Proceedings of the
X V E U R O P E A N
ANNUAL C O N F E R E N C E ON
HUMAN DECISION MAKING
AND MANUAL CONTROL
held at the
TNO Human Factors Research Institute
Soesterberg, The Netherlands
June 10-12, 1996
organized by
Proceedings of the
X V E U R O P E A N
ANN UAL C O N F E R E N C E ON
HUMAN DECISION MAKING
AND MANUAL CONTROL
held at the
TNO Human Factors Research Institute
Soesterberg, The Netherlands
June 10-12,1996
Program Chairman : Henk G. Stassen (Ed, TUDelft)
Organization: : Peter A. Wieringa (Ed, TUDelft)
Hans Godthelp (TNO)
Alexander P. de Vos (TNO)
P. Leo Brinkman (TUDelft)
Jan B. F. van Erp (TNO)
Human Factors Research Institute Man-Machine Systems Department
TNO Faculty of Mechanical Engineering
and Marine Technology
Soesterberg Delft University of Technology
The Netherlands The Netherlands
Library DATA Proceedings
Proceedings of the X V E U R O P E A N A N N U A L C O N F E R E N C E O N H U M A N DECISION MAKING A N D M A N U A L C O N T R O L :
held at the T N O Human Factors Research Institute, Soesterberg, The Netherlands, June 10-12,1996 Peter A. Wieringa, Henk G . Stassen (Eds).
Delft: Delft University of Technology, Faculty of Mechanical Engineering and Marine Technology
ISBN 90-370-0152-1
Subject Headings: man-machine Systems / human operators / process automation / human factors
Publisher: Delft University of Technology
Faculty of Mechanical Engineering and Marine Technology Library WbMT Mekelweg 2 2628 C D Delft The Netherlands Phone: Editors: Phone: e-mail: +31 15 2786765
P.A. Wieringa, H.G. Stassen +31 15 2786400
Preface
The 15th Annuai Conference on Human Décision Making and
Manual Control again shows a large variety of interesting présentations.
Ranging from "Modelling of Individual Car-following Behaviour", via
"Cognitive Support to Human Needs", to "A global Approach to solve Social
Aspects of Usability and Acceptability, especially in Developing Countries",
the conference offers ample possibilities for discussion and information
exchange between participants.
The range of présentations more or less reflects a tendency to
concéntrate not only on the individual operators and not only on modelling,
but on the user (in general) and on applications as well.
A strong point of these conferences is that they stimulate
interdisciplinary coopération, whereas in the beginning représentatives of
the technical disciplines on the one, and those from Human Factors and
Ergonomics on the other side showed some sort of shyness in evaluating
each other's approach. Nowadays both sides know that they cannot do
without the other.
The world in the past 15 years changed faster than ever before.
Therefore, I hope, the 15th "AnnuaJ Manual" will be succesfull in
coopération, especially where it concerns the problems of tomorrow.
Prof. J. Moraal
European Annual Manual Conferences
European Annual Conference on Human Décision Making and Manual
Control, for short the European Annual Manual, was first held in 1981. The
Conference was inspired on the idea to stimulate meetings between Ph-D
students and professionals working in the field of Man-Machine Systems in
the USA. The NASA -University Annual Conference on Manual Control was
already more than 15 years organizing such meetings.
Since 1981 a series of conferences have been held in eight European
countries, namely:
1981 : The Netherlands, Delft, Delft University of Technology
1982: Fédéral Republic of Germany, Bonn, Forschungsinstitut für
Anthropotechnik
1983: Denmarkt, Roskilde,
Ris0
National Laboratories
1984: The Netherlands, Soesterberg, Institute for Perception TNO
1985: Fédéral Republic of Germany, Berlin, Technical University of Berlin
1986: United Kingdom, Wales, Cardiff, University of Wales, Institute of
Science and Technology
1987:
-1988: France, Paris, Electricité de France
1989: Denmark, Lyngby, Technocal University of Denmark
1990: Italy, Ispra, CEC Joint Research Centre
1991 : Belgium, Lièges, University of Lièges
1992: France, Valencîennes, University of Valenciennes
1993: Germany, Kassel, University of Kassel
1994: Finland, Espoo, Technical Research Centre of Finland
1995: The Netherlands, Delft, Delft University of Technology
1996: The Netherlands, Soesterberg, TNO Human Factors Research
Institute
1lt has been announced that the 1997 European Annual Manual will be held
in France, Toulouse at EURISCO.
This symposium has been sponsored by the RoHMI network; a network of the Human Capital and Mobility programme of the E C on the design of Robust Human Machine Interaction.
Table of Contents
Preface
Session 1 Manual Control and Human Operator Modelling
1.1 J-M Pleijsant, M. Mulder, H. v a n de Vaart (TUDelft, A E ) .
How do Pilots perceive Time-To-Contact from the Gmund Surf ace?
Results ofa Visual Simulation Experiment.
1.2 M. Mulder (TUDelft, A E ) .
Modelling Manual Control of Straight Trajectories with a Perspective
Flight-Path Display.
1.3 J.H. H o g e m a (TNO, HFRI).
Modelling ofindMdual Car-Following Behaviour.
Session 2: Human Operator Decision Support Systems And
Interface Design
2.1 D. Holt, U. Kaymak, J.B. Klaassens, H.R. v a n Nauta Lemke
(TUDelft, E E ) .
A Fuzzy Decision Support System for Magnetic Component Design.
2.2 W. Sillevis Smitt, B. Bruggeman
Decision Support Systems and modern Maritime Air Defence. Fuzzy
Identification of targets(Royal Neth.Naval College).
2.3 M. Tiemann (UKassel, IMAT).
Evaluation ofa Methodology forHuman-Machine Interface
Development.
2.4 A.Nendjo Ella, M. Grislin, C. Kolski (UV, LAMIH / A I C A D A ) .
Towards a Global Approach to solve Social Aspects of Usability and
Acceptability, especially in Developing Countries.
]* Paper not presented at the Conference for personal reasons. Program committee proposed to include it into the ' Proceedings
Session 3:
REMOTE CONTROL
3.1 E.F.T. Buiëi (TUDelft, ME).
The Development ofa Man-Machine Interface for Space Manipulator
Displacement Tasks.
3.2 J . Herder, M. Horward, W. Sjoerdsma (TUDelft, ME).
Mechanica! Solutions versus Electronic Teleoperation; an example.
3.3 J.B.F. van Erp, P. Padmos (TNO, HFRI).
Real and Simulated Driving with Camera View.
3.4 J.B.F. van Erp, B. Kappé, J.E. Korteling (TNO, HFRI).
Visual Support in Operating Unmanned Aerial Vehicles.
Session
4 TASK ALLOCATION AND AUTOMATION OF COMPLEX SYSTEMS
4.1 M. Neerincx (TNO, HFRI).
Mapping Cognitive Support to Human Needs.
4.2 E. Cherifi, B. Riera, E. Hais (UV, LAMIH).
Filtering Information for the Supervision of Complex Systems.
4.3 A R . Paauw, Z.G. Wei, P.A. Wieringa (TUDelft, ME).
Expehmental investigation on the effect of Task Allocation on the
Human Operator.
