Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02615250 2008-01-14
1
Method and System for Modelling an Interface Between a User and the
Environment Thereof in a Motor Vehicle
The invention relates to a method and a system for determining a model
of an interface between an operator and his environment on board a vehicle.
In different sectors (aeronautics, automobile, maritime, etc.), the air,
terrestrial or maritime vehicles necessitate, for use thereof (control or
operation,
navigation, communication, environmental monitoring, system management,
etc.), instrument panels endowed with a plurality of interface elements.
To complete his task properly, the user of the vehicle under consideration
controls interface elements and must fully understand the functions performed
by
these interface elements, the information that they deliver and the procedures
describing sequences of actions (manual, visual, auditory) to be accomplished
in
relation with the interface elements.
Thus it is understood that, during control of a vehicle, the interaction
between the user and the interface elements disposed on board the vehicle has
great importance and is therefore the subject of much attention.
It therefore would be interesting to be able to evaluate this interaction in
novel and efficient manner, in order, for example, to be capable of improving
existing interface elements, of designing new such elements or of improving
flight procedures, or else of improving the arrangement of a plurality of
interface
elements relative to one another.
To this end, the present invention has as its object a method for
determining a model of an interface between a user and his environment on
board a vehicle, characterized in that it includes the following steps:
¨ construction of an interface model on the basis, on the one hand, of a
first type of information representative of interface elements of the
vehicle and, on the other hand, of a second type of information
CA 02615250 2008-01-14
2
representative of the knowledge that a user has concerning the use of
interface elements.
¨ acquisition of data representative of at least one human activity which
is sought during the interaction between the user and these interface
elements, the acquisition of these data being achieved by means of at
least one data-acquisition apparatus,
¨ analysis of the data thus acquired,
¨ adjustment of the interface model as a function of the data analysis.
The interface model is constructed on the basis of the duality of the user
and technical system and not merely of the technical information
representative
of the system, which makes it possible to obtain a very reliable model founded
on a set of information internally structured in such a way that it is
particularly
representative of the interaction between the user and his environment on
board
the vehicle, and especially the interface elements thereof.
By virtue of recorded data that reflect the visual and/or gestural and/or
vocal and/or physiological behavior of the user in relation to the interface
elements and to the interpretation of these data, it is possible to enrich the
previously constructed interface model and therefore to adapt it as closely as
possible to the context that it is supposed to represent.
As an example, it is possible to detect anomalies of functioning in the
interface elements, to evaluate new interface elements, to determine that an
interface element should deliver certain information or ensure certain
functions,
or else to determine that a new interface element performing one or more given
functions would be particularly useful.
According to one characteristic, the two types of information, the first type
of information of technical origin and the second type of information of human
origin, are furnished with an identical configuration to a dynamic database
having
a symmetric structure with respect to the user and the technical system.
CA 02615250 2008-01-14
3
To the extent that the information of both types is always furnished in the
same unique format, a gain in time and efficiency is achieved in the
processing
of such information and therefore in construction of the model.
According to one characteristic, both types of information are configured
according to the same multi-agent cognitive model.
Such a representation of the information proves to be particularly suitable
and efficient for construction of the interaction model.
According to one characteristic, the configuration of information of the first
type according to a multi-agent cognitive model comprises a step of
establishment of a link between procedures for use of the vehicle and the
interface elements of the vehicle.
In this way there is established a correspondence between the different
steps of the procedures for use of the vehicle (such as control) and the
interface
elements involved in each step in order to model such elements.
According to one characteristic, the configuration of information of the first
type according to a multi-agent cognitive model comprises a step of
identification
of functional zones on each interface element under consideration.
By defining such zones within one and the same interface element, it will
be possible to obtain a detailed model of each interface element and therefore
to
obtain thereafter, during the data-acquisition step, detailed information on
the
interaction between the user (such as the pilot) and the zones or even several
zones of different interface elements.
In this way the model will be even more complete and thus more reliable
by acquiring, for example, oculometric data relating to these zones of one and
the same interface element or of several such elements.
According to one characteristic, the configuration of information of the first
type according to a multi-agent cognitive model comprises the following steps
for
each interface element:
CA 02615250 2008-01-14
4
¨ determination of tasks relating to the use of the vehicle and performed
by the interface element under consideration,
¨ determination of agents of the multi-agent cognitive model relative to
the determined tasks,
¨ establishment of a link between the thus determined agents of the
cognitive model and the identified functional zones of the interface
element under consideration.
The model established in this way is particularly representative of the
interaction of the user (such as a pilot) with the interface element under
consideration, taking into account tasks assigned to the interface element and
that are determined, for example, by procedures for use of the vehicle (such
as
control).
According to one characteristic, the human activity sought during the
interaction between the user and the interface elements is selected among
vision, speech, hearing, motor function, and physiological manifestations and
reactions of the human body.
The acquisition and analysis of data reflecting very diverse human
activities furnish very useful information, with which, for example, the
interaction
model can be supplemented/modified.
According to one characteristic, the data-acquisition apparatus is an
oculometric apparatus recording visual data representative of the eye
movements of the user over the interface elements.
