Content uploaded by John Golias
Author content
All content in this area was uploaded by John Golias on Feb 08, 2014
Content may be subject to copyright.
Content uploaded by Constantinos Antoniou
Author content
All content in this area was uploaded by Constantinos Antoniou
Content may be subject to copyright.
Classi®cation of driver-assistance systems according to their
impact on road safety and tra c e ciency
JOHN GOLIAS*, GEORGE YANNIS and CONSTANTINOS ANTONIOU
Department of Transportation Planning and Engineering, National Technical University of
Athens, Athens GR-15773, Greece
(Received 15 February 2001; revised 11 July 2001; accepted 6 August 2001)
The aim was to examine driver-assistance systems that seem to have a
considerable potential for road safety and tra c e ciency improvement, and to
propose an impact-oriented classi®cation of these systems. A broad overview of a
series of driver-assistance systems under development or in some cases already
available is presented and it identi®es the basic characteristics of each system and
its expected impact on tra c e ciency and road safety. The latter is assessed on
the basis of appropriate evaluation criteria. Expert judgement and literature
evidence available are used in this context. This impact approach, in contrast with
the usually adopted user or system-oriented approaches, allows for more
appropriate identi®cation of the priorities in the ®eld of future research,
development and promotion of driver-assistance systems. The proposed
classi®cation allocates the driver-assistance systems in four diVerent categories
on the basis of whether tra c e ciency and safety impact are high or low. This
categorization reveals that 40% of the systems considered are expected to have a
high safety and low tra c-e ciency impact, while only 15% is expected to have
both impacts high.
1. Introduction
Today, the use of driver-assistance systems is of rapidly growing importance as
these systems are expected to improve road safety, increase road capacity and
attenuate the environmenta l impacts of tra c. The advent of new technologies
supporting vehicle intelligence (sensors, transmitters , communications , computers)
makes the use of driver-assistanc e systems less unapproachabl e to the wide public,
allowing for safer and more e cient driver experiences.
Road safety is inherently associated with any new technology introduced in
vehicles. The negative impacts of road accidents include fatalities, injuries, material
damages and disruption of tra c, and the introduction of appropriate driver-
assistance systems could alleviate to a certain degree these consequences, both in terms
of frequency and intensity (Sala and Mussone 2000). Driver-assistance systems have
the potential to improve road safety by in¯uencing tra c exposure, by reducing the
probability of crashes and by reducing injury consequences. More precisely, driver-
assistance systems support the modi®cation of the driving task by providing
information, advice and assistance, they in¯uence directly and indirectly the behaviour
* Author for correspondence; e-mail: igolias@central.ntua.gr
Trans port Revi ews ISSN 0144-1 647 prin t/ISSN 1464-5 327 onl ine #2002 Tayl or & Franc is Ltd
http://www.tandf.co.uk/journals
DOI: 10 .1080/014 4164011 0091215
TRANSPORT REVIEWS, 2002, VOL. 22, NO. 2, 179±196
of users of both equipped and non-equippe d vehicles, and alleviate accident
consequences by in-vehicle intelligent injury-reducing systems (Naniopoulos 2000).
Simultaneously, the deployment of driver-assistanc e systems is expected to
in¯uence tra c conditions. As such systems become increasingly popular and their
penetration levels increase, the tra c dynamics are expected to change accordingly.
These changes will be re¯ected to a variety of measures, including ± but certainly not
limited to ± tra c capacity of links, mean driving speed and optimized headway
distances. With the insightful use of driver-assistanc e systems, passenger car trips can
be carried out in a more e cient way or shifted to more e cient modes of transport.
This can be achieved, for example, throug h better real-time information for
in¯uencing pretrip choices, as well as oVering in-vehicle information on passenge r
cars and on-trip informatio n on public transport vehicles. Furthermore, better
vehicle control and speed adaptation, as well as a more e cient fuel-saving driving
style (by improved use of accelerator and gear shift), can be achieved by the use of
driver-assistance systems.
The speci®c contributio n of driver-assistance systems on road safety and tra c
e ciency is something still under consideratio n and research. However, several
attempts over the last decade (Brand et al. 1997) already revealed basic trends. Some
systems present a net potential for road safety improvement, while some others have
an eVect mainly on tra c e ciency improvement. In other cases, improvement for
road safety is often accompanied by lower tra c e ciency, whereas e ciency
improvement can sometimes lead to the increase of the number of road accident.
Furthermore, the level and rhythm of penetration of these systems as well as the
implementation policy followed are determining factors for the impact of these
systems to road safety and tra c e ciency.
Until today, several attempts concerning the classi®cation of driver-assistance
systems have been made. Very often, driver-assistanc e systems are classi®ed
according to the technologies (IT, wireless communications, etc.) and subsystems
(autonomous in-vehicle, supported by GPS/GSM communication , linked with road
infrastructure systems) used (EC 1997), or accordin g to the vehicle type (passenger
car, truck, bus) or type of road network (motorway , interurban, urban) they are
referring to. When road safety features are examined, the distinct phases in the
accident process (precrash, crash, post-crash) are often used for the classi®cation of
the driver-assistance systems (Heijer et al. 2000).
