Threat of human faults danger in car-driver interactions

Threat of human faults danger in car-driver
interactions
Mirko Novák
Conference WSEAS, Prague, March 13, 2005
1
Prof. Ing. Mirko Novák, DrSc.,
Czech Technical University in Prague,
Faculty of Transportation Sciences,
Department of Control Engineering and Telematics,
Joint Laboratory of System Reliability,
11000 Prague 1, Konviktská 20, e-mail: [email protected]
Abstract:
Human activity has to be considered as the less reliable
component of all the human subject – artificial system
interactions. This concerns also all the contemporary
transportation systems. The losses cause by human faults in
transportation reach extreme values and have tendency to
increase year by year.
A significant part of these losses comes from the decrease
of human subject attention when operating or using the
transportation system, namely driving the car.
Another considerably large part of these losses has its
origin in splitting of the driver attention between the functional
and marginal stimuli.
Also the masking of some minor stimuli by those, which
dominate in the set observed by driver, can cause serious
accidents.
In this key-note a brief discussion of these factors, having
serious negative influence on transportation reliability and
safety is mentioned and some ways for its minimization are
proposed.
1. Introduction
The very long experience with human beings
dealing with artificial systems leads to the regrettable
observation, that the human factor is very often the
weakest and most dangerous point in the interaction.
The reason for this is quite natural and easy to
understand:
– Unlike artificial systems, humans cannot operate
indefinitely without a break – human beings need to
relax, rest and sleep. Decreases of human vigilance
and attention while operation or using an artificial
system has always been - and still is - the most
frequent cause of system failures, accidents and
catastrophes.
- Another source of system failures lies in the
possibility that the human operator (or user) of a
particular artificial system may react too late, and
that his/her decision and reaction may be incorrect.
- No human subject is identical. Human behavior
is not fully deterministic; it varies from subject to
subject.
All these factors combine, with the result that the
reliability of human subject – artificial system interaction is
very limited, before all from the human side.
The price that we all pay for insufficiently
reliable and safe human – artificial system
interaction is tremendous.
If one takes the statistical data on road traffic
accidents into account, one finds that only in the
EU there were in 2002 rougly 42000 mortal
accidents per year, which represented the losses
of about 165 billion Euro.
The estimations how much of these losses
are caused by unsatisfactory level of human
operator (driver, dispatcher etc.) attention differ
according to used methodology – which is stil not
standardized, unfortunately – but can be taken
between 15 and 50%.
If we take the lower limit into account, we can
speak on losses about 25 billion Euro in the EU, which
are caused by low level of drivers attention.
Moreover, this losses have the tendency to
increase in time and also - surprisingly - with the level
of the the power and sophistication of the respective
transportation system.
We therefore urgently need to counteract this
human subject unreliability in interaction with artificial
systems.
Investigating the reliability of interaction
between a human subject and an artificial system,
we are interested before all in such situations, when
a particular human subject is exposed to the
influence of various external stimuli.
These situations are typical namely for the
human subject interaction with transportation
vehicle (car) and/or with the whole transportation
system.
A set of such external stimuli can be
characterized as
•
very variant,
and
•
changing significantly under influence of
many independent variables, namely
the
time.
The signals, with which the driver has to interact
when driving the car on the road, can be classified
in the following main groups:
»
»
»
»
»
»
a) Visual signals representing the observed external
scene of the road and its nearest environment, which
can have the influence on the situation in the front of the
car or on both its sides;
b) Visual signals observed by driver in the rear and
both side mirrors;
c) Acoustic signals concerning the traffic on the
neighborhood of the car;
d) Acoustic signals coming from the car body and car
engine;
e) Acoustic signals coming from the tires and wind;
f) Signals representing the drivers interaction with the
car control tools and auxiliary cockpit equipments
(communication and navigation before all);
g) Signals coming from the driver interaction with the
car crew.
2. Attention decreases
From the above-mentioned 7 groups of signals,
representing the most important stimuli influencing
the driver, only the first 3 can be considered as
directly related to driving activity.