Such an apparatus is particularly useful for describing the visual behavior
of the user (such as a pilot) when his gaze moves over different interface
elements as well as the exterior visual scene, or even particular zones within
one
or more interface elements.
This apparatus may be coupled with another apparatus making it possible
to record, in video form, for example, the gestures of the pilot, while the
position
of his gaze is tracked by the first apparatus. Audio recordings may also be
very
useful. In this way there is obtained a greater number of data to be processed
and a reater variety of data, thus making it possible to enrich the interface
CA 02615250 2013-02-14
=
model and to make it even more faithful to the context that it is suppose to
represent.
The interface model determined as briefly described hereinabove finds
applications in numerous areas (aeronautics, space, automobile, maritime,
etc.)
and can be used in numerous applications:
- improvement of one or more interface elements;
- design of one or more interface elements;
- evaluation of one or more interface elements;
- modification of a procedure for use (such as control) of the
vehicle;
- training of users (such as pilots) in the control of the vehicle.
The invention also has as its object a system for determining a model of
an interface between a user an his environment on board a vehicle,
characterized in that it includes:
- means for construction of an interface model on the basis, on
the one
hand, of a first type of information representative of interface elements
of the vehicle and, on the other hand, of a second type of information
representative of the knowledge that a user has concerning the use of
interface elements.
- at least one apparatus for acquisition of data representative
of at least
one human activity which is sought during the interaction between the
user and these interface elements,
- means for analysis of the data thus acquired,
- means for adjustment of the interface model as a function of
the data
analysis.
In a further aspect, the present invention also resides in a method for
determining a model for an arrangement of on-board instruments in a
vehicle, the method comprising: constructing, with a processor, the model of
the arrangement of the on-board instruments within an instrument panel
based on a first type of information representative of the on-board
instruments of the vehicle, and a second type of information representative
of the knowledge that the user has concerning the use of the on-board
instruments of the vehicle; acquiring data representative of at least one
CA 02615250 2013-02-14
5a
human activity which occurs during the interaction between the user and the
on-board instruments, the acquiring of the representative data is achieved by
at least one data-acquisition apparatus; analyzing, with a processor, the data
that is acquired; and adjusting the model for the arrangement of the on-
board instruments within the instrument panel as a function of the data that
was analyzed so as to obtain an adjusted model that is different from the
previously constructed model, wherein the configuration of information of the
first type according to a multi-agent cognitive model includes the following
steps for each on-board instrument: determining tasks relating to the use of
the vehicle and performed by the on-board instrument under consideration,
determining agents of the multi-agent cognitive model relative to the
determined tasks, and establishing a link between the determined agents of
the cognitive model and identified functional zones of the on-board
instrument under consideration.
In a still further aspect, the present invention also resides in a system
which determines a model for an arrangement of on-board instruments in a
vehicle, comprising: a processor; a construction unit which constructs the
model of the arrangement of the on-board instruments within an instrument
panel based on a first type of information representative of the on-board
instruments of the vehicle, and a second type of information representative
of the knowledge that a user has concerning the use of the on-board
instruments; at least one data acquisition unit which acquires data
representative of at least one human activity which occurs during the
interaction between the user and the on-board instruments; an analyzer
which analyzes the data thus acquired; and an adjustment unit which adjusts
the model for the arrangement of the on-board instruments within the
instrument panel as a function of the data analysis so as to obtain an
adjusted model that is different from the previously constructed model, a
configuration unit which configures information of the first type according to
a
multi-agent cognitive model performs for each on-board instrument:
determining tasks relating to the use of the vehicle and performed by the on-
board instrument under consideration, determining agents of the multi-agent
CA 02615250 2013-02-14
5b
cognitive model relative to the determined tasks, and establishing a link
between the determined agents of the cognitive model and identified
functional zones of the on-board instrument under consideration.
This system has the same aspects and advantages as those presented
hereinabove with regard to the method, and they therefore will not be repeated
here.
CA 02615250 2008-01-14
6
Other characteristics and advantages will become apparent in the course
of the description hereinafter, given solely by way of non-limitative example
and
referring to the attached drawings, wherein:
¨ Fig. 1 a generally represents an algorithm of the method of determining
an interface model according to the invention;
¨ Fig. lb schematically represents a system for use of the method
according to the invention;
¨ Fig. 2 schematically represents the process of construction of the
interface model according to the invention;
¨ Fig. 3 schematically represents an algorithm detailing the steps
illustrated in the algorithm of Fig. I a;
¨ Fig. 4 is a table illustrating the correspondence between a flight
procedure and the on-board instruments used in each step of the
procedure;
¨ Fig. 5 illustrates the identification of different information zones on
an
on-board instrument;
¨ Fig. 6 schematically illustrates the functions assigned to the zones
defined in Fig. 5;
¨ Fig. 7 represents, in table form, the link between the agents of the
cognitive model and the functional zones of the on-board instrument
illustrated in Fig. 5;
¨ Fig. 8 illustrates an example of construction of tables 16 and 18 of Fig.
la.
The invention finds a particularly advantageous application in aeronautics,
especially in the modeling of interface elements of an airplane cockpit.