When functional analysis of the driver-assistance systems characteristics is
attempted, these systems are initially classi®ed according to the type of user
(individual driver, professional driver, ¯eet owner, elderly drivers, etc.). Then these
systems are classi®ed according to the levels of driver tasks they are supporting, as
strategic (route/mode choice, etc.), tactical (vehicle manoeuvring , etc.) and
operational (steering , accelerator handling, etc.) (Michon 1985). For each level of
driver tasks, driver-assistance systems can be further classi®ed according to the
driver subtasks they are referring to, such as perception (seeing, hearing, feeling,
etc.), decision (for the various actions) and action (execution ) (Sternberg 1969).
Sometimes, driver-assistanc e systems are also classi®ed according to the human ±
machine interface they provide, like the provision of plain information, advices,
warning messages, communication with the environment and to the capability of
proceeding to a speci®c action.
The above classi®cations of driver-assistance systems re¯ect the current
directions of research and development, which up to now concentrated either on
180 J. Golias et al.
the systems’ improvement (technica l advances) or on the driver behaviou r
identi®cation (huma n ± machine interface) . Consequently, very often in the
literature, driver-assistanc e classi®cation follows either a system-oriented approach
or a user-oriented approach, fully responding to the increasing complexity of driver
assistance functions. However, these kinds of classi®cations cannot provide answers
on the usefulness of driver-assistance systems, as the impact to tra c e ciency and
road safety is not taken into consideration. Within this work, an impact-oriented
approach for the classi®cation of driver-assistanc e systems is attempted , where
priorities for future developments can better be identi®ed.
More precisely, a series of criteria for the identi®cation of safety and e ciency
impact are developed (Section 2), followed by the presentation and impact analysis
of the various driver support systems (Section 3) and vehicle support systems
(Section 4). The impact analysis was carried out mainly on the basis of expert
judgement, supported in some cases by evidence in the literature. The outcome of
this analysis allowed the formulation of an impact-oriented classi®cation of driver-
assistance systems in Section 5. The overall conclusions are given in Section 6.
2. Criteria for the identi®cation of safety and e ciency impact
The systems examined concerned either support of the driver or of the vehicle
and are summarized in the following list.
Driver support systems:
.Driver information (navigation routing, integrated navigation, real-time tra c
and traveller information) .
.Driver perception (vision enhancement, electronic mirror, parking and reversing
aid, state of the road surface systems).
.Driver convenience (driver identi®cation, hands-free and remote control,
automated transactions) .
.Driver monitoring (driver vigilance monitoring, driver health monitoring).
Vehicle support systems:
.General vehicle control (automatic stop and go, platooning).
.Longitudinal and lateral control (speed control, adaptive cruise control, road
and lane departure collision avoidance , lane change and merge collision
avoidance).
.Collision avoidanc e (rear-end collision avoidance, obstacle and pedestrian
detection, intersection collision warning).
.Vehicle monitoring (tachograph, alerting systems, vehicle diagnostics).
It is noted that, in practice, many of these driver-assistance systems are combined
with each other, creating complex systems and serving more sophisticated functions.
A good example is given by the platooning system, which integrates both lateral and
longitudinal control of the vehicle. The combination possibilities are very large and it
is not excluded that in the near future most of today’s distinct driver-assistance
systems will be integrated into one overall system.
The estimation of the impact of these driver-assistanc e systems to road safety and
tra c e ciency was based on a number of preset criteria, specially selected for the
purposes of this research. The criteria selected were those proposed in the
Classification of driver-assistance systems 181
international bibliography for measurement s of road safety and tra c-e ciency
impact, adapted for the purpose s of this research . Only criteria mutually
independent were selected allowing for a more straightforwar d impac t analysis.
For example, capacity increase was not considered as a tra c impact criterion, given
that it can automaticall y be translated into improvement of mean speed. The criteria
used are presented at the following list.
Criteria for estimating road safety impact:
.Avoidance of inappropriate speed.
.Keeping appropriat e longitudinal and lateral distances .
.Support of driver awareness.
Criteria for estimating tra c-e ciency impact:
.Speed adjustment.
.Headway adjustment.
The criteria for estimating road safety impact correspond to road accident
factors (Sabey and Taylor 1980), which are addressed by driver-assistanc e
systems. The ®rst criterion corresponds to the common problem of
inappropriate speed for the speci®c tra c and road conditions, whereas
keeping appropriate longitudinal and lateral distances with other vehicles and
road elements refers to the di culties of coordinatio n within tra c conditions,
especially complex ones. Finally, the support of driver awareness aims to deal
with the driver fatigue and the necessary concentration on the driving tasks.
