-world referents are called conformal HUDs.
Thus a conformal HUD with a flight-path symbol can explicitly show the pilot
where the aircraft is going relative to the surrounding world. In contrast, a pilot
flying with conventional flight instruments must infer the aircraft’s flight path
from a synthesis of the H-angle (Lintern & Liu, 1991), optical flow (Gibson,
1986), and possibly also the relative perspective gradient (Lintern, 2000).
Comparisons between HUDs and head-down displays in manual flight have
found that conformal HUDs use improved track, speed, and altitude maintenance
(Lauber, Bray, Harrison, Hemingway, & Scott, 1982; Martin-Emerson &
Wickens, 1997). The civil aviation community assumed that these HUD perfor-
mance advantages over conventional head-down instrumentation could reduce the
number of approach and landing incidents and accidents (Flight Safety Founda-
tion, 1991). Specifically, the commercial operational community assumed that
HUD use could mitigate two well-documented problems associated with approach
 HUD USE AND LANDING PERFORMANCE 93

and landing: visual approaches to runways without radio navigation aids or with
unreliable navigation aids, and the transition from instrument to external visual
cues for landing in low visibility (e.g., Newman, 1995).
In this article we focus on pilot performance during landing in low-RVR condi-
tions where transitioning from instrument to external cues for manoeuvring is an
issue. The presumed sources for the advantage in flight-tracking performance for
the HUD are that it eliminates the need for the pilot to move his or her gaze from
head-down instruments to the outside world to look for maneuvering cues (Stuart,
McAnally, & Meehan, 2003) and it minimizes scanning requirements (Mar-
tin-Emerson & Wickens, 1997). The transition from head down to the outside
world requires a change in visual accommodation (e.g., the visual depth of field
changes from less than 1 m to infinity). Because conformal HUD symbology is fo-
cused at infinity, HUD use eliminates the need for and time demand of visual ac-
commodation, simplifying the pilot’s task.
One possible negative effect of HUD in the landing situation is that inserting a
glass plate with symbols in front of a pilot may affect his or her ability to visually
acquire the approach lights, which is necessary to continue the approach below the
decision height. This could be either through attentional capture, also called cogni-
tive tunneling (Fischer & Haines, 1980; Weintraub, Haines, & Randle, 1985;
Wickens & Long, 1995), or merely by the fact that light transmission through the
HUD is not 100%. A commercial HUD will let about 85% to 90% of the incoming
light pass through the glass plate. A detrimental effect of HUD use during the land-
ing would then show up as a lower success ratio for HUD than for a conventional
flight deck without HUD.

APPROACH OPERATIONS IN REDUCED VISIBILITY

The majority of approaches (the segment of flight immediately before touch-

down) performed in U.S. and European civil air transport operation are con-
ducted as Category I instrument landing system (ILS) approaches. For a Cate-
gory I approach the two critical factors when making the decision of whether to
land are the decision height and the RVR. Decision height refers to the aircraft’s
vertical distance above the runway threshold where the pilot must make the de-
cision to land or make a go-around. The pilot must be able to see at least some
of the approach or runway lights at the decision height. The approach lights will
then guide the pilot to the runway.
The RVR is a measure of horizontal visibility defined by the length of visible
approach and runway lights in the ambient atmospheric conditions. If the RVR is
too low the pilot will not be able to see any of the approach lights at the decision
height and must make a go-around. Although the resulting missed approach is part
94 GOTEMAN, SMITH, DEKKER

of normal operations (and formal procedures), it adds an undesired additional risk