Several types of instrument panel are found in an airplane cockpit, for
example the principal instrument panel IP (or "main instrument pane!' in
English),
on which there are disposed several on-board instruments acting as interface
CA 02615250 2008-01-14
I
. ,
7
elements for the pilot known as PF ("Pilot Flying" in English) and the copilot
known as PNF ("Pilot Non Flying" in English), specifically, for example, the
instrument known as PFD ("Primary Flight Display' in English) and the
instrument known as ND ("Navigation Display" in English). There are also found
a central panel CP ("central panel" in English), an upper panel OP ("overhead
pane' in English) and a panel below the windshield GS ("glareshield panel" in
English).
The cockpit user, or in other words the pilot, uses all the interface
elements of the aforesaid instrument panels to carry out control of the
airplane,
navigation tasks and protective tasks to keep the airplane flying.
In order to facilitate execution of these tasks by the pilot and to permit him
to perform his activity with maximum safety, it has proved useful to determine
an
interface model between the pilot and his environment on board the airplane.
The algorithm of Fig. la illustrates the principal steps of the method
according to the invention for determining such a pilot-cockpit interface
model.
This algorithm is executed by a computer functioning in cooperation with
data/information storage means (databases, memories, etc.).
In the course of a first step El, it is provided that an interface model of
the
cockpit will be constructed on the basis of two types of information, a first
type of
information relating to the technical system and more particularly
representative
of interface elements of the cockpit, and a second type of information
relating to
the human and more particularly representative of the knowledge that a pilot
has
concerning the use of interface elements of the cockpit and the flight
procedures,
as well as of his behaviors (experience of the airplane pilot).
The pilot-cockpit interaction is based on interfaces having dynamic
character, including the behaviors of the user and of the technical system.
CA 02615250 2008-01-14
-
8
This step is based on the use of information of both technical origin and
human origin in order to take into account the user-technical system pair
during
construction of the interaction model.
As illustrated in Fig. 2, the aforesaid two types of information are
furnished to a dynamic database 10 endowed with a pilot (human)-technical
system structure, symmetric relative to the interaction axis separating the
base
part 12 relating to the human aspect and the base part 14 relating to the
technical aspect.
It will be noted that the information is deposited in this database in a
manner structured according to an identical configuration which is defined,
according to each aspect (human and technical), by an input-output level
detailing all the inputs and outputs used and by a processing level detailing
the
different subsystems used.
The model is constructed beginning with identification of the inputs and
outputs of the human side and of the technical side, after which subsystems
are
identified at the information processing level.
The symmetric construction of the model of interaction between the
human-technical system makes it possible to apply the same methods to all
entities present. Since the technical system as well as the human are
envisioned
as complex systems and are broken down analogously into subsystems (in other
words, if vocal alarms (belonging to the vocal subsystem) and graphic alarms
(belonging to the graphic subsystem) on the technical system side are
considered), it is necessary to consider the subsystems by means of which the
human will perceive, be aware and process these alarms: on the human side
these subsystems are assimilated with auditory and visual modes, with
attention,
with the system for processing symbols, with short and long term memory and
with decision making.
The information of technical origin (first type) and of human origin (second
type) are configured in identical manner according to the same multi-agent
cognitive model, and the known UML language ("Unified Modeling Language" in
English) is used to formalize the pilot-cockpit pair.
,
CA 02615250 2008-01-14
9
In the multi-agent cognitive model, there will be defined agents making it
possible to describe the knowledge processes of the pilot in relation to the
interface elements of the cockpit.
This multi-agent representation is particularly suitable for describing
processes that can evolve simultaneously.
In fact, a pilot may have to analyze visual information (at the input to the
human side and at the output from the technical system side) at the same time
as auditory information, such as sound alarms.
This multi-agent representation is also particularly suitable for the case of
tracking information that follows a sequential path and that can arise between
different independent interface elements.
Furthermore, this representation is also useful for establishing an
appropriate hierarchy and classification of information for the purpose of
facilitating later analysis of data representative of human activities
involved in the
interaction between the pilot and interface elements.
In cognitive modeling based on agents and resources, the agents of the
cognitive model are determined by their roles, their responsibilities, their
resources or functions and the objectives to be attained.
According to this multi-agent approach, the area of application or in other
words the use of interface elements of the airplane cockpit is analyzed in
terms
of needs to be satisfied in a given context.
The agents are oriented by one objective, and they make it possible to
take into account what is wished relative to the constitutive scheme of the
pilot's
thought processes.
For example, the pilot believes that, to change flying level, he needs a
certain number of conditions in order to assure that his maneuver will be
successful: visibility, status of the engines, atmospheric conditions...
The pilot therefore wishes to obtain such information in order to be able to
accomplish his task, and he will therefore use the cognitive resources
furnished
to him by the interface elements (on-board instruments).
In this way he supplements his awareness of the situation and can
visualize the future and act accordingly.
CA 02615250 2008-01-14
All of these aspects therefore will be amenable to being evaluated in an
experimental framework based on multi-agent cognitive modeling.
These agents that contribute to cognitive processes involve perception,
comprehension and mental representation of the interface elements of the
cockpit.
Thus each agent satisfies the objectives established by means of action
plans, which in aeronautics, for example, are procedures defining the use of
on-
board instruments in the crew operating manual known as FCOM ("Flight Crew
Operating Manual" in English) which provides in particular for review of
different
checklists, of landing and takeoff phases, etc.