The criteria for estimating tra c-e ciency impact correspond to two basic
tra c parameters, according to tra c engineering theory (Pline 1992). The
adjustment (most often increase) of vehicle speed can result to a higher
network e ciency either for the speci®c vehicle and/or for the complete
network when several vehicles adjust their speed. Additionally , the adjustment
(most often decrease) of vehicle headway can lead to more e cient tra c
conditions as distances between vehicles are optimized. Speed and headway
adjustment can both re¯ect the road capacity increase and the delays decrease.
In the following sections, the presentation of the systems examined adds a new
dimension to the existing descriptive information , namely the analysis of the
potential of each system for the improvement of road safety and tra c e ciency.
The estimation of this impact was based on a combination of expert judgement
and related ®ndings from bibliography research and used the above impact analysis
criteria. Expert judgement was necessary, as available quanti®ed impact analysis
results are very limited. For reasons of analysis organizatio n not necessarily related
to causal consideration s the systems review follows a categorization in driver
(Section 3) and vehicle support systems (Section 4).
3. Driver support systems
3.1. Driver information
The classic systems for driver information are those related to navigatio n
routing, which provide location and route guidance input to the driver (Srinivasan
and Jovanis 1997). Utilizing existing technology such as Radio Data System (RDS)
182 J. Golias et al.
and Tra c Message Channels, these systems combine static information from data
sources such as compact discs or DVDs, with limited information pertaining to the
tra c conditions along the vehicle route. Such systems often have the capability to
perform simple routing operations and provide continuou s driving directions,
based on a number of simple criteria that the driver has preselected (e.g. minimum
distance, minimum estimated travel time, minimum distance on highways, etc).
Navigation routing systems can assist drivers in planning their route or ®nding
their way in areas they are not particularly familiar with, leading thus in tra c
e ciency and safety improvements related not only to the vehicle speed and
longitudinal and lateral distances but also to the reduction in distance driven.
Furthermore, the driver is expected to have an increased level of awareness, as the
result of the system support. The equipped vehicle speed and headway with the
preceding vehicle could also be slightly improved. Nevertheless, as traditiona l
navigation systems do not incorporat e real-time information, but instead are
limited to using historical tra c information, they are not expected to provide
directly signi®cant safety or tra c e ciency bene®ts.
A number of integrated navigatio n systems have emerged in the past few years
that support the driver through a variety of additional services such as signing,
warning or even intervening in the driving process (e.g. by temporaril y taking control
of the vehicle), in the event of unsafe driving conditions, such as unsafe travel speed
for the geometry ahead. Additional integration capabilities relate to the incorpora-
tion of environmenta l and road surface condition s (Hayward et al. 2000). These
systems support the driver on his/her driving pattern by suggesting changes when
unsafe driving behaviour is detected and consequently lead to safety bene®ts by the
avoidance of inappropriat e speed for the speci®c road and tra c conditions. No
signi®cant tra c-e ciency impact is anticipated from these systems as speed and
headway bene®ts are not expected to be considerable.
Real-time tra c and traveller information systems combine the information
available to users of traditional navigatio n systems with real-time travel-relate d
information, which they receive from the infrastructur e (e.g. through dedicated radio
channels, roadside beacons or wide-area transmissions) (Spelt 1997, Wieck 1997). In
general, the information that the user receives can be either descriptive (Kantowitz et
al. 1997) or prescriptive (guidance ) (Bonsall and Parry 1991). Descriptive systems
broadcast current or future tra c conditions, as well as information on incidents or
other relevant information to assist drivers in the determination of their routes.
Some systems provide the user with a set of available routes and additional
information (e.g. travel times for each route) to assist the driver to select the optimal
route. A limited number of systems have the ability to let the user specify criteria for
the determination of the best route (e.g. fastest route, route that avoids interstates,
route that avoids toll roads). Anothe r category of systems generally referred to by
the term prescriptive, determine the best path and broadcast it to the user. Real-time
tra c and traveller information systems can exploit information such as vehicle
location, previous route guidance instructions , safety and advisory information, and
other real-time updates on conditions such as congestion, work zones, environmenta l
and road surface conditions (Brand et al. 1997, Davison et al. 1997). Unlike
traditional navigation systems, real-time information systems consider prevailing
tra c conditions, thus providin g a means of overcoming unexpected tra c
irregularities. Furthermore, such systems can exploit information on planned events
(such as sports games or scheduled road maintenance work). Therefore, the safety
Classification of driver-assistance systems 183
bene®ts from the introduction of such systems in the speed, distance and driver
awareness are not expected to be measurable, but the speed increase of the equipped
vehicles could result in signi®cant tra c e ciency gains. Possible reduction in
distance driven is expected to have a positive impact on both tra c e ciency and
safety. Nevertheless, signi®cant headway improvements are not anticipated, as real-
time tra c and traveller information systems do not assist vehicle control.
It is noted that negative safety consequences from the above driver informatio n
systems are possible, as they require in-vehicle screens, which may distract the
driver’s attention from his primary driving task. Research on driver behaviour
towards the use of these systems, especially during the period of the introduction of
the new systems is considered necessary not only for the identi®cation of the overall
safety impact but also for the appropriat e redesign of the systems, if necessary .