Let denote them by word “functional stimuli” Sf.
The other 4 groups of stimuli have no such direct
relation to driving reliability and we shall call them as
“marginal”.
•
•
•
•
When driver observes the functional stimuli in the
course of his/her driving,
he/she becomes subsequently tired,
his/her attention level LAT decreases,
reaction time TR prolongs and the
probability Pcorr of his/her correct response on certain
stimulus of the group Sf decreases.
This increases, of course, the danger of an
accident.
The resistance of the driver against the influence
of attention decrease is very individual,
however it can be
•
measured, classified and by
•
special training methods (namely by those, based
on the so- called bio-feedback technology) it can be
also significantly improved.
Besides this, there is possible to construct various
warning systems, the aim of which is
• to continuously measure the level of driver attention in
the course of driving,
• to predict its further development and
• to inform and to warn the driver if the decrease of
his/her attention falls under certain limit.
Because the direct measurement of the values LA
in the course of driving is quite problematic, various
indirect attention level indicators are used.
Among them, those based of the sophisticated
analysis of drivers EEG (electroencephalographic)
signals play the dominant role (because allow the
most specific, reliable and fast determination of the
actual state of driver attention). Though the attention
classifiers and predictors derived from EEG analysis
are very individual and must be developed for each
particular driver specially, they operate considerably
reliable for long time (up to few years) if the respective
human subject does not undergo some mental
decease or accident.
While various warning systems appearing on the
market are based usually on the so called secondary
attention level indicators (eye movement, skin
impedance, face grimaces analysis etc.), only those,
using the brain electromagnetic radiation as the key
source of information of his/her attention can be
considered as satisfactory specific and reliable.
The presence of any marginal stimulus in the set observed
by driver can lead to the splitting of his/her attention, so that
the part of it, oriented to functional stimuli Sf related directly to
driving decreases.
Another way for improving the driving reliability is therefore
to modify the arrangement of the car-cockpit interior so that the
influence of the marginal stimuli Sm on the driver attention
splitting is minimized.
However, because each human subject has in any instant
only a limited capacity of his/her attention at disposal, the
presence of signals belonging to any of marginal group lead to
splitting of the driver attention into two parts:

that, which is immediately concentrated to driving and

that, which is consumed by the sum of all the marginal
signals.
The total attention disposable capacity CAT of the particular
driver driving certain car at some specific situation can be
expressed as
tf
………………………..(1)
C AT   LAT  t dt
t0
The value of the integral (1) varies from driver to driver and
depends also on his/her physical and psychical condition.
In any case this disposable capacity is subsequently
exhausted, of course not in monotonic manner. In the instant
tk only the part
tk
CrAT  C AT   L AT  t  dt
………………………………….(2)
t0
remains.
The procedure of the driver disposable attention
capacity CAT depends also on the type of the respective car,
on the type and quality of the road, on traffic intensity and on
the whole set of environmental conditions of driving.
Let us denote
the intensity of partial visual signals concerning the transportation
situation in front of the car as sfi,,
and
the intensity of individual visual signals with regard to the marginal
situation as smj.
The total intensity of visual signals coming to the driver’s brain
on the basis of the road observation, can be then expressed as:
in f
Sr =
s
i 1
fi

i  nm
s
j 1
mj
= Sf + Sm, ……………………...(3)
where i = 1…nf is the number of significant signals, and j = 1…nm
is the number of marginal signals.
In reality, the no human subject can observe all the acting
stimuli at once, thus, also his/her short-term memory has to be
taken into account in the course of the entire recognition
process.
Therefore the driver attention is switched into the abovementioned two parts by the influence of Sf and Sm.
The dynamic such attention switching is considerably
complicated and its detail analysis was till now not finished.
In any case, the necessity that driver reacts on some
marginal signal detour his/her attention for some instant out of
the main direction, i.e. the observing of the signals Sf and
reaction on them.