On the user's (pilot's) side, these action plans correspond to the mental
representation that the user has of written flight procedures. This
representation
varies as a function of experience.
As already mentioned hereinabove with reference to Fig. 2, the cognitive
architecture is based on two principal levels, or in other words the input-
output
level and the information processing level.
Agents are classified by level (input-output or processing) and by type
(input-output channel or processing system).
Thus several types are encountered on the same level: agents of the
visual type, agents of the auditory type, etc. are available at the input-
output
level, and attentional agents, mnemonic agents, decision-making agents, etc.
are
available at the processing level.
As indicated hereinabove, the agents are characterized by one or more
roles, responsibilities and resources.
More particularly, the role of an agent is defined relative to a task or sub-
task (for example, relative to control of the vehicle) that must be performed.
The responsibilities of the agent are to execute the task or sub-task, and
the resources used permit effective execution of the task or sub-task.
CA 02615250 2008-01-14
_
11
Thus, for example, a three-dimensional scene can be represented by a
set of agents, each of which is in charge of a particular characteristic of
the
scene, such as relief or textures. The textures correspond to the grid density
of
the relief, which can be variable or constant depending on the terrain
database,
that is, the grid elements can have the same size everywhere or that the grid
elements can have different sizes depending on the relief zones represented,
the
colors and the symbology.
As for all agents, the resources of these agents are classified by level
(input-output or processing) and by type (input-output channels or processing
system).
Thus, for example, the relief of the aforesaid three-dimensional visual
scene that can be represented by one agent may draw on varied resources used
for detecting and analyzing the valleys, rivers, woods, roads, building
structures,
etc. of the visual scene.
The determination of the agents of the multi-agent cognitive model is
performed according to the steps of the method indicated hereinafter, which
steps are carried out iteratively by means of two approaches, the approach
going
from the top to the bottom, known as "top down", and the approach from the
bottom to the top, known as "bottom up."
The "top down" approach is based on facts that may be known about the
pilots as well as on the manner in which they use the interface elements of
the
cockpit, and it can facilitate the classification into agents.
The "bottom up" approach is based on the interface elements of the
cockpit and the visual indications, which are grouped in order to reveal
responsibilities and agents.
CA 02615250 2008-01-14
,
12
Top down
1 ¨ 1. Identification of tasks
¨ 2. Identification of subsystems used to perform each task
¨ 3. Identification of agents within each subsystem
¨ Identification of links between the agents
¨ Identification of resources of each agent
¨ Identification of links between the resources and the other agents
¨ 3. Pairing of categories and agents
I
¨ 2. Grouping of resources into categories
¨ 1. identification of resources linked to elements of the visual scene
Bottom up
The modeling of the cockpit according to this multi-agent cognitive model
makes it possible to define the elements of the visual scene at a fine level
of
granularity, which takes into consideration constitutive elements of each
interface
element (on-board instruments), or in other words information zones of these
interface elements, and not each interface element as a set (coarse level of
granularity).
Within the scope of this model, the resources of agents defined in this
way are assigned to the processing of interface elements.
In general, the formalization of the pilot-cockpit pair is not confined to
representing disparate entities, but proposes to define links between these
entities, as represented in Fig. 2, by organizing the entities in the form of
tables
16, 18 containing resources, agents, linking agents and plans, both on the
human side and on the technical system side. It will be noted that the linking
agents make it possible to define direct links to specific resources of
another
agent. Without these linking agents, it would be possible to link only agents
and
not resources to agents.
As illustrated in Fig. 2, the modeling of the technical system is
represented on the left of Fig. 2 by the modeling of PFD interface element 20,
which will be detailed hereinafter, while on the right side of this same
figure there
is represented the architecture of the cognitive modeling of human side 22 on
. CA 02615250 2008-01-14
,
13
the two main levels, or in other words input-output level 24 and level 26,
where
information processing takes place.
Each of these levels can be broken down into several subsystems, such
as that of vision, audition, language and motor function for the first and
that of
attention, in English "long term memotY' (LTM), "work memorY' (WM) and
"decision making", for the second.
As soon as the interface model has been constructed on the basis of the
two types of information (information representative of interface elements and
information representative of human knowledge and behavior with regard to the
use of interface elements), the algorithm of Fig. la provides a data
acquisition
step E2.
In the course of this step, there are acquired data representative of one or
more human activities (such as vision, speech, audition, movement of human
limbs, kinesthesia, physiological manifestations and reactions of the human
body, etc.) that are involved in the pilot's interaction with the interface
elements.
Thus, for example, at a given instant, the pilot not only gazes at a zone of
an interface element of the cockpit, the item or items of information being
detected by an oculometric apparatus and automatically integrated into a
results
database, but also acts simultaneously on the control column and/or on other
devices, the corresponding item or items of information being collected by a
video or other recording system and also being stored.
It will be noted that, depending on the nature of the human activity in
question, there is used a suitable data acquisition apparatus (oculometer,
video
recorder, electrodermal sensor, etc.).
After these data have been acquired, they are studied in the course of the
following step (step E3), for example by the expert pilot or pilots who have
undergone the test aimed at in step E2.