4.4 Z.G. Wei, A.R. Paauw, P.A. Wieringa (TUDelft, ME).
Subjective evaluation of Task Allocation: An application ofthe
Analytic Hierarchy Process.
4.5 N. Moray (UV, LAMIH).
A Method for Identifying Coupling between Humans and Machines.
Session 5
HUMAN RELIABILITY ASSESSMENT
5.1 T.W. van der Schaaf, C E . Shea, A.S. Nyssen (TUE / UL).
MECCA: Médical Errors and Complications Causal Analysis.
5.2 T.W. van der Schaaf, M. Frese, D. Heimbeck (TUE / UvA).
Human Recovery and Error Management.
Session 1
Manual Control and Human Operator
Modelling
How Do Pilots Perceive Time-to-Contact from the
Ground Surface?
Results of a Visual Simulation Experiment
J.M. Pleijsant, M. Mulder, U.C. van der Vaart
Faculty of Aerospace Engineering, Delft University of Technology
The Netherlands
P.C.W, van Wieringen
Faculty of Human Movement Sciences, Free University of Amsterdam
The Netherlands
Abstract
This paper describes an experiment on the pilot's perception process during the landing manoeuvre -or flare - of an aircraft. In particular the rôle of the so-called Time-to-Contact, TTC is considered. When approaching an obstacle, TTC is the time remaining to collision if no action were taken. In ail time-constrained tasks like car driving or braking, subjects tend to use TTC as a eue by which their actions are triggered. Previous research on manned simulator landing tasks with only a runway outline scène visible indicated that pilots indeed use some kind of TTC strategy but that the timing of the flare was also related to the height above the ground. The present experiment was designed to examine whether addition of ground texture to a simulated Visual runway scène would improve the perception of TTC, and hence the timing of the flare. The results suggest that addition of texture indeed enables a pilot to improve the landing performance due to a significantly better perception of TTC, as compared with a runway outline only.
1. Introduction. Time-to-Contact
The application of flight simulators for training pilots is still growing. It is important
that the information presented by the simulator be as realistic as possible so that a
pilot would perform as in real flight. Still, there are noticeable différences between
simulated and real flights caused by the tact that real and Visual motion are not
perfectly matched in simulators. In order to achieve high fidelity, designers usually
attempt to create extremely detailed visual scènes but as this requires extensive
amounts of computer power, unacceptable time delays may resuit. If one wants to
avoid or minimise time delays it is important to identify what visual information is
essential to obtain a realistic moving scène. This is a challenge for both aerospace
engineers and psychologists.
Gibson (1979) asserts that highly detailed scène éléments provide less information
on egomotion than the optic flow field created by dynamic transformations of those
éléments. Pilots are able to perceive information from the optic flow about the
direction of heading or the aiming point (AP), and about the so-called
Time-to-Contact (TTC, Lee 1980). TTC,
x , or the tau margin is the time remaining to
collision if no action were taken.
When a subject approaches a wall or an object, TTC is decreasing with time. In
différent types of time-constrained tasks like car driving or braking, subjects tend to
use T T C as a eue by which their actions are triggered. That T T C or
x triggers the
onset of the landing flare was demonstrated by flight simulation research at Delft
University of Technology (Advani et al. 1993). In that experiment the timing of the
flare was also related to the height above the ground surface. Height above the
ground corresponds to a certain value of the angular size or optical a n g l e s , which
is the angle between two points sideways trom the AP and the pilot's eyes. The
dependency onip may have been caused by the low visibility of the simulated night
approach, in which the only visible cue was provided by the runway outline. In order
to further examine the influence of TTC on the timing of the flare the present
experiment was designed to test the hypothesis that addition of ground texture to a
synthetic runway scène would increase the visible optie flow field and thus would
improve the perception of T T C as compared with a runway outline only.
2. Landing an Aircraft
During the final approach to a runway a pilot has to fly the aircraft along a glide path
with a slant angle of about 3 [deg] (Figure 1). Generally the vertical component of
the airspeed vector, the sink rate C, is too large for a smooth landing. Assuming a
typical approach speed V = 60 [m/s], the glide path angle
y = -3 [deg] results in
C « 3 [m/s]. At touchdown this sink rate is highly unacceptable for both passengers
and undercarriage. Hence C should be reduced before touchdown, which is done by
executing the flare manoeuvre. The flare is initiated by pulling the steering wheel
backwards, resulting in a more positive pitch angle (Figure 1). The increase of the
pitch angle coupled with an approximately proportional increase of the lift force
effectively reduces the sink rate. The onset of the flare requires a précise coupling
of timing and action. A flare that is too late or too weak results in a hard landing, a
flare initiated too early or done too strong may resuit in a soft landing or no landing
at all. The latter is highly undesirable since the aéroplane has a natural tendency to
return to a climbing flight after such a missed touchdown. In order to avoid this and
to establish an early firm contact between wheels and runway so that effective
wheel braking can start immediately, pilots usually aim at a reasonably firm
touchdown with a certain positive sink rate.
3. Visual Information from the Ground Surface
In most T T C experiments the existence of a T T C strategy for the onset of a collision
avoiding action has been demonstrated by using approaches to a plane
perpendicular to the observer's line of movement. In Figures 2 and 3 perpendicular
approaches to a square object are shown. Assuming an approach along a straight
line with constant velocity V, the tau margin T as introduced by Lee (1980) equals
the real TTC. It has been pointed out thatx can be described in terms of the optical
angle
8 by means of the following relationship:
n
x sinGcosG
ra-v"
de
= T<
1>
dt
in which 0 represents the optical angle between a random point and the aiming
d9
point (AP), and in which — i s the rate of change of 0. In the following the variable
dt
x will be used to designate the optical angle ratio and can be regarded as the
perceived TTC. This x information is directly available from the optie flow field,
<£»
since the inverse ratio represents the relative velocity of the optical image
8
expanding across the retina. Here the relationship between TTC and optical angle
information was shown by using the aiming point related angle9. However, such a
relation can also be shown for other optical angles for instance the optical angle ç
between two points from the lower side of the square (Figure 2). As a conséquence,
the optical angle ratio can be applied to every point of the object resulting in the
same
x . Hence, TTC can be perceived from the entire optic flow field during a
perpendicular approach.
Next we consider a slant approach to a square object. The optical sizes of the
square using a 1 second interval are shown in Figure 4.
Figures 4 and 3 seem rather similar, but the optic flow field of the slant surface does
not expand isotropically as in Figure 3. Rather, this flow field can be divided into two
différent areas (Lee 1974): the upper area is that part of the ground surface in
between the horizon and the aiming point, whereas the lower area is covered by the
remaining part of the visible ground surface, see Figure 5.
It can be seen that the perceived TTC from the upper area XUA
i slarger than the
real TTC at the AP, whereas XLA. from the lower area is smaller than TTC. The
border line between both areas (the line parallel to the horizon and passing through
the AP, therefore called the aiming line, AL) is the flow line providing the real TTC,
because it is perpendicular to the line of movement.