This is the reason, why, as for the driver’s attention, we are
interested in the ratio
 = Sf/Sm,…………………………………...(4)
and also in the ratio
  LAT / LAT
m
, ………………………………(5)
f
representing the corresponding splitting of the drivers attention
level LA.
The attention splitting is not only the result of the impact of
the environment, irrelevant to driving, but also by some other
stimuli, such as the lights of the surrounding buildings, the road
illumination lights, the sun shining at low angles, the lights
reflecting on the wet road surface, road semaphores and
marks, as well as various signs, and, particularly the
advertisements.
Disturbing signals coming from the cockpit interior are to be
taken into account as well. It is obvious that often the number of
these disturbing signals is quite high.
At present many car manufacturers are interested in
improvement of cockpits of their cars with respect to higher
driving safety and reliability.
Many various solutions appear.
An extreme example of such a solution is shown in Fig. 1. It
presents a view of a experimental cockpit design Here, the
driver is forced to deal with many stimuli undesirable for driving.
The main part of them badly distract his/her attention from
focusing on the situation on the road.
Fig.1: Example of a scene loaded with a quantity of signals from the
car interior which distract the driver’s attention from the situation on
the road
Fig. 2: The view of the cockpit of the Skoda Superb car, in which the deviations of
the driver’s view are shown from the standard point of the observation of the
external scene on the road to the views on internal control display (yellow
arrow), navigation display (blue arrow) and radio display (green arrow).
In Fig. 2 is the cocpit of the Skoda Superb car.
Here the situation is much simpler and the driver can
control in much easily than the one shown in Fig. 1.
However, also here the navigation tools, which forms an
un-negligible part of the information content sources, involved
in the cockpit of modern car, has to bee optimized.
In Fig.2 the navigation display is too small, and unsuitably
located – actually, it can only be noticed when the driver diverts
his sight from the situation on the road.
This is evident from the length of the blue vector shown in
Fig. 2., representing the necessary deviation of the driver’s
sight, one counts on, while having a look in the direction of the
navigation display, and, at the same time, fulfilling the activities
of driving. Similar situation occurs when dealing with the car
control display (yellow arrow in Fig. 2) or the radio receiver
panel (green arrow in Fig. 2).
Investigation of the time period needed to distract the driver
attention from the situation on the road to the navigational
system, located in the previously mentioned manner is
extremely important, together with the presented information
recognition, because during this particular time interval the
driver is not able to observe the situation on the road, and, in
fact, operates as if being blinded.
The same applies for the operation of mobile phones within
hand-free sets or radio-receivers as parts of a conventional
arrangement and also for the operation with the buttons and
switches, used for the control of the car and cockpit interior
functions.
In Fig. 4, the schematic sketch shows to which extent the
location of the above mentioned instruments influences the
attention of the driver and his/her reaction time.
Fig. 4: The driver’s interaction with a conventionally located navigation display
(pink arrow) and a radio set (yellow arrow) as compared with the location of
this panel in the left lower corner of the windshield and the respective control
elements at the top of the front panel (blue rectangle and green arrows).
-
Though the proposed attention splitting
measurement methodology seems to be practically
applicable and will with high probability help to the car
designer to receive a considerably objective
recommendation for optimization of the position of the
selected car interior control tool, many open problems
remain.
These concern not only the influence of the
location, type, shape and color of other car cockpit interior
tools, but also the
proper choise of the proband set (using the results of the
analysis of the anamneses questionnaires) and
further refinement of the determination methods.
Also the problem of time (and other independent variables)
dynamics has to be taken into account.
Refferences
[1]…Novák M., Faber J., Votruba Z.:
Neural Network World,
Monography No. 2, Prague 2004
[2]…Faber J., Bouchner P., Hrubeš P., Machan J.,
Nedoma P., Novák M.: Methods for
investigation of driver attention splitting caused
by car cockpit tools
Research Report No. LSS 191/04, CVTU,
Prague, 2004