(International Civil Aviation Organization [ICAO], 1993).
The approach light lengths differ from runway to runway. Geographical con-
straints sometimes make the standard full length of 720 m (Full Facilities) impos-
sible to achieve. Fewer approach lights means less guidance and later contact with
the approach lights during the approach. Consequently the length of the approach
lights is reflected in existing regulations. For example, to commence a Category I
ILS approach to a runway equipped with a 720 m length of approach lights, the
RVR measured at the runway must not be less than 550 m (Federal Aviation Ad-
ministration [FAA], 2002; Joint Aviation Authorities [JAA], 2004b). This RVR is
expected to allow the pilot to see a visual segment of the ground that contains
enough of the runway approach lights to judge the aircraft’s lateral position,
cross-track velocity, and position in roll when the aircraft is at decision height.
This condition is sketched in Figure 1. The same condition viewed from the pilot’s
position in cockpit is sketched in Figure 2. The figure depicts how the visible row
of Full Facility (720 m) approach lights appear from under the aircraft’s nose and
stretch forward toward the runway until they are obscured by the limited visibility.
In Figure 3 we have sketched the pilot’s view of the same approach lights length in
lower than standard (450 m) RVR. Here the shorter ground visual segment is seen
as fewer approach lights are available for guidance toward the runway.
For shorter approach light lengths there is less guidance. In Figure 4 we have
sketched what the pilot sees at decision heights for approach lights with 420 m
length in standard (700 m) RVR. In Figure 5 the pilot view of the same approach
lights is shown in lower than standard (550 m) RVR. In all of these approach light

FIGURE 1 The visual segment as seen from the aircraft from the decision height at a given
runway visual range.
FIGURE 2 The visual segment as seen through the left cockpit window in a B–737 at decision
height when flying in over a Full Approach Lights System (at least 720 m length) in standard 550
m runway visual range.

FIGURE 3 The visual segment as seen through the left cockpit window in a B–737 at decision
height when flying in over a Full Approach Lights System (at least 720 m length) in lower than
standard 450 m runway visual range.

95
 FIGURE 4 The visual segment seen through the left cockpit window in a B–737 at decision
height when flying in over an Intermediate Approach Lights facility (450 m length) in standard
700 m runway visual range.

FIGURE 5 The visual segment seen through the left cockpit window in a B–737 at decision
height when flying in over an Intermediate Approach Lights facility (450 m length) in lower than
standard 550 m runway visual range.

96
 HUD USE AND LANDING PERFORMANCE 97

lengths–RVR combinations the pilot does not see the runway threshold at decision
height. Only later as the aircraft comes closer to the runway threshold can the pilot
see and aim the aircraft at the designated touchdown point, normally 305 m down
the runway. Apart from the obvious effect that the approach lights will come in
view later with lower RVR, other effects of shorter approach lights lengths could
also come into play. If runway length has been shown to influence the perceived
descent path (Lintern & Walker, 1991) it is also possible that a reduced length of
approach lights can have similar effects, adding a source of uncertainty to the ver-
tical control of the aircraft.

THE ISSUE OF RVR AND HUD USE

Large commercial airlines such as American Airlines, All Nippon Airways,

Delta, Scandinavian Airlines, and Southwest Airlines are introducing HUDs to
the flight deck at an increasing rate, making the effect of HUD use on touch-
down performance an operationally relevant question. Generalizing from earlier
research (Martin-Emerson & Wickens, 1997), HUD use with conformal
symbology should improve tracking performance during approach and landing
in restricted visibility. Showing the instantaneous flight path should improve the
pilot’s ability to control the aircraft velocity in relation to the real world below
decision height and this, in turn, should reduce the variance of the touchdown
point. However, the landing situation is different from cruise or approach when
the pilots can concentrate on the HUD symbology at the expense of outside
cues. During landing the pilots have to concurrently process both outside cues
and HUD cues to get any benefit from the HUD. The operational benefit from a
reduction in touchdown variability could be that regulations would allow ap-
proach operations using HUD in lower than standard RVR conditions. The cur-
rent operating minima were not set bearing HUD operations in mind and may be
too restrictive for operations using HUD, a fact that the existing legislative text
acknowledges (JAA, 2004b). However, setting RVR for approach too low will
ultimately reduce approach success rate. In low RVR with very few external
cues available at decision height, there is a risk that the pilots will focus their at-
tention on the HUD symbology to the extent that they might not perceive the
few visible approach lights at decision height. Pilots who do not pick up the out-
side cues may thereby initiate a go-around when the approach actually could
have been continued, an outcome that is not desirable from an operational stand-
point. Squeezed between operational pressure for lower minima for the HUD’s
presumed advantage and possible negative effects of the HUD, regulators are
being pressed to generate rules that are insufficiently grounded in research.
In light of the increasing use of HUDs in commercial aviation and the paucity
of the existing knowledge base, we studied whether the established improve-
98 GOTEMAN, SMITH, DEKKER