4.
Runway Outline versus Ground Texture
Before we can compare runway outline and ground texture, we should define both.
A runway outline can be regarded as a visible trapézoïdal shape, sharply
distinguished from the surroundings. Ground texture can be defined as a spatial
array of patches, lines or points varying in size, shape, posture, colour or brightness
(partly adopted from Bookout and Sinacori 1993).
In conformity with the optical angleGas introduced in Figure 2 the optical angle
e
R W (= y
^ty)
canbe defined for a slant approach to a runway outline (Figure 6)
being the angle between a point sideways from AP, the pilot's eyes and AP itself.
For large values of TTC, the angle 8RVV is very small. Approaching the runway
and thus reducing TTC, the value of 0
R W will rapidly increase as can be seen in
Figure 7 where the horizontal line represents the pilot's line of motion to AP. Further,
the slope of the S R W -path corresponds to its rate-of- change (^RW ^ AJong the
dt
aiming line, TTC can only be perceived by observing the ratio:
TA1 e
RW
nx
1
dt '
Next consider a ground surface containing a runway outline and texture éléments.
In this case ail the texture éléments along the aiming line provide optical information
to the pilot as given by Eq. (3). This is illustrated by Figure 8 where a number of
9
paths as a function of TTC are shown. Since more optical information is available in
a textured visible environment a better perception of TTC can be expected if ground
texture is présent compared to the case of a runway outline only.
Results of landing experiments in Simulators as reported in the literature are
somewhat contradictory. Addition of large spaced checkerboard texture to a
simulated runway scène did not improved landing performance as compared to an
outline only scène in one particular experiment (Harris et al. ,1978). Rather landing
performance deteriorated when texture was visible. From other experiments
(Bennett et al. 1986; Warren and Riccio 1985; Wolpert et al. 1983; Zacharias 1985)
it appeared that a terrain following task, which is slightly different from a landing
approach, is best executed when the outline of a road or runway is visible. The
présence of ground texture again appeared to reduce the pilot's performance.
5.
Method
The experiment was done by using a Silicon Graphics Iris Indigo Workstation with a
Silicon Graphics 17" colour monitor. The expérimental subjects were positioned at a
viewing distance of 33 cm resulting in an eye-field-of-view (EFOV, Mulder 1994) of
approximately 49°(azimuth) by 37° (élévation). Together with a 17 Hz image
update rate the expérimental design suffices the Workstation simulation
requirements as suggested by Batson et al. (1992).
In order to properly test the hypothesis that addition of ground texture improves the
perception of T T C three different synthetic runway scènes were displayed on the
monitor (Figure 9):
A. runway outline
B. ground texture
C. runway outline and ground texture
The ground texture was represented by a random line pattern consisting of about
5m long lines. The lines were randomly oriented and located across the ground
surface. At the start of the simulation 1000 lines were visible. During a simulation
run a continuously decreasing part of the ground surface was visually available to
an expérimental subject, using a constant EFOV. In order to obtain a sufficiënt
number of texture éléments remaining visible during the approach, the éléments
were divided around the AP according to a normal distribution. Further, a density
gradient was added to the texture to enable a sufficiënt perception of the flat ground
surface, as has been pointed out by Cutting and Millard (1984).
Assuming that in real flights the onset of the flare would mainly be based on TTC,
the possible temptation of a subject to trigger the flare at a certain height
corresponding to a value of the optical angle0RW , should be prevented. Hence two
runway widths were used, resulting in a total number of five synthetic runway
scènes. The dimensions of both runways varied in width (W = 40m and W = 60m)
and in the distance between the runway threshold and the aiming point (LT = 200m
and LT = 300m).
The synthetic scènes were approached along a straight line using three different
approach speeds (V = 50m/s, 60m/s and 70m/s) and two different glide path angles
(y = -2° and y = -4°). The combination of the different values of V and y resulted in
six different sink rates C, ranging from C = 1.74m/s to C = 4.88m/s. Because every
sink rate was applied to every runway scène, the experiment consisted of thirty
different simulation conditions, which were selected by the computer in random
order.
6. The pre-programmed Flare Manoeuvre
As explained earlier, the approach is to be succeeded by the flare manoeuvre in
order to obtain a smooth landing. The profile of the flare executed by transport
aircraft has been shown to be approximately exponential in nature (Roskam 1979).
The flare in the present experiment was pre-programmed using the following
exponential équation for the height:
H T O - H f - H o ^ l
- e ™
0* ) • (4)
In this expression H(t) is the eye height [m] above the ground surface and
Tf < t [sec]. Further, the index f represents the moment of onset of the flare, and
the index opt represents the optimal moment of flare initiation. If flare initiation
timeTf is taken to be zero for a perfect or optimal flare, soTf = 0, and Hf = H
o pt
then:
H(t) = H
o p t. e
ijT°
0^
. (5)
The sink rate C(t) = -
d l" j ^ [m/sec] during the exponential flare of Eq. (5) is then,
dt
C ( „ . - ^ H L . e " W .
( 6 )"TCopt
Notice that the sink rate at flare initiation (t=0) is:
c
m
m
T T Co p t
At the moment of touchdown, (t - tt.d) the eye height above the ground is equal to
the eye height AH above the wheels, so,
-(
t td)
H ( t
t d) = AH = H
0.e
TTC° P
t, (8)
and the touchdown sink rate C^d is:
-(
ttd)
C t d - - H ( t t d
. ) - = ^ - . e • O)
•TCopt
After some elaboration It can be shown that the sink rate C
t^ at touchdown for the
'ideal' exponential landing flare can be expressed by:
C t d = -H(t
t.d) = = £ r - • (10)
• '^opt
It follows from Eqs (6) through (10) that the flare is completely determined by the
parameters Hn, T T C
0p t and AH. After extensive pre-experiment testing it was
decided to set T T C
o pt at 6 [sec]. Next, AH was set at a value of 3 [m], resulting in
an 'ideal' or optimum touchdown sink rate of:
Notice that now with T T C
o pt and A H being set, the value of HQ follows directly
from Eq. (7). In fact HQ is set by the various initial sink rates C(0) of the experiment.
A margin for acceptable touchdowns was set by :
0.05 s Ct.rj. s 0.90 [m/sec].
This touchdown margin can be translated to a TTCf margin: the TTC Funnel
(Figure 10). In this figure the required TTCfat fiare initiation for an acceptable
landing has been plotted against the approach sink rate C(0). The horizontal line
represents the optimal C
t¿ « 0.33 [m / sec]. The upper curve represents the
minimal C
t^ = 0.05 [m / sec] and the lower curve stands for the maximal
• Ct.d.
=° -
9 0t
mI
se cl • ^
rom t n i sf ' 9
u r e ¡ t c an De s e e n t n a t t n ehigher the approach
sink rate, the narrower the TTCf margin will be. Henee it can be expected that
approaches with large values of C(o) (i.e. the steepest approaches) are more
difficult to land than others.