3.2. Driver perception
Vision enhancement systems can augment the driver’s vision under condition s of
reduced visibility, e.g. due to fog, rain, snow or darkness. Such systems require in-
vehicle equipment to sense, process and display the information, ranging from
specially designed headlights to infrared and radar sensors (Mahach et al. 1997).
Data collected from in-vehicle systems can be combined with information obtained
from road sensors and transmitted wirelessly to the vehicle to sense the danger of
imminent collisions. The information is gathered in an on-board computer that
processes it and compares it to preprogrammed safety limits. Generated collision
warning signals can then be displayed on head-up displays (HUD) or transmitted as
audible signals through the vehicle audio system to the driver (Chassant 2000).
Vision enhancement systems may have positive impacts in the maintaining of safe
longitudinal and lateral distances by the driver. As drivers are more aware of the
tra c dynamics of the vehicles around them (both in their direction, as well as the
opposite direction), then they can maintain safer distance s from these vehicles. In
terms of speed and driver awareness, vision enhancement is not expected to result
into identi®able safety bene®ts. Furthermore, as these systems improve the visibility
ahead they are expected to have signi®cant impact on the speed of the equipped
vehicles, as the drivers will be more comfortable driving in higher speeds. On the
contrary, vision enhancement systems are not expected to have signi®cant impacts
on the tra c e ciency through headway reduction.
Rear-view mirrors have been core safety equipment in cars for several decades.
Nevertheless, limitations inherent in the materials used, as well as the structural
elements of the vehicles, result in poor viewing conditions and the existence of blind
spots that often lead to accidents. Video cameras and radar/infrared/thermal sensors
can substitute traditional rear-view mirrors and display an image of the entire rear
view of the vehicle in a special screen in the dashboard . Such systems are collectively
referred to as electronic mirrors (EADM 1999). As watching a screen can distract the
drivers from their primary driving tasks, several systems can process the electronic
mirror images and warn the driver ± using an audio or visual alarm ± in the case of a
dangerous situation , e.g. when the distance between vehicles in the same or adjacent
lanes has dropped below a predetermined safety threshold. Such electronic mirror
systems assist the driver in maintainin g safe distances with vehicles in the same or
adjacent lane and are particularly helpful in dealing with vehicles that are behind the
equipped vehicle, e.g. during lane change manoeuvres. As these systems focus
primarily in the tra c behind the equipped vehicle, they are not expected to have a
184 J. Golias et al.
great positive impact on the vehicle speed or headway. The automatic driver warning
capability in the event of unsafe driving environmen t increases the awareness of the
driver and thus may lead to additional safety gains.
Manoeuvres involving reverse operations present additional di culty to most
drivers. Parking and reversing aid systems provide continuous information and early
warning to drivers performin g such operations . To accomplish this task these
systems use variou s types of sensors to achieve short-range obstacle detection and
tracking. Parking and reversing aid systems assist the driver’s perception of areas
that are out of sight during the execution of delicate manoeuvres and warn the driver
of potentially unsafe close distance to obstacles, oVering a reasonable time-window
for the driver to react accordingly and avoid collision. The tra c or safety impacts of
such systems are not particularly evident, but similar systems may be popular as they
absorb part of the inconvenienc e associated with parking and reverse manoeuvres. It
is obvious, that as these systems do not apply to the majority of the driving tasks,
their impact on safety and tra c e ciency may not be traceable.
A promising category of driver information systems refers to systems collecting
and analysing information on the road surface conditions using vehicle-mounted or
®xed infrastructure road sensors (Cremona et al. 1994). The collected informatio n
can then be transmitted to the in-vehicle system, where it is further processed. If
unsafe road surface condition s are identi®ed ahead of the vehicle, then the driver can
be noti®ed via audio or visual message. Furthermore, the information can be
transmitted to the appropriate tra c information centres, from where it can be
disseminated to other road users as well as the appropriate authorities. The impacts
of such systems could be signi®cant as drivers’ awareness of the road surface
conditions can have considerable positive impacts on both safety (as a combinatio n
of appropriate driving speed adoption, appropriate headway adoption, and
increased driver awareness) as well as tra c e ciency, through better driving
behaviour in relation to speed.
3.3. Driver convenience
Driver convenience is a key factor determining their performance. Systems
already commercially available in high-end luxury cars oVer the capability to identify
the driver (from a choice set of a few precon®gured drivers) and automatically adjust
the seat, the steering wheel, the rear and side mirrors, the temperature, etc. to the
particular driver’s preset preferences. The impact of driver identi®cation systems on
driver awareness can be signi®cant, as the driver is not distracted by trying to adjust
the driving environmen t (seat, steering wheel, rear and side mirrors). Furthermore,
the driving environment always adjusts to the optimal settings for each driver,
whereas manual recon®guration ± often demanding several complex or time-
consuming activities ± could be bypassed, e.g. during short trips, thus compromising
driver’s performance and safety. As these systems do not aVect driving decisions
directly, they are not expected to produce direct safety or tra c e ciency bene®ts
due to speed and/or headway improvements .