ment in tracking performance when using an HUD would be reflected in the

landing as a smaller touchdown footprint on the runway in standard and lower
than standard RVR conditions. We sought to ascertain and quantify whether the
HUD-related advantages for flight-path tracking during cruise established by
earlier research is carried over to the landing when the pilot has to concurrently
process HUD symbology and outside cues. Data exist for certification trials for
HUDs currently used in civil aviation. However, those certification trials were
aimed at showing HUD performance in a low-visibility mode with guidance
down to and including the landing flare, where there was no need for external
visual cues at all (JAA, 2004a). Such data do not export well to touchdown per-
formance using HUD in Category I flight operations where obstacle clearance
below the decision height is maintained by visual maneuvering. We know of no
other studies with commercial pilots in high-fidelity simulations that have tested
whether HUD use is associated with performance advantages related to landing
in restricted visibility operations predicated on visual cues for maneuvering. We
were given the opportunity when a European Air transport carrier installed
HUDs in an existing fleet of Boeing 737 (B–737–NG) aircraft. In connection
with the normal training program we were able to compare pilot performance
during approach and landing when using an HUD and when using conventional
head-down instrumentation.
As there is less guidance with shorter approach lights we investigated the effect
of HUD on touchdown performance for two different approach lights conditions
as defined in the European regulations: Full Approach Light facilities (≥ 720 m)
and Intermediate Approach Light facilities (420 to 719 m; JAA, 2004b). See Table
1 for the RVR and approach light lengths in the two experiments. In the first exper-
iment, we assessed HUD use in visibilities at and below the required minimum
RVR at simulated facilities with a system of approach lights of full length (900 m).
In the second experiment, we assessed HUD use with visibilities at and below the
required minimum RVR at simulated facilities with a system of approach lights of
intermediate length (420 m). We predicted that the HUD would reduce the size of

TABLE 1
RVR Standards for Facilities With Full and Intermediate Systems of Approach Lights

Experiment 1 With Full Facilitiesa Experiment 2 With Intermediate Facilitiesb

Low RVR Standard RVR Low RVR Standard RVR

450 m 550 m 550 to 600 m 700 m

Note. RVR = runway visual range.

a900 m. b420 m.
 HUD USE AND LANDING PERFORMANCE 99

the footprint both longitudinally and laterally and that an HUD in itself would not
reduce the approach success rate at the RVRs used in the experiments.

EXPERIMENT 1

Method

Forty-eight pilots from a major European airline volunteered to participate. All

were qualified to fly the B–737–700 aircraft and had completed their HUD train-
ing for the operator. The HUD training sessions consisted of 1 day of theory and
two simulator sessions of 4 hr duration each. Experience on the B–737–700 var-
ied from 50 hr to more than 1,000 hr. There is every reason to believe that these
participants are representative of the population of commercial pilots to whom
regulators need the data to generalize.

Apparatus. We used a CAE B–737–700 training simulator with aircraft

aerodynamics and visual angles valid for B–737–700 to collect data. This six-axis
full-motion simulator is approved for low-visibility operations down to an RVR of
200 m. The simulator’s visual system had a field of vision of 180°/40° with a focal
distance greater than 10 m.
The head-down instrumentation was a B–737–700 instrument panel in pri-
mary flight display (PFD) configuration with flight director guidance. This in-
strumentation was available in both the with-HUD and the no-HUD conditions.
The HUD installed in the simulator was a Rockwell-Collins Flight Dynamics
HGS–4000® (Head-up Guidance System), certified for low-visibility operations
down to and including an RVR of 200 m. The HUD symbology and functional-
ity used in the experiment met the production-line standard specification for the
instrument meterological conditions (IMC) mode used when conducting Cate-
gory I ILS and nonprecision approaches (see Figure 6). This mode was deliber-
ately chosen to improve the external validity on the form of operational usability
of the study. In normal operations, the vast majority of ILSs are only approved
for Category I operations. The HUD provided conformal display of flight path
(velocity) and flight-path guidance. The flight path was displayed as a circle
with slanted wings. Flight-path guidance was displayed in the form of a ring in-
side the flight-path symbol. Flight-path guidance was not available in the con-
ventional head-down instrumentation.
The radio navigation facilities simulated in the study conformed to the ICAO
(1996) standard for ILS radio navigation aids for Category I approaches transmit-
ting a radio beam for both vertical and lateral reference. The integrity of the trans-
mitted beam is guaranteed to keep the aircraft within allowable airspace, safe from
100 GOTEMAN, SMITH, DEKKER