CA 02615250 2008-01-14
14
In the course of analysis of the acquired data, the test subject examines
the results and proceeds to interpret them by trying to determine whether an
action he took at a given instant of the test was quite appropriate and
whether it
took place at the right time.
More generally, he explains the relationship between gathering of
information/absence of gathering of information and action/absence of action.
In the course of interpreting these results, the subject of the test
determines, for example, why his gaze followed a given eye movement over one
or more consecutive interface elements and/or over one or more zones of the
same interface element.
Depending on the results of this analysis and on the interpretation thereof
by the subject of the test, and if appropriate by other experts of different
disciplines, it is possible to validate the modeling framework for the pilot-
cockpit
interface just as it was constructed or to adjust it.
Thus, for example, it may be found that an interface element is lacking
that would permit the pilot to complete his control, navigation or other task
successfully, or else that a navigation interface element (display, etc.) is
lacking.
It may also be found that the granularity level adopted during construction
of the interface model is too fine and therefore little representative of the
real
context, or on the other hand that the granularity level is too coarse and
therefore
does not permit sufficient pertinent information representative of this
context to
be obtained.
The interpretation of the test results also makes it possible to reveal
dysfunctioning of interface elements or of flight procedures.
As an example, this may be observed after considerable fatigue and
elevated stress of the test subject has been recorded. It is therefore
possible to
improve the interaction model as a function of the results of step E3.
In this way an iterative procedure is followed by repeating the loop
represented in Fig. la between step E4 and step El until there is obtained the
CA 02615250 2008-01-14
desired interaction model that is as representative as possible of the
environment on board the airplane.
As soon as the interface model has been determined by the method
according to the invention in conformity with the set objectives, then this
validated model can be used, for example, for training future pilots in a
flight
simulator or else for improving interfaces proposed by the system (layout,
sequencing of information, spatial and multi-modal redundancy, etc.).
It will be noted that Fig. lb represents a system 30 for determining a
model according to the invention, representative of the interaction between
user
32 and interface elements 34. This system comprises a computer 36 having
inputs and outputs for cooperation with user 32 and interface elements 34, as
well as with a data acquisition apparatus 38 (such as an oculometric
apparatus),
which transmits the acquired data that are to be analyzed to computer 36.
The algorithm of Fig. 3 illustrates in more detail the steps of the algorithm
of Fig. 2 by showing the symmetric formalization of the pilot-cockpit pair.
Construction of the interface model of the technical system side begins
with a first step E10, in the course of which there is established a link
between
the flight procedures defined in the FCOM manual and the interface elements of
the cockpit (on-board instruments such as PFD, ND, etc.) that the pilot (PF)
and
copilot (PNF) must consult for each action described in the flight procedure
in
question.
Among these procedures, there are found in English the "takeoff'
procedure, the post-takeoff procedure, in English the "climb" procedure, the
"cruise" procedure, the "descent preparation" procedure, the "descent"
procedure
the "standard approach" procedure, the "non-precision approach" procedure and
the "landing" procedure.
CA 02615250 2008-01-14
16
By associating the on-board instruments involved in each action described
in the "climb" flight procedure of the control manual of an Airbus A340, there
is
obtained the table illustrated in Fig. 4. showing, for example, that the pilot
must
consult the instrument known as ECU ("Flight Control Unit" in English) of the
GS
panel in SET-value mode and the PFD instrument of the main panel IP in
CHECK mode in order to read the BARO indication (barometric reference).
Similarly, while climbing, the pilot must consult the PFD instrument of the
main panel to see the speed and altitude information as well as the airplane's
altitude.
As soon as the flight procedures have been linked to the cockpit interface
elements in question, the algorithm of Fig. 3 provides a subsequent step E12,
in
the course of which information zones of each interface element of the cockpit
are identified and functions performed by these zones are determined.
As an example, the different information zones on the interface element
known in English as the PFD ("Primaty Flight Display') are identified with
reference to Fig. 5.
This figure is divided into two parts: the PFD interface element is
represented in the left part and the different information zones of this
interface
element as well as their location thereon are identified in the right part.
In this way there are counted nine information zones marked by the
numbers 1 to 9 in the right part of Fig. 5, and which will be designated
hereinafter
by the references Z1 to Z9.
After the zones of each interface element have been identified, the roles
and responsibilities (functions of different zones, taking into account tasks
and
sub-tasks that relate to control of the aircraft and wherein each interface
element
is used) are determined in the course of the following step E14. On the basis
of
this determination of the roles and responsibilities of zones, it will be
possible to
determine the agents of the multi-agent cognitive model.
,
= CA 02615250 2008-01-14
17
Thus, for example, for the PFD interface element, there are distinguished
three fundamental tasks, which are control of the airplane (Ti), navigation
(T2)
and protection to keep the airplane flying (T3).
Within each of these three tasks, it is possible to determine more precise
sub-tasks:
¨ indicating values of parameters of the airplane (T11),
¨ indicating selected values or points (originating from the FMGS: "Flight
Management and Guidance System" in English) (T12),
¨ indicating the flight tendencies (T13),
¨ giving the indications of radionavigation instruments and of the FMGS
(T21),
¨ permitting the indications furnished by the FMGS to be followed easily
(T22),
¨ presenting the flying limits (T31), and
¨ alerting (T32).