Electronic appliances, such as cell phones, fax machines, personal computers,
etc. ®nd their way into passenge r cars. Operation of such systems invariably
distracts the driver, substantially limiting the driver’s capability to respond
properly to emergency situations before an accident. Various systems have been
developed to overcome this issue, including hands-free interfaces and remote
control units located on the steering wheel (Hofmeister et al. 1997). While the
Classification of driver-assistance systems 185
former oVer the highest safety advantages, the latter are often a signi®cant step
towards the right direction for situations where control can not be achieved
through a hands-free interface; at least the driver’s hands do not depart from the
steering wheel. Such systems may have positive impacts on driver’s awareness, as
potential distraction s from the non-driving-relate d information sources are
minimized. A potential safety pitfall, however, is that the drivers of equipped
vehicles may be more inclined to use such systems, on the assumption that they do
not interfere with the driving performance ; even though hands-free systems may
alleviate the distractions of the users from their driving-related tasks, it is clear that
any non-related task impedes the driving performance . As the systems belonging in
this category are not directly involved in the driving process, they are not expected
to signi®cantly aVect road safety and tra c e ciency.
Automated transactio n systems facilitate high-speed electronic transactions ,
often without even the need for the vehicle to slow or stop. Example s of such
systems include electronic toll collection, parking fee payment, and parking/
restricted area entry permit veri®cation (Venable et al. 1995, Abe et al. 1996).
Some of the key technologies that have made such systems possible are smart-
cards (Blythe 1997), GSM and dedicated short-range communication s (DSRC)
(CEN 2000). As these systems have substantial infrastructure requirements and
serve a number of vehicles, standardization is required that will ensure the
interoperability of such systems (Hamet 1999). The positive safety impact of
automated transactio n systems is related to the elimination of unnecessary and
sometimes dangerou s deceleration and acceleration areas. As this is an indirect
eVect, its magnitude is not expected to be considerable . Automated transaction
systems improve tra c e ciency by making transactions such as tolling
transparent to the tra c; instead of having to stop to pay the toll fare, vehicles
can drive through the control station, often at speeds up to 180 km/h.
Furthermore, as the vehicles are not obliged to decelerate and accelerate (as
they would have to do in the case of a non-automated transaction), headway-
related tra c e ciency gains are expected as well.
3.4. Driver monitoring
Driver vigilance monitoring systems continuously monitor the driver’s perfor-
mance for possible signs of condition s that may endanger the driver, such as
drowsiness as a consequence of fatigue (Sugasawa et al. 1996), alcohol abuse,
medication, etc., or lack of attention , e.g. due to stress (Grace 1999, Boverie et al.
2000). When the system detects potential problems with the driver’s awareness then
the system uses audio and visual warnings to draw the driver’ s attention and restore
an acceptable level of alertness. Such systems are expecte d to have high impact in
maintaining driver’s awareness especially during long interurban trips and can
consequently lead to considerable safety bene®ts.
Driver health monitoring systems take driver monitoring a step further by
monitoring several parameters of the driver’s health conditions and combining the
results to estimate the current health level of the driver. From this information, the
system assesses the condition of the driver and if it appears to be below certain
preselected `safe’ levels then the driver and possibly some external entity, e.g. a
doctor or the police, are noti®ed (Hernandez et al. 1998). Obviously, such systems
could have signi®cant awareness-related safety impact as the driver is noti®ed when
his health is deteriorating.
186 J. Golias et al.
4. Vehicle support systems
4.1. General vehicle control
Automatic stop-and-g o systems allow vehicles to automatically stop when this is
necessary, e.g. the preceding vehicle has stopped, and start again when the conditions
allow it (Carrea 1993). Such systems could oVer signi®cant safety bene®ts in
hazardous situations and in situations where frequent stop-and-g o is required, such
as congested conditions. A key issue with automatic stop-and-go systems that are
being developed has been a relatively high false alarm rate; overcoming this problem
could lead to a wider use of such system s (Brand et al. 1997). Automatic stop-and-go
systems could have signi®cant positive safety impacts by managing the vehicle speed
and longitudinal distance better than the human driver. Nevertheless, these systems
generally would not improve the driver’ s awareness, especially as the driver could
increasingly rely more on the system’s performance. Finally, as stop-and-go usually
applies to congested conditions limited signi®cant tra c e ciency gains are
expected, mainly through improved headway maintenance.
Another system in this category, albeit with a lower level of maturity is
platooning, a situation where each vehicle travels keeping a constant headway from
the preceding one, either through external speed control or through electronic speed
control by the vehicle itself. A special case of this function is the tow-bar application,
where the vehicles (usually trucks) are electronically coupled and each follows the
preceding one. Platooning application areas are usually restricted to highway and
motorway network sections with a reduced speed limit (usually up to 85 km/h). The
more advanced platoonin g systems, where the vehicles are externally controlled by
the infrastructure, are called intelligent vehicle highway systems (IVHS) (Brand et al.