FIGURE 6 The HUD symbology of the instrument meterological conditions (IMC) mode
used during the approaches in Experiments 1 and 2. Courtesy of Rockwell-Collins Flight Dy-
namics. Reprinted with permission.

obstacles down to 200 ft height, corresponding to lowest allowable decision

height. Approaches in Experiment 1 were flown to a simulated runway with a sys-
tem of approach lights 900 m in length. This length falls within the full facilities
system category of European aviation regulation (JAA, 2004b).

Design. The design used was a two-factor experiment with repeated mea-
sures on HUD use (HUD, no-HUD). The between-subject factor was RVR at two
levels, the standard minimum RVR (550 m) for full facilities system and lower than
standard RVR (450 m). Each participant flew one approach with HUD and one ap-
proach without HUD to a runway with a full system (900 m) of approach lights. The
order of HUD use was counterbalanced across participants to control for potential
order and carryover effects. This design enables us to assess whether and HUD per-
formance advantages in standard RVR carry over into lower than standard RVR
conditions.

Procedure. Each pilot manually flew two approaches using standard operat-
ing procedures of the airline. The scenarios started as a 6 nm final to the runway in
lower than standard or standard RVR in a simulated 10 kt left crosswind. In the
with-HUD condition, the pilots kept the aircraft on lateral and vertical by following
the flight-path guidance ring with the flight-path symbol on the HUD. At 50 ft
above the runway threshold, the guidance cue was automatically removed and the
pilots performed the landing flare using external visual cues in conjunction with the
HUD flight-path symbol. The HGS–4000® IMC mode incorporating automatic re-
 HUD USE AND LANDING PERFORMANCE 101

moval of the guidance cue was deliberately chosen to ensure that the pilots could
not attend solely to the HUD symbology in the with-HUD conditions.
In the no-HUD condition the pilots kept the aircraft on lateral and vertical track
by following the flight director bar guidance. At decision height they continued the
approach and landing using the external cues only.
All approaches were recorded to determine approach success. Approach and
landing plots for approaches ending with a landing were printed using the air-
craft’s center of gravity as the reference to determine the size of the touchdown
footprint.

Results

Approach success rate. Each of the 48 pilots attempted two approaches.

Thirty-seven made two successful landings, one with HUD and one without. Three
made go-arounds in both conditions. Two pilots landed with the HUD and made
go-arounds without the HUD; six landed without the HUD and made go-arounds
with the HUD.
The McNemar change test is the appropriate statistical procedure for testing the
null hypothesis that pilots were equally likely to (a) land with the HUD and
go-around without it, and (b) land without the HUD and go-around with it (Siegel
& Castellan, 1988). Because the observed test statistic, calculated from the values
given earlier (2 and 6), is 1.125 and is less than the criterion, χ2(.05, 1) = 3.84, we
cannot reject the null hypothesis. Accordingly we infer that HUD use had no im-
pact on approach success rate.