As soon as these tasks and sub-tasks have been determined, the role
and responsibilities of the different zones of each interface element and, for
example, of the PFD are determined.
In Fig. 6 there are identified and represented, in a table, different
functions or responsibilities of previously identified zones.
Thus, on the basis of zone Z1 designated as "FMA" ("Flight Mode
Annunciator" in English), it is possible to identify four sub-zones that
furnish
information on the control mode (for example, automatic pilot mode) and on
radionavigation.
Zone Z2 designated as "VA", furnishes information on the airspeed and
can be divided into two sub-zones.
,
,
' CA 02615250 2008-01-14
18
Zone Z3 designated as "AA" and which can be divided into two sub-zones,
furnishes information on the attitude of the airplane (pitch, trim, roll,
guidance,
control column, etc.).
Zone Z4 designated as "ANv" and which can be divided into three sub-
zones, serves as altimeter and furnishes information on the vertical speed of
the
airplane.
Zone Z5 designated as "ILS-GS" (ILS for "Instrument Landing System"
and GS for "Glide Slope" in English) furnishes information on the vertical
position
of the instrument landing system ILS relative to the slope GS.
Zone Z6 designated as "ILS-Loc", furnishes information on the ILS
horizontal position relative to the, in English, "localize(.
Zone Z7 designated as "Mil", furnishes information on the Mach number
of the airplane as well as navigation information.
Zone Z8 designated as "H/T" ("heading/track zone" in English), furnishes
information on the guidance and heading of the aircraft.
Finally, zone Z9 designated as "Ref/Alt", furnishes information on the
altimetric reference.
It will be noted that the names of the zones are derived from the definition
of the agent's role, which will be defined later.
By virtue of the table of Fig. 6 and of the determination of tasks and sub-
tasks, it is possible in the course of the following step E16 to determine the
cognitive agents used to build the cognitive model according to the criteria
related to control and navigation.
Returning to the example of modeling of the PFD interface element, there
are determined the cognitive agents that make it possible to describe the
cognitive processes of use of different zones of the PFD interface element as
represented in Fig. 7.
Thus there are determined agents related to the analysis of the vertical
motion (altitude, V/S), to the analysis of the horizontal motion (speed and
,
, CA 02615250 2008-01-14
19
heading), to the analysis of attitude A/C, to the tracking of FMGS
instructions, to
the orientation/ILS, to the FMA, to the color code and to the alert.
As indicated in the "agent responsibilities" column, for example, agent Al
has the role of analyzing the vertical motion of the airplane by focusing on
the
parameters of altitude and vertical speed and to perform this role, it is
responsible for the values of the vertical parameters and for the symbols of
these
parameters.
To perform this role, agent Al relies on four cognitive resources related to
the responsibility of the values of the vertical parameters on the one hand
and on
two cognitive resources related to the responsibility of the symbology on the
other hand. This permits the agent to achieve the tasks that are related
mainly to
control of the apparatus (T11 and T12) and that are localized in zone Z4 of
the
PFD interface element.
In the course of the following step E18, the system inputs and outputs are
identified relative to the use context, in other words, information proposed
by the
system (interface elements such as the PFD) is identified at a given instant
relative to a given use situation, such as takeoff or climb.
In the course of the following step E20 of the algorithm, it is provided that
system information (such as interface element PFD) situated at the processing
level will be identified.
Fig. 8 illustrates in detail the construction of tables 16 and 18 of Fig. 2
according to the structure comprising plan, linking agent, agent and
resources,
both for the technical system side and the human side, during altitude
monitoring
relative to instrument PFD.
Thus, on the technical system side (Table 18), it is determined, within the
framework of a plan relating to the "climb" procedure illustrated in Fig. 4,
that the
resources employed are zones Z4 and Z9 of the PFD (Fig. 7), the agent is agent
Al of the PFD and the linking agent is agent A3 of the on-board instruments
known in English as "Electronic Flight Instrument System" (EFIS).
Table 16 (human side) will be described later.
CA 02615250 2008-01-14
,
In parallel with the description that has just been given in the course of
steps E10 to E20, there is established an interaction model of the human side
with reference to steps E22 to E28, which will now be described.
Information of human origin is furnished, for example, by means of
interviews with experts in control or in flight procedure. In the course of
these
interviews describing given situations (in other words, use of instruments,
such
as ND and PFD, that present two-dimensional information, compared with the
use of an instrument that would present the same information directly in three
dimensions), the experts are asked to indicate the actions that they would
envision taking, the checks to be performed, the information that they would
need in order to act, etc.
In the course of a first step E22, it is provided that there will be
identified,
at the input-output level of the interface model on the human side, modes of
interaction with the technical system, meaning, for example, the input-output
channels constituting human vision, human language, audition, kinesthesia,
etc.
In the course of this same step, there are also identified the necessary
resources for undertaking the appropriate maneuver, or in other words, for
example, for perceiving (gazing at) the altitude information furnished by the
corresponding zone of interface element PFD, hearing (listening to) the
auditory
"TERRAIN" alarm ("call-out" in English) (meaning that the airplane is outside
the
safety zone relative to the terrain, or in other words is too low), pulling
back on
the control column or else opening the throttle once again.