Seven subjeets without any previous flying expérience participated in the
experiment. After several training sessions, each subject completed ten replications
of all thirty experimental configurations. The subjeets were instructed to initiate the
fiare - by pressing the spacebar - in order to obtain a landing with a sink rate as
close as possible to Ct
>fj. = 0.33 [m/sec] . After each trial, subjeets were provided
with feedback information of the sink rate at touchdown.
7. Results
As has been explained earlier, a satisfactory perception of TTC corresponds to a
touchdown within the limits set for acceptable touchdown sink rates. Henee the
percentages of achieved touchdowns per display offer a first indication for the
quality of the perception process for each display.
In Table 1 touchdown percentages of ail simulation runs conducted by the seven
subjeets are shown. From this table it appears that the displays containing texture
enabled subjeets to achieve more touchdowns due to a possibly improved
perception of T T C as compared with displays without visible texture.
Table 1. Percentage of successful touchdowns
-0.90 < H(tt.d.) <--05 [m/sec]
Display
Successful
Touchdowns
I
small runway
51 %
II
small runway + texture
5 7 %
II
texture only
6 0 %
IV
large runway
5 5 %
This preliminary conclusion is confirmed by extensive multivariate Analysis of
Variance (ANOVA) computations, using a significance level at
a = 0.05.
The application of ANOVA's for statistical analysis of the experimental data has
been allowed by Kolmogorov-Smimov goodness-of-fit tests. These tests clearly
revealed the approximately normal distribution of the data, a requirement for
correctly interpreting the ANOVA results.
Recalling the TTC Funnel, one may assume that the best perception of TTC would
be revealed by a constant TTC strategy independent of the approach speed V, the
glide path angle y and the runway size. The results presented below are illustrated
by the data of one typical subject, as shown in Figure 11 through 15 . These figures
show typical mean values and standard deviations of TTCf and \pf at onset of the
flare as functions of the initial sink rate C(0). For the two runway outline only
displays (I and IV, Figure 11) TTCf at onset of the flare was significantly influenced
by V for all subjects. Except for two subjects, y did not affect TTCf. Further, the
runway size significantly affected TTCf for most subjects.
The three displays containing visible ground texture (II, III and V, Figure 12)
predominantly enabled subjects to use constant TTC strategies. TTCf was neither
affected by V nor by y for most (five) subjects. The significant differences for the
others mainly occurred at the lowest and highest values of the approach sink rate
C(0). In spite of the seemingly invariable TTC strategies, the runway size
significantly influenced TTCf for some subjects, indicating that they were gazing at
both texture and runway.
Besides TTCf the effects of the simulation variables on the optical angle % at
onset of the flare were also examined.
For the displays without texture (I and IV, Figures 13 and 14) % was not
significantly influenced by V for all subjects and was affected by y for only two
subjects, suggesting a constant \|> strategy. However, tyf was significantly affected
by the runway size for all subjects, which may seem rather paradoxical.
As a confirmation of the exposed TTC strategy for the textured displays (II, III and V,
Figure 15), ^ f was significantly influenced by V for all subjects and was influenced
by Y for some subjects. Finally, tpf was significantly affected by the runway size as
might have been expected.
8. Discussion
The results indicate that the availability of visible ground texture in a simulated
runway scene yields improved perception of T T C as compared with a runway outline
only scene. This enhanced perception allows a pilot to initiate the flare on the basis
of T T C only.
Without texture, subjects seem to prefer constant ^ strategies to trigger the flare,
such that a 'small' strategy is used when approaching a small runway, whereas a
'large' ip strategy is used for a larger runway.
Although these results are clear cut, the question remains why texture provides
improved perception of TTC compared to a runway outline only scene. The present
results are rather surprising, especially regarding the previously mentioned simulator
experiments (Harris et al. (1978), Bennett et al. (1986), Warren and Riccio (1985),
Wolpert et al. (1983), Zacharlas (1985)), the outcome of which that the outline of a
runway or road improves pilot's performance if compared to ground texture.
A suitable explanation for the different results of the present and the other TTC
experiments appears to be well possible by a further detailed analysis of the
differences in the tasks and the displays used. Such a detailed analysis, however, is
beyond the scope of the present paper.
9. Conclusions
The presence of visible ground texture in simulated landings was shown to
significantly improve the perception of TTC and hence the landing performance as
compared with a runway outline only scene. The presence of a visible runway
appeared to have negligible effect on the perception of TTC. Hence without ground
texture as a visible cue, subjects seemed to time the onset of the flare manoeuvre
on the basis of the optical angle
tp.
Because the real TTC is only provided by the aiming line, the conclusion may be
drawn that pilots mainly perceive TTC from this line. However to validate this
hypothesis, a next experiment should be conducted, in which the effects of texture
patterns along several lateral lines - corresponding to virtual flow planes - on the
perception of TTC can be identified.
References
Advanl, S.K. and U.C. van der Vaart, R.Th. Rysdyk, J. Grosz (1993)
What Optical Cues Do Pilots Use to Initiate the Landing Flare? Results of a Piloted
Simulator Experiment. Proceedings of the AIAA Flight Simulation Technologies
Conference, Monterey, CA (USA), August 9-11, AIAA-93-3561 -CP.
Batson, V.M. and R.L Harris Sr, J.A. Houck (1992)
Effects of Display Parameters on Pilots' Ability to Approach, Flare and Land.
Proceedings of the AIAA, AIAA-92-4139-CP.
Bennett, C.T. and W.W. Johnson, J.A. Perrone, A.V. Phatak (1986)
Synthetic Perspective Optical Flow: Influence on Pilot Control Tasks.
Bookout, G. and J. Sinacori (1993)
Texture as a Visual Cueing Element in Computer Image Generation, Part I
-Representation of the Sea Surface. Proceedings of the AIAA Flight Simulation
Technologies Conference. Monterey (CA), August, 9-11, AIAA-93-3560-CP.
Cutting, J.E. en R.T. Millard (1984)
Three Gradients and the Perception of Flat and Curved Surfaces. Journal of
Experimental Psychology, Vol.113, No.2.
Gibson, J.J. (1979)
The Ecological Approach to Visual Perception. Boston (MA): Houghton Mifflin
Company.
Harris Sr, R.L. and M.C. Waller, S. Salmirs (1978)
Runway Texturing Requirements for a Head-Down Cathode Ray Tube Approach
and Landing Display. Proceedings of the AIAA Flight Simulation Technologies
Conference. Arlington (TX), September, 18-20. AIAA-78-1588.
Lee, D.N. (1974)
Visual Information during Locomotion. In: McLeod, R.B. and H. Picks (Eds),
Perception: Essays in Honor of J.J. Gibson. Ithaca (NY): Cornell University Press.
Lee, D.N. (1980)
Visuo-Motor Co-ordination in Space-Time. In: Stelmach, G.E. and J. Requin (Eds),
Tutorials in Motor Behavior. Amsterdam: North-Holland Publishing Company.
Mulder, M. (1994)
Displays, Perception and Aircraft Control. Report LR-762. Delft: Delft University of
Technology, Faculty of Aerospace Engineering.
Roskam, J. (1979)
Airplane Flight Dynamics and Automatic Flight Controls, Part II. Lawrence (KA):
University of Kansas.