Touchdown performance. As noted previously, 11 of the 48 pilots con-

ducted one or two go-arounds. The simulator failed to capture the location of the
landing footprint for another 7 pilots. As a result the data set for comparing the
touchdown footprints across the HUD and no-HUD conditions consists of 30 pairs
of approaches. Of these 30, 15 were flown using the HUD in the first approach and
15 using the conventional head-down instruments (no-HUD) in the first approach;
15 were flown in standard RVR (550 m) conditions and 15 in lower than standard
RVR conditions (450 m).
Lateral touchdown performance was measured as the absolute lateral deviation
from the runway centerline at touchdown. The data for the main effect of HUD use
are shown in Figure 7. Landings were closer to the centerline when pilots used the
HUD. The two-factors repeated-measures analysis of variance (ANOVA) re-
vealed a strong effect for HUD use on the lateral component of the touchdown
footprint, F(1, 28) = 9.05, p < .006, η2 = .12 indicating a power of .80 at α = .05.
The main effects for RVR and order of HUD condition were not significant. There
was, however, a marginally significant interaction between HUD use and the order
102 GOTEMAN, SMITH, DEKKER

FIGURE 7 Graph showing the

main effect for HUD use and the
order of its use on the lateral size
of the touchdown footprint at sim-
ulated facilities with a full system
(900 m) of approach lights, Exper-
iment 1.

FIGURE 8 Graph showing the

interaction of HUD use and the or-
der of its use on the lateral size of
the touchdown footprint at simu-
lated facilities with a full system
(900 m) of approach lights, Exper-
iment 1.

of HUD use, F(1, 28) = 3.31, p < .08. As shown in Figure 8, both the group that
used the HUD on their first approach (HUD first) and the group that used the HUD
on their second approach (no HUD first) landed nearer the centerline when they
used the HUD. Thus there is a strong hint of asymmetric transfer in HUD use on
the lateral size of the footprint. Pilots appear to benefit from previous exposure to
the HUD.
Tests for the equality of variance across conditions of HUD use require the
samples to be independent, a condition violated by the repeated-measures design.
Were the samples independent, the difference in the observed variances in the lat-
eral touchdown footprint would be significant at α = .10 with lower variability
when landing with the HUD.
Longitudinal touchdown performance was measured as the distance from the
runway threshold at touchdown. Contrary to expectations and in contrast with the
 HUD USE AND LANDING PERFORMANCE 103

results for lateral performance, the analysis of variance revealed that the effect of
HUD use was not significant. Neither the between-subject main effect of RVR nor
its interaction with HUD use was significant. Thus there is no indication of asym-
metric transfer in HUD use on the size of the longitudinal footprint.

EXPERIMENT 2

Method

The method, procedure, and design used in the second experiment were identical
to those used in the first experiment with the few exceptions discussed here. The
different criteria for standard RVR across facility types preclude collapsing and
analyzing the data as a single experiment.

Participants. Forty-five pilots from the same major European airline volun-
teered to participate. None of the volunteers had participated in Experiment 1. All
were qualified to fly the B–737–700 aircraft and had completed their HUD training
for the operator. Experience on the B–737–700 varied from more than 50 hr to more
than 1,000 hr.

Simulated ground facilities. Approaches in Experiment 2 were flown to

simulations of a runway with 420 m of approach lights. This length falls within the
intermediate facilities category of approach lights as defined by European aviation
regulations (JAA, 2004b).

Design. Once again, the design was a two-factor experiment with repeated
measures on one factor, HUD use. Each participant flew one approach with HUD
and one approach without HUD to a runway with a system of approach lights of 420
m length. The between-subject variable was RVR at two levels, a standard mini-
mum RVR (700 m) for intermediate facilities and a lower than standard minimum
RVR (550 m to 600 m).

Results

Approach success rate. Each of the 45 pilots attempted two approaches.

Thirty-two made two successful landings, one with HUD and one without. Four pi-
lots made go-arounds in both conditions. Three pilots landed with the HUD and
made go-arounds without the HUD; six pilots landed without the HUD and made
go-arounds with the HUD. Once again we used the McNemar change test to test the
null hypothesis that pilots were equally likely to (a) land with the HUD and
go-around without it, and (b) land without the HUD and go-around with it. Because
104 GOTEMAN, SMITH, DEKKER

the observed test statistic, calculated from the values given previously (3 and 6) is
0.44 and is less than the criterion, χ2(.05, 1) = 3.84, we infer that HUD use had no
impact on approach success rate.