It is appropriate to note that it is also optionally possible to remain
confined to the input-output level in restricted use situations such as
takeoff, for
example. Such situations are distinct from flight phases and therefore from
specific procedures or subsections of procedure. It is possible to examine
these
specific phases in depth by studying how they proceed under difficult
conditions
¨ bad weather, engine failures, defects in the presentation of information,
user
stress or fatigue.
In the course of the following step E24, it is provided that the multi-modal
interactions at the input-output level of the human cognitive model, or in
other
words the interactions between the different channels (vision, audition, etc.)
will
,
be identified.
CA 02615250 2008-01-14
21
Thus, for example, there are identified the interactions between the
different previously identified modes on the basis of cases surveyed during
interviews, such as that consisting in perceiving altitude information on the
interface element in question and hearing the auditory alarm and thus pulling
back on the control column.
In the course of the following step E26, it is provided that the processing
level of the cognitive model on the human side will be defined.
This is achieved by providing on the one hand that those actions to be
taken by the pilot and/or those decisions to be made that have proved
particularly difficult or delicate to implement will be identified in relation
to how
they are modeled by the technical system side, and on the other hand that
hypotheses on the processing applied to these sensitive or difficult parts
will be
established.
Thus, for example, it is supposed that the user will make the right decision
relative to the visual and auditory information that the interface elements of
the
technical system furnish to him in the collision alert case.
In the course of this same step E26, there is identified the processing of
information according to the different previously identified modes (input-
output
channels).
Table 16 representative of the modeling of the human side corresponding
to table 18 of the technical system side is constructed, within the framework
of
monitoring of the airplane's altitude relative to the PFD instrument, on the
basis
of the defined plan, or in other words the use of the PFD and the control of
the
airplane.
In this table the resources used at the input-output and processing levels
are determined/identified.
Thus there are identified the visual inputs and outputs, or in other words
the altitude monitoring furnished by the PFD and the corresponding processing,
or in other words work memory (WM) and long term memory (LTM) as well as
decision making.
The corresponding agent is the "PFD" and the aforesaid resources are
linked to the "flight plan tracking" agent.
In this way the multi-agent pilot-cockpit cognitive model is constructed
symmetrically.
CA 02615250 2008-01-14
22
The following step E28 makes it possible to supplement the human
cognitive model and to validate it with an expert in the discipline
(specialist in
cognitive psychology, in physiology, in language).
In the course of the following step E30, it is provided that the model
representative of the human-technical system (pilot-cockpit) pair will be
validated
with experts of the different disciplines involved, or in other words experts
in flight
procedure, pilot experts, designers and experts in human factors (experts in
vision, audition, language, kinesthesia, etc.).
It will be noted that steps E28 and E30 may optionally be combined.
Once the model has been constructed, there is initiated the previously
described step E2, in the course of which the methods of human factor analysis
are used in order to collect data reflecting corresponding human activities by
means of an experimental protocol.
For this purpose it is possible to employ several analysis methods, as
indicated hereinabove, wherein visual data representative of the pilot's eye
movements in the course of time over one or more of the interface elements of
the cockpit are acquired by means of an oculometric apparatus (more
particularly, the position of the pilot's gaze is tracked from one zone of one
interface element to another zone of another interface element), video data
representative of the movements of the pilot as he acts, for example, on the
control column, are acquired by means of the video system of the cockpit,
and/or
auditory data are acquired by means of an audio recording apparatus.
By virtue of the framework or model defined in the course of the preceding
steps, it is possible to link, for example, the two types of human activity
data
(oculometric data and data relating to the motor function of the human body),
inasmuch as a common receptacle (database) has been positioned to receive
the two types of information of technical origin and human origin.
It will be noted that the experimental protocols comprising the basis of the
different human factor evaluations are derived from the previously constructed
CA 02615250 2008-01-14
23
interaction model, and they also supply this model by the results that they
produce.
Furthermore, it will be noted that the use of a common receptacle capable
of cross-referencing methods and evaluations ensures coherence, homogeneity
and traceability of the collected data.
More particularly, oculometric apparatus 38 of Fig. lb makes it possible to
record the position of the pilot's gaze at a visual scene, thus making it
possible to
track the different visual elements traversed by the pilot's gaze at the
interface
elements of the cockpit as well as at the external visual scene.
The oculometric apparatus includes an analog device, or in other words
the oculometer, which records the movements of the pilot's eye. The oculometer
includes three components, specifically a camera recording the eye movements,
an infrared source emitting an infrared beam into the eye, and a camera
recording the visual scene viewed by the pilot.
Thus the video data acquired by the camera recording the eye
movements and the video data acquired by the camera recording the visual
scene viewed by the pilot are superposed and the position of the pilot's gaze
is
represented by a pointer (such as a circle or a crosshair), which moves over
the
visual scene.
The use of the oculometer alone, although sufficient for the external visual
scene, does not provide sufficient precision if it is desired to record
particularly
fine details of the pilot's eye movement, for example reading of texts or
gathering
of information on specific zones of screens.