1997). Platooning systems could oVer safety bene®ts as the vehicle’s speed and
position are electronically controlled. Platooning systems widespread use could
result in signi®cant tra c e ciency bene®ts as vehicles ± that have formed platoons ±
achieve higher speeds and shorter headways.
The most well-known longitudinal control system is probably the variable speed
limiter or the intelligent speed adaptation. DiVerent implementations range from
external speed recommendations to an automatic speed reduction function. The
latter may be imposed directly to all (equipped) vehicles within the control area, e.g.
through a communication centre, or indirectly, e.g. by managing the local tra c
lights accordingly. Stop and go functions may be also included in this category,
especially when implemented in conjunctio n with an infrastructure-base d system
(Davison et al. 1997, Oei 1998b). Speed-limiting systems may provide signi®cant
safety improvements as they oVer a way to eVectively control speed and longitudina l
distance between cars. The safety impacts of such systems through improvement of
driver’s awareness are not expected to be high. Obviously, such speed-limitin g
systems could not have considerable positive impacts on tra c e ciency.
Adaptive cruise control (ACC) is a more elaborat e longitudinal control system
which adjusts vehicle speed to maintain a safe separation with the preceding vehicle
(Martin 1993, Winner et al. 1996). Adaptive cruise control senses the presence and
relative velocity of moving vehicles ahead of the equipped vehicle and adjusts the
speed of travel accordingly (Oei 1998a, Hayward et al. 2000). Adaptive cruise control
systems could have signi®cant positive safety impacts as they eVectively control the
speed and longitudinal distance between vehicles, ensuring that no rear-end
collisions occur. However, driver’s awareness is not in¯uenced in a way that could
Classification of driver-assistance systems 187
lead to signi®cant safety bene®ts. Furthermore, as the speed and headway are
automatically control they are optimized and can therefore provide better tra c
performance compared with human control.
It is noted that negative safety consequences from the introduction of these
general vehicle control systems may be encountered, as the driver could be surprised
in his driving task by the automatic actions undertaken by the systems. Furthermore,
depending on the settings of the various vehicle control systems (maximu m speed
and headway) negativ e impacts on tra c may appear. Only further driver behaviour
research and appropriate redesign of the systems will allow the minimization of any
negative impact due to these systems.
4.2. Collision avoidance
Road and lane departur e collision avoidance is a lateral control system
providing warning and control assistance to the driver through lane or road edge
tracking and by determining the safe speed for road geometry in front of the
vehicle (Pomerleau et al. 1997). The system continuously calculates the vehicle’s
optimal position and compares it with the actual vehicle position. If the system
detects deviations exceeding the de®ned safety thresholds, then it creates
audiovisual warnings for the driver. More advanced systems could feature
extended functionality , including particular suggestions /directions to the driver
on how the particular problem can be overcome or actual control interventions to
restore the vehicle’s intended path. Such systems may have a positive impact on
safety by ensuring that vehicles do not inadvertently depart from their desired lane,
a situation that can often result in an accident. The safety impacts of these systems
on speed and driver awareness are not expected to be signi®cant. Similarly, as these
systems are only activated in the event of an emergency situation (road or lane
departure) the direct impact on the tra c e ciency is not expected to be traceable.
Lane change and merge collision avoidance systems are another type of lateral
control systems providing various levels of support for detecting and warning the
driver of vehicles and objects in adjacent lanes. While this information can be
useful during normal driving conditions, it is particular valuable during lane
change or merge manoeuvres. Systems in this category track vehicles in adjacent
lanes and use this information to warn the driver when their position and/or speed
makes the planned lane change/merge manoeuvre unsafe. More sophisticated
systems may include speed and steering control interventio n for enhanced collision
avoidance. Such systems ensure that lateral separatio n between vehicles in adjacent
lanes are always maintained and may therefore have signi®cant positive impacts in
the reduction of tra c accidents (Mazzae et al. 1995, Young 1995, Campbell et al.
1996). The optimized lane change and merging capabilities of these systems may
lead in signi®cant tra c e ciency gains, related to better headways . These systems
are expected to have little or no impact on the vehicle’s speed.
Rear-end collision avoidance systems sense the presence and speed of
vehicles and objects in the vehicle’s lane of travel and provide to minimize the
risk of collisions with vehicles and objects found in front of the equipped
vehicle (Ganci et al. 1995, Woll 1995). More sophisticate d versions can include
longitudinal control through vehicle braking and speed adaptation, and
ultimately even lateral control by oVering lane change capabilities to avoid
collisions. Rear-end collision avoidance systems could oVer safety improvements
by monitoring the lane in which the vehicle is travelling for slow moving
188 J. Golias et al.
vehicles and other obstacles and adjusting the equipped vehicle’s speed and
travelling lane accordingl y to avoid a collision. It is obvious that systems that
adjust the vehicle’s speed ± such as rear-end collision avoidance ± can not have
signi®cant safety or tra c e ciency bene®ts associated with optimized speed. In
addition, the impact of such systems in driver awareness is not expecte d to
produce enough potential for safety bene®ts.