Touchdown performance. As noted earlier, 13 of the 45 pilots conducted

one or two go-arounds. The simulator failed to capture the location of the landing
footprint for another 4 pilots. As a result, the data set for comparing the touchdown
footprints across the HUD and no-HUD conditions consists of 28 pairs of ap-
proaches. Of these 28 pairs of approaches, 9 were flown using the HUD in the first
approach and 19 using the conventional head-down instruments (no-HUD) in the
first approach; 14 were flown in standard RVR (700 m) conditions and 14 in lower
than standard RVR conditions. The opportunistic nature of data collection pre-
cluded balancing the order of HUD use.
The main effect for HUD use, shown in Figure 9, is the only factor in the
two-factor repeated-measures ANOVA to achieve statistical significance. As in
Experiment 1, the effect for HUD use on the lateral component of the touch-
down footprint is strong, F(1, 26) = 14.9, p < .001, η2 = .10, indicating a power
greater than .70. Once again, landings were closer to the centerline when pilots
used the HUD. Because participant order was not fully counterbalanced, it is not
possible to assess the potential for asymmetric transfer effects. As in Experiment
1, there was no significant effect for HUD use or RVR on the longitudinal com-
ponent of the touchdown footprint. Once again, the difference in the observed
variances across conditions of HUD use would be significant at α = .10 if the
data sets were independent.

FIGURE 9 Graph showing the

main effect of HUD use on the lat-
eral size of the touchdown foot-
print at simulated facilities with an
intermediate system (420 m) of
approach lights, Experiment 2.
 HUD USE AND LANDING PERFORMANCE 105

DISCUSSION

There are three findings. First, HUD use per se did not influence the pilots’ deci-
sion to land or go-around at the decision height. The lack of an effect of HUD
suggests that the additional information in the HUD did not distract the pilots’
attention or interfere with their decision making during the most critical portion
of the approach and landing sequence. Second, HUD use significantly reduced
the size of the lateral component of the touchdown footprint for all RVR condi-
tions. We cannot ascertain exactly where this effect resided. As the control con-
dition was the standard instrumentation of the B–737–NG, we can only say that
inserting an HUD with the properties that the tested HUD had, will give this ef-
fect. Arguably it can be said that the difference between the HUD and the
no-HUD conditions lay in the presence of a conformal flight-path vector in the
pilots’ primary field of view during the landing. Third, in contrast to its effect on
the lateral component of the touchdown footprint, it appears that the HUD did
not influence the size of the longitudinal footprint. The first two findings con-
form to our hypotheses. Here we reexamine our hypotheses about the impact of
HUD use on the touchdown footprint and offer an explanation for its differential
impact on the lateral and longitudinal components.
Experience engenders automaticity. For commercial pilots, hours of practice
and numerous repetitions make touchdowns at standard facilities in good weather
relatively routine. Nevertheless, the ubiquitous and ever-varying direction and ve-
locity of wind is likely to preclude the development of true automaticity at touch-
down. Crosswinds introduce an element of uncertainty regarding drift (the shift in
lateral location of the aircraft relative to the runway’s centerline). For the pilot to
detect drift the visual ground segment needs to be long enough to determine the
aircraft’s movement over the ground. That means that to detect drift at all, a notice-
able lateral displacement must take place and not all of this displacement can be
corrected before touchdown. Such exogenous uncertainty is likely to have condi-
tioned pilots (and regulators) to tolerate a certain amount of variance in the lateral
touchdown footprint.
The HUD largely eliminates uncertainty about drift. The addition of a
conformal flight-path vector projected over the runway provides instantaneous
feedback about aircraft drift and actual flight path. The additional information en-
ables precise control of the flight path during approach and landing and reduces the
variance in lateral displacement practically to nil.
Control of the longitudinal component of the touchdown footprint is largely
an effect of how pilots handle the aircraft’s energy (operationally manifested as
sink rate) in the final seconds before landing. The pilot uses information pro-
vided by the optic flow from the looming runway to control the aircraft’s energy
(Lee, 1974). It is important to note that pilots of large commercial air transport
aircraft are also aided by radio altimeter callouts that count down from 50 ft to 0
106 GOTEMAN, SMITH, DEKKER