A magnetic field generator is therefore associated with the oculometer in
order to impart maximum precision.
The magnetic field generator is used as a frame of reference in three-
dimensional space to sense the position of the pilot's head relative to the
coordinates of the different surfaces and planes that make up his real
environment. In this regard, the surfaces and planes in question are those
corresponding to, the screens and to the control panels of the cockpit
comprising
' CA 02615250 2008-01-14
=
24
regions of interest, which themselves can be broken down into zones and sub-
zones of interest, as was seen hereinabove, for each interface element.
To analyze the head movements of the pilot, there are therefore used a
magnetic field generator and a receiver fixed to the pilot's head and these
components, in combination with the aforesaid analog device (oculometer),
make it possible to obtain maximum precision of the position of the user's
gaze
at a visual scene.
More particularly, the receiver fixed to the pilot's head furnishes the exact
position of the head in the three-dimensional model.
The distance between this head receiver and the camera recording the
scene, as well as the distance between the head receiver and the eyes of the
pilot, is then introduced into the three-dimensional model. The first of the
aforesaid distances is necessary to achieve calibration of the camera relative
to
the scene, and the second of these distances is necessary to calibrate the
analog device (oculometer).
It will be noted that the adaptation of the aforesaid oculometric apparatus
to the cockpit, in order to impart maximum precision by combination of data
furnished by the position of the pilot's head and data furnished by the
position of
his gaze, takes into account the geometric study of the cockpit and the study
of
the posture of the pilot.
While performing the geometric study of the cockpit, the Applicant noticed
that, for installing the magnetic field generator on a support in the cockpit,
it was
advisable to ensure that the distance between the generator and any metal
surface was sufficiently large to minimize the magnetic interferences that can
be
produced with the oculometric apparatus.
In addition, while configuring different components comprising the
oculometric apparatus inside the cockpit, the Applicant found that the
distance
between the magnetic field generator and the receiver for the position of the
pilot's head had to be strictly shorter than the distance between the receiver
for
the position of the pilot's head and any metal surface, again to reduce the
magnetic interferences as much as possible.
CA 02615250 2008-01-14
,
It is appropriate to note that the postural study of the pilot makes it
possible to define the limits of his volume of movement and therefore the
distances between the head receiver and the magnetic field source.
By virtue of the aforesaid oculometric apparatus, it is possible to record
very precisely the ocular movements (behaviors), such as the fixations,
saccades
and pursuits that characterize the way in which the pilot gazes at the
specific
elements of an aeronautical visual scene (on-board instruments and exterior
visual scene). The constituent components of an oculometric apparatus, or in
other words the analog device, the magnetic field generator and a helmet
supporting the head receiver, can be obtained from Senso-Motric Instruments
GmbH, Warthestrasse 21, D-14513 Teltow, Germany.
As already indicated hereinabove, in the course of step E3 following the
data acquisition step, such data are analyzed with the test subject or
subjects
(pilots) in order to verify the coherence and reliability of the test results.
Thus, according to an example borrowed from the automobile industry
(since the invention can in fact be applied to disciplines other than
aeronautics),
by using an oculometer in a teaching motor vehicle, the instructor and the
student, by viewing the video data recorded with the oculometer once the
lesson
has ended, can better understand why the student did not look in the rear-view
mirror before turning.
After all of the data collected in the course of step E2 have been analyzed
and interpreted in the course of step E3, they are then validated at a first
intradisciplinary collective level with the experts of the discipline in
question (for
example, a population of pilots in the case of aeronautics), and are then
validated at an interdisciplinary collective level with the experts of
different
disciplines (human factor experts, engineers, pilots), in order that these
data will
be shared with all interested parties.
Thus the test data are explained and shared at three levels, an individual
level, an intradisciplinary level and an interdisciplinary level.
This validation with the experts makes it possible to return to the definition
of the framework determined during the first steps (construction of the multi-
CA 02615250 2008-01-14
26
agent cognitive model) and to adjust and refine the model as a function of the
test results and of the interpretation thereof by the experts.
As soon as the model is validated, it is possible to deduce therefrom
potential improvements to the pilot-cockpit interface elements and to the
procedures for using these interface elements (such as flight procedure,
etc.), or
to use this model to teach the pilots to train themselves in the interface
elements
of the cockpit.
By way of example, the method according to the invention makes it
possible to determine at which instant a display system mounted high up above
the pilot's head ("head up display' in English) should be used to optimize the
use
thereof. The method according to the invention also makes it possible to
determine whether such a display system is actually being used by the operator
in a particular type of vehicle.
In another example, the method according to the invention makes it
possible to check that the pilot is mentally constructing a three-dimensional
visual representation of the position of his vehicle in space, and is doing so
solely on the basis of two-dimensional information furnished by on-board
instruments.
The method according to the invention can then be used as the basis for
designing a new instrument that furnishes a three-dimensional visual
representation of the position of the vehicle in space.
The method is particularly advantageous for determining which
information furnished by interface elements of the on-board panel is actually
useful.
In fact, by virtue in particular of the acquisition and analysis of data, for
example oculometric data, the method makes it possible to separate information
indispensable to the user from information that is not particularly useful or
is
even redundant.