Obstacle and pedestrian detection systems oVer similar services by warning the
driver when pedestrians, vehicles, or obstacles are in close proximity to the driver’s
intended path (Butsuen et al. 1996, Kamiya et al. 1996, Sugasawa et al. 1996,
Papageorgiou et al. 1998). Informatio n from on-board sensors or infrastructure-
based sensors is used to detect obstacles and pedestrians and speed and direction
information from the on-board computer is used to estimate the vehicle’s path. The
combination of the above informatio n may generate alarms notifying the driver of
potentially unsafe conditions. Even though the operation of such systems is similar
to that of the rear-end collision avoidance systems, obstacle and pedestrian detection
systems oVer diVerent safety bene®ts. In particular, as obstacle and pedestrian
detection systems provide warnings to the driver when potential collision danger is
imminent, positive safety impacts related to vehicle speed and driver awareness may
be achieved. Furthermore, as these systems may only limit the vehicle speed (as a
result of obstacle and pedestria n warnings), no signi®cant tra c e ciency bene®ts
are expected.
Most accidents happen at intersections. Intersection collision warning systems
utilize a cooperation of vehicle and infrastructure systems to provide warning to the
driver for potential collision at an intersection (Lloyd et al. 1996, Brand et al. 1997).
A special category of intersection collision avoidance systems is railroad crossing
collision avoidance systems, which provide in-vehicle warnings to drivers approach-
ing railroad crossings when a train is approaching (Luedeke 1997, Polk 1997). Initial
implementation of this feature is anticipate d for buses and trucks carrying hazardous
cargo. Extensions to other vehicles may become feasible when this technology
becomes more cost-eVective. One way to achieve this is by creating economies of
scope, i.e. combine the system with other services. Intersection collision warning
systems may provide considerable safety bene®ts by limiting on time the speed of the
equipped vehicle, when collision danger is sensed at a downstream intersection.
Furthermore, such systems may improve the driver’s awareness by notifying, e.g. by
audio or visual messages, the driver about potentially dangerous conditions that the
system has identi®ed. Finally, these systems are not expected to aVect headway and
speed adjustment in a way that could in¯uence tra c e ciency.
4.3. Vehicle status monitoring
A simple system for vehicle status monitoring is the tachograph recording
equipment which can record, store, display, print, and output data related to driver
activity, as well as log informatio n describing the beginning and end of each trip,
control activities performed during the trip, e.g. by the police, etc. Tachograph
systems can also log additiona l information allowing for more sophisticated analysis
when the tachograph information is analysed, e.g. the ability to detect operating
violations, such as driving for a period exceeding the maximum designated period
without a stop or exceeding the speed limit for the speci®c type of vehicle. The
existence of a tachograp h in the vehicle may force the driver to be more alert, drive at
safer speeds and maintain optimal distances from other vehicles, thus improving
Classification of driver-assistance systems 189
safety. It is, however, obvious that the contribution of such auditing equipment in
the improvement of tra c e ciency is not expected to be considerable .
A large number of alerting systems have been developed aimed at alerting the
emergency services (e.g. police, ambulance, ®re brigades, highway patrols) in case of
a tra c incident. Furthermore , some of these systems oVer also dedicated support
services, to which the troubled drivers get connecte d automatically. Most of the
systems feature either a cell-phone technology connection or satellite-based
communications (Benson and Clima 1996, Heddebau t and Rioult 1998). Alerting
systems can either be absolutely automated and/or require a ± more or less ±
substantial intervention from the driver of the vehicle in emergency (Sobolewski and
Deeter 1997). For example, automatic collision noti®cation system calls auto-
matically for emergency services when the vehicle is involved in a serious crash and it
also supports manual operations to call for emergency or other types or roadside
assistance. The vehicle location is automatically transmitted with the call for
assistance in either the automatic or manual mode of operation. Position accurac y
will be su cient to determine direction of travel on a divided highway. Beyond the
noti®cation/alert services, these systems take advantag e of the existing infrastructure
to bundle a number of other services, including stolen vehicle tracking, remote door
unlock, roadside assistance in case of car breakdown, route support upon request,
remote diagnostics of vehicle malfunctions and driver condition. Alerting systems do
not intervene with the driving tasks and therefore are not expected to yield
signi®cant safety or tra c-related bene®ts.
Vehicle diagnostic information systems are an extension of current vehicle
monitoring and self-diagnostic capabilities, oVering elaborate engine condition
information services such as oil pressure and coolant temperature gauges (Bannatyne
and Warshawsky 1997). Such systems monitor vehicle safety-related functions,
compare the readings with the expected operating conditions and warn the driver or
the operator if potentially irregular operation is detected. Diagnostic informatio n
services can be considered as an added diagnostic capability for existing functions
like monitoring braking system integrity, sensor and actuator performance, and the
communication system. Like the alerting systems, vehicle diagnostic systems do not
intervene with the driving tasks and therefore no signi®cant safety or tra c-related
bene®ts are expected.