Warren, R. and G.E. Riccio (1985)
Visual Cue Dominance Hierarchies: Implications for Simulator Design. Society of
Automotive Engineers (SAE) Technical Paper Series, Long Beach, October, 14-17.
Wolpert, L. and D.H. Owen, R. Warren (1983)
Eyeheight-Scaled versus Ground-Texture-Unit-Scaled Metrics for the Detection of
Loss in Altitude. In: Jensen, R.S. (Ed.), Proceedings of the Second Symposium on
Aviation Psychology. Columbus (OH).
Zacharias, G.L (1985)
Modelling the Pilot's Use of Flight Simulator Visual Cues in a Terrain-Following
Task. Technical Report No.R8505, prepared for J.B. Sinacori Associates under
USAF ASD Contract no. F33615-81-0515.
Figure 3 Expansion of square surface during a perpendicular approach; optical sizes
of square with a 1 second interval
Figure 4 Expansion of square surface during a slant approach; optical sizes of
square with a 1 second interval
Figure
5 Side view of the slant approach defining the optical angle
(bAP
Runway Outline
0 2 4 6 8 10 12
TTC [sec]
Figure 7 Optical angle
8R Was a function of T T C
Ground Texture
TTC [sec]
Figure 8 Optical angles
8-^as a function of T T C
Figure 9 Synthetic runway scènes: A. runway outline, B. ground texture, C. runway
outline and ground texture
T T C F u n n e l
8TOUCHDOWN
C R A S H
NO LANDING
max,
CTDOpt.
CTDm i n .
CTD 1.5Z5
33J5
C[m/s]
4 v 5TTCf at onset of flare
Display IV : large runway
one subject
1.74 (50;2) 2.09 (60;2) 2.44 (70;2) 3.48 (50;4) 4.18 (60;4) 4.88 (70;4)
C [m/s]
Figure 11
TTC
fversus
C at onset of the flare. Display IV: large runway (results oî one
subject)
TTCf at onset of flare
Display V : large runway + texture
one subject
1.74 (50;2) 2.09 (60;2) 2.44 (70;2) 3.48 (50;4) 4.18 (60;4) 4.88(70,4)
C [m/s]
Figure 12
TTC, versus C at onset of the flare. Display V: large runway and texture
(results of one subject)
PSlf at onset of flare
Display I : small runway
one subject
1.74 (50;2) 2.09 (60;2) 2.44 (70;2) 3.48 (50;4) 4.18 (60;4) 4 . 8 8 (70;4)
C [m/s]
Figure 13 ij>, versus
C at onset of the flare. Display I: small runway (results of one
subject)
PSlf at onset of flare
Display IV : large runway
one subject
1 . 7 4 ( 5 0 ; 2 ) 2 . 0 9 ( 6 0 : 2 ) 2.44 (70;2) 3.48 (50;4) 4.18 (60;4) 4 . 8 8 (70;4)
C [m/s]
Figure 14 tp, versus
C at onset of the flare. Display IV: large runway (results of one
PSIf at onset of flare
Display V : large runway + texture
one subject
13 12- 11-O) 10-0) 31 9 cö o_ 8- 7- 6-5 1 . 7 4 ( 5 0 , 2 ) 2.09 (60;2) 2.44 (70;2) 3 . 4 8 (50;4) 4.18 (60;4) 4 . 8 8 (70;4) C [m/s]Figure 15
% versus C at onset of the flare. Display V: large runway and texture
(results of one subject)
M O D E L L I N G M A N U A L C O N T R O L O F S T R A I G H T T R A J E C T O R I E S
W I T H A P E R S P E C T I V E F L I G H T - P A T H D I S P L A Y
Max Mulder
Delft University of Technology, Faculty of Aerospace Engineering, P.O. Box 5058, 2600 GB Delft, The Netherlands
e-mail: m.mulder@lr.tudelft.nl
Abstract. A perspective flight path display shows the flight trajectory to be flown in a synthetic three-dimensional world. The application o f such a display has important conséquences for a pilot because the guidance information is presented via spatial sources o f information. To understand pilot manual controi behaviour with a perspective display, it is essential to investigate the manner in which pilots use thèse oprical eues. The paper describes the approach chosen and discusses results o f one o f the experiments. Keywords. Perspective flight-path display, Manual controi, human operator modelling
1. I N T R O D U C T I O N
A perspective flight-path display, s h o w i n g t h e p l a n n e d trajectory t o t h e p i l o t i n a synthetic three-dimensional w o r l d ( F i g u r e 1), is not a new concept. Since t h e early 1950's i t h a s been h y -pothesized t h a t such a p i c t o r i a l display c o u l d m e a n , i n m a n y ways, a n i m p o r t a n t i m p r o v e m e n t i n informationtransfer t o t h e p i l o t . Its a p p l i c a -t i o n was i m p r a c -t i c a l , however, due -t o -technical l i m i t a t i o n s .
B a s i c a l l y , t w o developments i n technology m a d e the p i c t o r i a l display concept p r a c t i c a l . F i r s t , r a p i d i m p r o v e m e n t s i n c o m p u t e r technology m a d e sufficiently detailed real-time graphies possible.
Second, the advance o f new p o s i t i o n i n g
Systems,
such as GPS ( G l o b a l P o s i t i o n i n g System) a n d
MLS ( M i c r o w a v e L a n d i n g S y s t e m ) , p r o v i d e d t h e c a p a b i l i t y o f m e a s u r i n g the p o s i t i o n of the aircraft w i t h a sufficiënt u p d a t e rate.
T h e a p p l i c a t i o n o f a perspective display i n t h e cockpit has i m p o r t a n t conséquences. I n a conven-t i o n a l cockpiconven-t conven-the p i l o conven-t m e n conven-t a l l y reconsconven-trucconven-ts conven-t h e aircraft's s p a t i a l a n d t e m p o r a l s i t u a t i o n f r o m a n u m b e r o f planar, i.e. t w o - d i m e n s i o n a l , displays. W i t h a perspective flight-path display this infor-m a t i o n is presented i n a spatial f o r infor-m a t [Theunis-sen a n d M u l d e r , 1995].
A t t h e Delft U n i v e r s i t y o f Technology, research
Fig. 1. The Tunnel-in-the-Sky display
is being conducted t o investigate t h e i m p l i c a t i o n s a perspective flight-path display has o n p i l o t ' s m a n u a l controi behaviour. T h e m a i n subject o f interest is n o t , however, whether such a display w i l l lead t o a n i m p r o v e d m a n - m a c h i n e interface. R a t h e r , t h e research i m p e t u s is t o détermine hous a pilot is able t o controi a c o m p l e x d y n a m i c System (the aircraft) along a spaceconstrained t r a -jectory, w i t h a perspective flight-path display as p r i m a r y source o f i n f o r m a t i o n . O n c e this ques-tion has been answered, a n a t t e m p t c a n be m a d e to represent t h e i m p o r t a n t characteristics o f t h e
XVth European Annual Conference on Human Décision Making and Manual Controi (1996)
p i l o t i n a m a t h e m a t i c a l model
T h i s paper describes one of the experiments w h i c h have been conducted. T h e intent is to show clearly the m a n n e r i n w h i c h the research issues are ap-proached. Section 2 discusses the d e c o m p o s i t i o n of the p r o b l e m i n t o different subsets. T h e n re-sults f r o m a cue-inventory are presented i n sec-t i o n 3. In order sec-to invessec-tigasec-te some of sec-the hy-potheses f o l l o w i n g f r o m this cue-inventory, an ex-p e r i m e n t has been conducted. Section 4 describes the setup of t h i s experiment w h i l e sections 5 a n d 6 discuss the t i m e - d o m a i n a n d frequency-domain results f r o m the experiment respectively. F i n a l l y , section 7 contains a discussion of the approach followed a n d conclusions.