ft (runway contact) in 10-ft decrements. The initiation of the landing flare has
been shown to be a function of time to contact (Mulder, Pleijsant, van der Vaart,
& van Wieringen, 2000). A small change in the timing of a landing flare at the
nominal glide slope of 3° results in large longitudinal differences. The HUD
mode used in the experiments provided no flare guidance and no additional in-
formation that could be used to guide the pilots when to initiate the landing
flare. So, if the pilots are relying more on timing of a flare maneuver than on ve-
locity cues from the HUD to initiate the flare, it is easy to understand why we
failed to detect any effect of the tested HUD with a flight-path symbol on longi-
tudinal touchdown performance. It remains to be seen whether similar results are
found for HUD modes with flare guidance, and which visual representations are
actually most effective in prompting pilots to reduce sink rate at the optimal
height above the runway.

APPLICATION

The effect of the tested HUD use on landing performance was to reduce the lat-
eral size of the touchdown footprint. This effect was seen in both standard and
lower than standard RVR conditions. Left crosswind is known by operators to
produce landings on the leeward side. This leeward displacement may not be
critical when visibility is restricted by fog but can be crucial when visibility is
restricted by rain or snow in strong crosswind conditions. Carriers that operate
in higher latitudes (with more snow and inclement weather) are aware of the
danger of misaligning the aircraft’s velocity with the runway, especially during
night when a combination of sideways-drifting precipitation and landing lights
may produce an illusion of sideways movement. Some operators specifically
recommend their pilots to switch the landing light off in such situations (SAS,
2005). The message that an aircraft equipped with a HUD showing a flight-path
vector is likely to reduce the tendency of drifting with the wind during landing
compared to the standard instrumentation condition is important for carriers that
routinely operate in these weather conditions.
The absence of a detectable influence on the approach success rate in the HUD
condition is also operationally important. European and U.S. authorities are cur-
rently working on a set of harmonized aerodrome operating minima. These min-
ima are expected to allow a lower than standard minimum RVR to commence
approach for HUD-equipped aircraft (JAA, 2004c). The effects of HUD use on
landing performance were seen in standard and lower than standard RVR condi-
tions in both Experiments 1 and 2. This study carries a clear and important opera-
tional implication: Based on empirical performance data, the minimum RVR for
approaches using HUD with flight-path symbology could be set lower than for ap-
proaches without an HUD while still maintaining a similar or even smaller touch-
down footprint.
 HUD USE AND LANDING PERFORMANCE 107

CONCLUSIONS

Data from the experiments reported here show that HUD with a conformal ve-
locity symbol (flight path) improved lateral touchdown performance, likely be-
cause the conformal flight-path vector in conjunction with the visual ground
segment makes it easier for pilots to determine and correct for aircraft drift. We
did not find an effect of HUD use on longitudinal touchdown performance,
probably because the pilots flared the aircraft using a time-to-contact strategy,
rather than using the flight-path vector available in the HUD modes studied
here. The beneficial effects of HUD use on landing performance were seen in
both in standard and lower than standard RVR conditions in both experiments,
which implies that the minimum RVR for approaches using an HUD could be
set lower than for approaches without an HUD. Finally, we did not find any det-
rimental effect of HUD use on approach success rate.
These results could be critical in supporting regulatory decision making about
the visibility required to commence an approach using an HUD-equipped aircraft.
In fact, it is precisely this type of systematic empirical data that is often lacking as a
basis for developing well-informed regulations, in Europe as well as the United
States. We believe that empirical results like these should be a critical ingredient in
safety regulators’ rule writing for (the application of) new technology in commer-
cial aircraft operations. When it comes to HUDs, the need for informed regulation
of HUD use is already pressing, and will only increase as HUD use proliferates.
The relevance of the findings here could even support the expansion of such tar-
geted empirical studies, for example, to Category II conditions, or to refine them to
find, for example, the lowest acceptable RVR using HUDs.