2. T O W A R D S A C O N T R O L - T H E O R E T I C M O D E L O F P I L O T M A N U A L C O N T R O L
B E H A V I O U R W I T H A P E R S P E C T I V E F L I G H T - P A T H D I S P L A Y
I n [ M u l d e r , 1995] the m e t h o d o l o g y of the research project has been discussed. B e l o w , the m a i n points of interest are briefly repeated.
2.1. Research goal
O u r goal is to u n d e r s t a n d a n d , u l t i m a t e l y , to m o d e l the m a i n characteristics o f p i l o t m a n u a l control b e h a v i o u r w i t h a perspective flight-path display. T h e p a r t i c u l a r research interests are es-p e c i a l l y the information-transfer between the p l a y a n d the p i l o t , i n c l u d i n g the influences of dis-p l a y design variables a n d the effects of a d d i t i o n a l display symbology.
2.2. Modelling approach
T h e p i l o t ' s task is to follow the reference trajec-tory, a t y p i c a l guidance task. B a s e d o n the char-acteristics of the reference trajectory, the guidance task can be d i v i d e d i n t o a n u m b e r of phases (or
sub-tasks):
1. to m a i n t a i n a straight section o f the trajec-tory,
2. to m a i n t a i n a curved section of the trajectory, 3. to control a t r a n s i t i o n between a straight a n d a curved section of the trajectory a n d vice versa.
In other words, or, f r o m a system-theoretical p o i n t of view, f o l l o w i n g the p l a n n e d trajectory leads to m a i n t a i n i n g a series of different system steady-states (or references) a n d c o n t r o l l i n g transitions between these steady-states. T h e m o d e l l i n g
at-tempts are s t r u c t u r e d a c c o r d i n g to t h i s decompo-sition p r i n c i p l e .
M a i n t a i n i n g a certain state of the s y s t e m against disturbances is s i m p l y a r e g u l a t i o n task. T o m o d e l this task, t w o w e l l k n o w n m o d e l l i n g m e t h o d o l o -gies can be a p p l i e d : T h e classical a p p r o a c h , re-s u l t i n g i n a m u l t i - c h a n n e l verre-sion o f M c R u e r ' re-s crossover-model [ M c R u e r et a l . , 1965], a n d a n
op-timal control approach, u s i n g the O p t i m a l C o n t r o l M o d e l ( O C M ) of K l e i n m a n et a l . [ K l e i n m a n et a l . 1969]. T h e t r a n s i t i o n phase between t w o different steady-states can be m o d e l l e d u s i n g a n extension to the conventional O C M , i.e. the O p t i m a l C o n -t r o l a n d P r e v i e w M o d e l ( O C P M ) , w h i c h has been described i n [ M u l d e r , 1995].
In the f o l l o w i n g , the discussion w i l l be restricted to the regulation task o f f o l l o w i n g straight sections of the t u n n e l trajectory.
3. O P T I C A L C U E S I N A S T R A I G H T T U N N E L S E G M E N T
3.1. General
A perspective flightpath display shows the p l a n -ned trajectory i n a s y n t h e t i c t h r e e - d i m e n s i o n a l w o r l d . T h e task o f the p i l o t is to c o n t r o l the air-craft a l o n g this p a t h . T o fulfil his task, the p i l o t estimates the state o f the aircraft w i t h respect to the trajectory a n d , based o n the estimated state, decides u p o n a n d activates the necessary c o n t r o l action(s). I n order to u n d e r s t a n d the i n t e r a c t i o n between the p i l o t a n d the display i t is essential to get grip of this state e s t i m a t i o n process. T h i s has been investigated f r o m two different points of view.
In [ M u l d e r , 1994] i t was e x a m i n e d w h a t effects a s p a t i a l display has o n the c o n t r o l b e h a v i o u r of a pilot: T h e man i n the m a n - m a c h i n e interface was taken to be the central element. M a i n questions that were addressed were the a v a i l a b i l i t y , the use-fulness a n d the p o t e n t i a l u t i l i z a t i o n o f a l l sorts of s p a t i a l , or o p t i c a l sources o f i n f o r m a t i o n present i n the real w o r l d a n d / o r i n a perspective display. In [ M u l d e r , 1996] the machine side was the m a i n issue. A n a t t e m p t was m a d e to m a k e a n i n v e n t o r y of a l l s p a t i a l cues i n a basic perspective f l i g h t - p a t h display. Here, irrespective o f the h u m a n operator, m a t h e m a t i c a l relations are derived t h a t express the state of the aircraft w i t h respect to the ref-erence trajectory i n t e r m s of these o p t i c a l cues. O b v i o u s l y , m o s t o f the available cues w i l l p r o b -ably be neglected b y the operator for reasons o f their p e r c e i v a b i l i t y thresholds or s i m p l y because
XTE/TAE -5 [deg] 0 [deg] + 5 [deg]
- 5 [ m ]
0 [ m ]
+5 [m]
Fig. 2. Distortion of the symmetry of the tunnel display they are t o o far-fetched for the operator t o be rec-ognized as p o t e n t i a l sources o f i n f o r m a t i o n . S o m e of the cues, however, are so evident a n d c a n be so easily perceived f r o m the display t h a t i t is m o s t probable that they are used b y the operator. It are these p a r t i c u l a r cues w h i c h are t h e m a i n sub-ject o f investigation.
3.2. Straight tunnel sections
In t h i s paper, the discussion w i l l be restricted t o sections of the trajectory that are straight a n d infinitely l o n g . F u r t h e r m o r e , wind-effects are ne-glected. A s has been discussed i n [Mulder, 1995], i n case o f no p o s i t i o n errors (and relatively s m a l l aircraft a t t i t u d e angles) the t u n n e l image w i l l b e
symmetrie. A n y déviation f r o m the trajectory leads t o a d i s t o r t i o n o f this s y m m e t r i e c o n d i t i o n ( F i g u r e 2 ) . T o m i n i m i z e the discrepancy between the a c t u a l a n d the p l a n n e d trajectory, the opera-tor m u s t m a i n t a i n a s y m m e t r i e t u n n e l image. In [ M u l d e r , 1996] i t is shown t h a t there are m a n y o p t i c a ! cues i n t h e display t h a t are directly related to t h e déviation o f the aircraft p o s i t i o n a n d a t t i -tude f r o m the référence trajectory. F o r example, the aircraft h e a d i n g error, or Track-Angle-Error
(TAE), c a n be perceived f r o m t h e t r a n s l a t i o n of t h a t p a r t o f the t u n n e l w h i c h is located at a large
distance f r o m t h e v i e w p o i n t , as was observed i n [ G r u n w a l d a n d M e r h a v , 1976] a n d [Theunissen, 1994]. I n t h i s paper t h e discussion w i l l be re-stricted t o t h e o p t i c a l cues related t o a p o s i t i o n error.
3.3. Optical cues
T h e l a t e r a l a n d v e r t i c a l p o s i t i o n errors c a n b e es-t i m a es-t e d u s i n g a large n u m b e r o f o p es-t i c a l cues o f w h i c h the most salient ones are i l l u s t r a t e d i n F i g -ure 3. F i r s t o f a l l , we have the relative l a t e r a l displacements e,j a n d o f the t u n n e l frames i a n d j located at distances D,- a n d Dj ( w i t h
Dj = Di + AD, a n d A D the fixed distance
be-tween t w o successive frames). C h a n g e s i n these relative l a t e r a l displacements f r o m t h e zero-error c o n d i t i o n are a f u n c t i o n of l a t e r a l p o s i t i o n error
X o n l y ( i n a p p r o x i m a t i o n ) :
Scij = +KX
\ DiDj ) (1)
(2) w i t h « a display constant dépendent o n the field-of-view o f t h e perspective p r o j e c t i o n a n d the size of the display screen.
T h e same holds for t h e relative v e r t i c a l
displace-XVth European Annual Conference on Human Décision Making and Manual Controi (1996)
Fig. 3. Position error eues in a straight tunnel section ments m¡ a n d vy, w h i c h are a f u n c t i o n o f t h e vertical p o s i t i o n error V o n l y ( i n a p p r o x i m a t i o n ) :
* = - K W )
/Dj -Dj\ \ DiDj ) ova = +KV (3) (4) Note that i n t h e f o r m u l a s stated above, the dévi-ations are f o r a l l possible pairs o f t u n n e l framesi a n d j. A s one c a n see, however, the a p p l i
-cability o f these cues détériorâtes fast when the distances i n v o l v e d b e c o m e larger. T h e
t e r m r a p i d l y decreases t h e a m p l i t u d e s of the dis-placements below threshold.
T h e second set o f cues resuit f r o m t h e l o n g i t u d i -n a l li-nes co-n-necti-ng t h e i -n d i v i d u a l t u -n -n e l frames. T h e angles t h a t the projections o f these lines m a k e w i t h the h o r i z o n are also m e r e l y a f u n c t i o n o f lat-eral a n d v e r t i c a l p o s i t i o n error o n l y : V X (5) Ui = W W (5) V X (6) U>2 =
w
+w
(6) V X (7) U>3 = +w
+w
(7) V X (8) W4 = +w w
(8) w i t h W the (square) t u n n e l w i d t h .A t h i r d eue results f r o m t h e i m a g i n a r y line
Con-necting the intersections of the a l t i t u d e pôles w i t h the b o t t o m o f the t u n n e l frames. T h e resuit is a n angular eue w h i c h changes a l m o s t i d e n t i c a l l y w i t h the latéral p o s i t i o n error:
W5 = - 2 *
W (9)
T h i s eue w i l l be neglected i n t h e f o l l o w i n g , be-cause i n t h e displays used i n t h e e x p e r i m e n t de-scribed below the a l t i t u d e pôles are n o t presented.
3.4. Discussion
T h e linear a n d angular o p t i c a l eues discussed above are b o t h a f u n c t i o n o f v e r t i c a l a n d l a t e r a l p o s i t i o n error only. T h e r e are t w o f u n d a m e n t a l différences between these cues.
F i r s t o f a i l , w h e n the aircraft m o v e s t h r o u g h the t u n n e l , t h e t u n n e l frames t r a n s l a t e t o w a r d s t h e perceiver, w h i l e t h e l o n g i t u d i n a l lines connecting the frames appear t o d o n o t . T h i s i s a n i m p o r -t a n -t fac-t, since -t h e m o -t i o n o f -t h e -t u n n e l frames c o u l d prevent a n accurate e s t i m a t i o n o f the o p t i -cal displacement eues: T h e p i l o t c o n s t a n t l y has t o shift attention towards a n e w set o f frames. T h e angular eues, however, o n l y change because o f a changing p o s i t i o n error, m a k i n g a shift i n
atten-XVth European Annual Conference on Human Décision Making and Manual Control (1996)
Fig. 4. Display A (X = 10 [m], TAE = -5 [deg], W = 45 [m])
Fig. 5. Display B (X = 10 [m], TAE = -5 [deg], W = 45 [m])
t i o n forwards a n d backwards i n t o the t u n n e l i m -age unnecessary.
Second, as one c a n see f r o m t h e formulas stated above, i t is clear t h a t a l a t e r a l a n d a vertical p o s i t i o n error b o t h determine a n y one o f the four a n -gles. T h i s is i n contrast t o t h e fact that a change i n t h e relative l a t e r a l (vertical) displacements o f the t u n n e l frame lines is o n l y a f u n c t i o n o f t h e lateral (vertical) p o s i t i o n error. I n other words, the angular cues are coupled a n d the linear cues
uncoupled w i t h respect t o t h e l a t e r a l a n d vertical
p o s i t i o n errors. T h i s is also a n i m p o r t a n t fact. T h e c o u p l i n g o f the angular cues means that a n y change i n a n y o f t h e four angles c a n be t h e re-sult o f b o t h a v e r t i c a l a n d a l a t e r a l p o s i t i o n error. T h e linear cues, o n the other h a n d , are uncoupled a n d o n l y change w i t h i n t h e same d i m e n s i o n (i.e. l a t e r a l o r vertical) as t h e o c c u r r i n g p o s i t i o n error. 3.5. Hypotheses and experiment justification T h e discussion above shows t h e virtues a n d dis-advantages o f the two p r i m a r y sets o f cues t o esti-m a t e a p o s i t i o n error o n a straight section o f the t u n n e l display. T o e x a m i n e the usefulness o f b o t h sets a n d t h e extent t o w h i c h t h e above m e n t i o n e d characteristics influence t h e i r relative usefulness, a n experiment h a s been conducted.
T h r e e displays were defined, w h i c h are a l l abstrac-tions o f the basic tunnel-in-the-sky display:
1. D i s p l a y A, s h o w i n g o n l y t h e l o n g i t u d i n a l lines connecting the t u n n e l frames (Figure 4 ) , 2. D i s p l a y B, showing only t h e tunnel frames
themselves ( F i g u r e 5),
3. D i s p l a y C, a c o m b i n a t i o n o f displays A a n d
B ( F i g u r e 6).
Fig. 6. Display C (X = 10 [m], TAE = -5 [deg], W = 45 [m])
It is clear that i n d i s p l a y A o n l y t h e angular cues are available, w h i l e i n d i s p l a y B o n l y t h e linear
displacement cues are present 1. F r o m display C,
b o t h sets of cues c a n b e perceived. F u r t h e r note that i n a l l displays t h e aircraft a t t i t u d e , i.e. p i t c h angle 0, r o l l angle § a n d h e a d i n g a n g l e2 \b can a n d
w i l l be perceived i n i d e n t i c a l fashion.