ACKNOWLEDGMENTS

We thank Staffan Carlsson and Zacharias Soudah for assistance in data collec-
tion, Jan-Olof Ehk for statistical advice, and the pilots who volunteered their
time and skill. Rockwell-Collins Flight Dynamics Company provided simulator
time to conduct the experiment.

REFERENCES

Federal Aviation Administration. (2002). Criteria for approval of Category 1 and Category 2 Weather
Minima for Approach, AC 120–29A. Washington, DC: U.S. Department of Transportation.
Fischer, E., & Haines, R. F. (1980). Cognitive issues in head-up displays (NASA Tech. Paper No. 1711).
Moffett Field, CA: NASA Ames Research Center.
Flight Safety Foundation. (1991). Special safety report: Head-up guidance technology (HGST). A pow-
erful tool for accident prevention. Flight Safety Digest (FSF/SP–91/91). Alexandria, VA: Author.
Gibson, J. J. (1986). The ecological approach to visual perception. Hillsdale, NJ: Lawrence Erlbaum
Associates, Inc.
108 GOTEMAN, SMITH, DEKKER

International Civil Aviation Organization. (1993). OPS—Aircraft operation (Doc 8168) (Vol. 2, 4th
ed.). Montreal, Quebec, Canada: Author.
International Civil Aviation Organization. (1996). Annex 10, Aeronautical telecommunication: Vol. 1.
Radio navigation (5th ed.). Montreal, Quebec, Canada: Author.
Joint Aviation Authorities. (2004a). All weather operations JAR–AWO. Hoofddorp, The Netherlands:
Author.
Joint Aviation Authorities. (2004b). JAR–OPS 1 Subpart E—All weather operations. Hoofddorp, The
Netherlands: Author.
Joint Aviation Authorities. (2004c). NPA–OPS 41 incorporating HUD, EVS and CRD. Hoofddorp, The
Netherlands: Author.
Lauber, J. K., Bray, R. S., Harrison, R. L., Hemingway, J. C., & Scott, B. C. (1982). An operational eval-
uation of head-up displays for civil transport aircraft (NASA Tech. Paper No. 1815). Moffett Field,
CA: NASA Ames Research Center.
Lee, D. (1974). A theory of visual control of braking based on information about time to collision. Per-
ception, 5, 437–459.
Lintern, G. (2000). An analysis of slant for guidance of landing approaches. International Journal of
Aviation Psychology, 10, 363–376.
Lintern, G., & Liu, Y. (1991). Explicit and implicit horizons for simulated landing approaches. Human
Factors, 33, 401–417.
Lintern, G., & Walker, M. B. (1991). Scene content and runway breadth effects on simulated landing ap-
proaches. International Journal of Aviation Psychology, 1, 117–132.
Martin-Emerson, R., & Wickens, C. D. (1997). Superimposition, symbology, visual attention and the
head-up display. Human Factors, 39, 581–601.
Mulder, M., Pleijsant, J.-M., van der Vaart, H., & van Wieringen, P. (2000). The effects of pictorial de-
tail on the timing of the landing flare: Results of a visual simulation experiment. International Jour-
nal of Aviation Psychology, 10, 291–315.
Newman, R. L. (1995). Head-up displays: Designing the way ahead. Aldershot, England: Ashgate.
Scandinavian Airlines. (2005). SAS operations manual: Part A. Stockholm, Sweden: Author.
Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the behavioral sciences. New York:
McGraw-Hill.
Stuart, S. W., McAnally, K. I., & Meehan, J. W. (2003). Head-up displays and visual attention: Inte-
grating data and theory. Human Factors and Aerospace Safety, 1, 103–124.
Weintraub, D. J., Haines, R. F., & Randle, R. J. (1985). Head-up display (HUD) utility: 2. Runway to
HUD transitions monitoring eye-focus and decision times. In R. W. Swezey (Ed.), Proceedings of
the Human Factors Society 29th annual meeting (pp. 615–619). Santa Monica, CA: Human Factors
Society.
Wickens, C. D., & Long, J. (1995). Object versus space-based models of visual attention: Implications
for the design of head-up displays. Journal of Experimental Psychology: Applied, 1, 179–193.

Manuscript First Received: September 2005