T o analyze the usefulness o f the o p t i c a l cues f r o m the three displays, t w o a d d i t i o n a l variables were i n t r o d u c e d i n t h e experiment:
I T h e effect o f control channel: T h r e e control channels were a p p l i e d :
1 The angular cues could be estimated from the imaginary
lines connecting the tunnel frames' vertices. Results from a pilot questionnaire revealed, however, that this was not the case.
2 Since the reference heading angle of the tunnel is set zero,
the aircraft heading angle equals the track-angle-error.
XVth European Annual Conference on Human Decision Making and Manual Control (1996)
(i) Roll: l a t e r a l t r a c k i n g only, t h e v e r t i c a l p o -sition error V was kept zero,
(ii) Pitch: v e r t i c a l t r a c k i n g only, the l a t e r a l p o -s i t i o n error X w a -s kept zero,
(iii) Dual: a c o m b i n a t i o n o f (i) a n d ( i i ) , i.e. b o t h the l a t e r a l as the v e r t i c a l p o s i t i o n er-rors h a d t o be m i n i m i z e d simultaneously. II T h e effect o f forward motion:
T w o situations were e x a m i n e d :
(i) No (forward) motion, i n w h i c h the l o n g i -t u d i n a l p o s i -t i o n o f -t h e aircraf-t was fixed,
resulting i n a hovering t a s k3,
(ii) (Forward) motion, i n w h i c h the l o n g i t u d i -n a l p o s i t i o -n o f the aircraft was set free, re-s u l t i n g i n a conventional t u n n e l t r a c k i n g task.
These a d d i t i o n a l e x p e r i m e n t a l variables c a n be used t o e x a m i n e t h e usefulness o f the t w o sets o f p o s i t i o n e s t i m a t i o n cues according t o the follow-i n g a p r follow-i o r follow-i hypotheses:
I T h e presence o f f o r w a r d m o t i o n has n o effect on the p e r f o r m a n c e4 w i t h display A.
II T h e presence o f forward m o t i o n deteriorates performance w i t h display B.
III T h e a d d i t i o n o f a n a d d i t i o n a l control channel deteriorates performance w i t h display A. I V T h e a d d i t i o n o f a n a d d i t i o n a l c o n t r o l channel
deteriorates performance w i t h display B, b u t to a significantly less degree t h a n t h e perfor-m a n c e deterioration for the saperfor-me c o n d i t i o n o f display A.
T h e effects o f the independent variables m o t i o n a n d control c h a n n e l o n display C are expected t o be a m i x t u r e o f those effects o n displays A a n d
B. Since display C is a c o m b i n a t i o n o f displays
A a n d B, the relative virtues a n d disadvantages
of the c o m b i n e d sets o f cues could compensate for each other.
4. E X P E R I M E N T 4.1. Goal of the experiment
T h e goal of the e x p e r i m e n t was t w o f o l d . F i r s t o f a l l , the v a l i d i t y o f the hypotheses stated i n t h e former section m u s t be e x a m i n e d . Second, the observed control b e h a v i o u r o f t h e subjects m u s t be described a n d e x p l a i n e d w i t h a m a t h e m a t i c a l 3 In the hovering task the lateral and vertical position
er-rors result in an apparent motion in a plane perpendicular to the tunnel centerline.
4 Performance is defined here as the accuracy with which
the reference trajectory can be followed (position errors).
m o d e l .
T e s t i n g t h e hypotheses does n o t d e m a n d m u c h i n -genuity i n the e x p e r i m e n t a l design. O n e c o u l d for instance define a conventional c o m p e n s a t o r y t u n nel t r a c k i n g task, measure a l l performance v a r i -ables o f interest, a n d d o a post-hoc analysis o f the
empirical d a t a u s i n g s t a t i s t i c a l tests.
T h e efforts t o describe t h e observed b e h a v i o u r v i a a m a t h e m a t i c a l m o d e l , however, are n o t so s t r a i g h t f o r w a r d . B o l d l y , one c o u l d state t h a t m a k -i n g models -is relat-ively s -i m p l e , w h -i l e v a l -i d a t -i n g t h e m c a n be extremely difficult. N a t u r a l l y , t h e m o d e l v a l i d a t i o n c o u l d b e restricted t o u s i n g t h e e m p i r i c a l t i m e - d o m a i n d a t a o n l y . M o s t , i f n o t a l l , of the widely-used operator m o d e l l i n g techniques, however, have s h o w n their a p p l i c a b i l i t y especially i n the frequency-domain.
T h e frequency-domain d a t a , i.e. the operator frequencyresponse functions, c a n be h a r d t o o b t a i n . I n o u r s i t u a t i o n t h i s is especially t r u e b e -cause:
• t h e p i l o t is o p e r a t i n g i n closed-loop, w h i c h introduces a l o t o f subtleties i n t h e identifi-c a t i o n proidentifi-cedure,
• w i t h the t y p e o f displays discussed here, the p i l o t is essentially a m u l t i i n p u t , m u l t i -o u t p u t s y s t e m .
In order t o o b t a i n t h e f r e q u e n c y - d o m a i n d a t a , a n identification m e t h o d w i l l be used t h a t was devel-oped i n [Lunteren, 1976] a n d a p p l i e d i n [Paassen, 1994]. T h e a p p l i c a t i o n o f this m e t h o d h a s i m p o r -t a n -t consequences for -the d e f i n i -t i o n o f -t h e exper-i m e n t , as w exper-i l l b e dexper-iscussed exper-i n t h e next sectexper-ion. 4.2. Identification procedure
4.2.1. method. I n [Lunteren, 1976], a n o n -p a r a m e t r i c i d e n t i f i c a t i o n m e t h o d is develo-ped t o estimate p i l o t frequency-response functions i n closed-loop. A l t h o u g h a f u l l discussion o f t h i s m e t h o d is b e y o n d t h e scope o f t h i s p a p e r , t h e m a i n concepts w i l l b e briefly addressed.
B a s i c a l l y , for each p i l o t i n p u t s i g n a l a n i n d e p e n -dent reference (or, i n o u r case, disturbance) sign a l , u s u a l l y a s u m o f sisignusoids, has t o b e i sign t r o duced into t h e closed p i l o t / v e h i c l e l o o p . I n a w e l l -chosen e x p e r i m e n t a l setup, a l l variables o f inter-est t h e n c o n t a i n m u c h power at t h e frequencies o f the disturbance signals a n d o n l y l i t t l e power at a l l other frequencies. T h e operator's frequency-response f u n c t i o n c a n t h e n be e s t i m a t e d b y inter-p o l a t i o n a n d inter-p r o inter-p e r m a n i inter-p u l a t i o n o f t h e F o u r i e r coefficients o f the F F T - e d f r e q u e n c y - d o m a i n d a t a i n a m a n n e r described i n d e t a i l i n [Paassen, 1994].