5. Impact-oriented classi ®cation of driver-assistanc e systems
On the basis of the above presentation and impact analysis of driver-
assistance systems , a high or low impact value was assigned to each of the
preselected criteria (see Section 2) for each driver-assistance system examined. In
this way, table 1 was prepared, which clari®es the degree of safety and e ciency
impact of the systems considered.
It is noted that the estimation of the impact of driver-assistanc e systems refers
only to direct impac t on road safety and tra c e ciency. Certainly, there is often
indirect impact, as for example the avoidance of an accident (i.e. the decrease in the
number of road accidents) may lead to the avoidance of a related congested situation
(i.e. less congestion) . However, for the purposes of this research, only direct impact
was considered in the analysis of each system.
It is noted that in some cases, driver-assistance systems with signi®cant quantities
of information on in-vehicle screens (e.g. navigation routing, electronic mirrors, etc.)
may distract the driver’s attention and thus they may have negative consequences on
190 J. Golias et al.
Table 1. Assessment of road safety and tra c e ciency impact of various driver-assistance systems.
Road safety Tra c e ciency
Keeping
appropriate
Avoidance of longit. and Support
inappropriate lateral driver Speed Headway
speed distance awareness adjustment adjustment
DRIVER driver navigation routing L L L L L
information integrated navigation H L L L L
real time tra c and traveller information L L L H L
driver vision enhancement L H L H L
perception electronic mirror L H L L L
parking and reversing aid L L L L L
state of the road surface systems H H H H L
driver driver identi®cation L L H L L
convenience hands-free and remote control L L H L L
automated transactions L L L H H
driver driver vigilance monitoring L L H L L
monitoring driver health monitoring L L H L L
VEHICLE general automatic stop and go H H L L L
vehicle control platooning L L L H H
speed control H H L L L
adaptive cruise control H H L H H
collision road and lane departure collision avoidance L H L L L
avoidance lane change and merge collision avoidance L H L L H
rear end collision avoidance L H L L L
obstacle and pedestrian detection H L H L L
intersection collision warning H L H L L
vehicle tachograph L L L L L
monitoring alerting systems L L L L L
vehicle diagnostics L L L L L
H: high, important impact; L: low, limited or insigni®cant impact.
Classification of driver-assistance systems 191
the safety level. In the framework of this research, it is assumed that any negative
impacts are limited. However, further research is require d towards determining the
actual driver behaviour due to in-vehicle screens, which will clarify the overall net
balance of these systems in relation to safety.
The above impact analysis of driver-assistance systems led to the formulation
of an impact-oriente d approach for the classi®cation of the driver-assistanc e
systems, as shown in table 2. Here, a system is put in a category of high impact
only if at least one of the criteria has a value of high impact. Within each part of
table 2, the systems are put in order of importance, with those having the more
values of high impact at the top.
It is interesting to note from table 2 that only four driver-assistance systems
present high values for both road safety and tra c-e ciency impact (state of the
road surface systems, adaptive cruise control, lane change and merge collision
avoidance and vision enhancement), with the ®rst two systems presenting high-
impact value for four out of ®ve criteria. Twelve other systems present a high road
safety impact but limited tra c-e ciency impact, whereas only three systems present
important high tra c-e ciency impact and limited safety impact. Finally, another
®ve systems present limited impact to both road safety and tra c e ciency. The fact
that there are more systems with importan t impact on road safety (16 systems) than
systems with important impact on tra c e ciency (seve n systems) was expected, as
the primary objective for most of the driver-assistance systems is the improvemen t of
road safety.
6. Conclusions
Today, the current information technology application s together with several
developments in the car-industry allow for fast and e cient communication
Table 2. Impact-oriented classi®cation of driver-assistance systems.
Road safety impact
High Low
Tra c High state of the road surface systems automated transactions
e ciency adaptive cruise control platooning
impact lane change and merge real time tra c and traveller
collision avoidance information
vision enhancement
Low automatic stop and go navigation rou ting
speed control parking and reversing aid
obstacle and pedestrian detection tachograph
intersection collision warning alerting systems
integrated navigation vehicle diagnostics
electronic mirror
driver identi®cation
hands-free and remote control
driver vigilance monitoring
driver health monitoring
road and lane departure
collision avoidance
rear end collision avoidance
192 J. Golias et al.
networks, real-time processing of large amount of data and user friendly human ±
machine interfaces (Rumar et al. 1999), which make possible the reliable and cost-
eVective use of several driver-assistance systems. In this way, driver-assistanc e
systems are incorporate d in the new vehicles, responding better to the diverse needs
of the road users for safer and more e cient tra c conditions. Subsequently, the
existence of strong evidence that the impact of driver-assistance systems to road
safety and tra c e ciency is positive, will progressively make the use of these
systems much more attractive, as their ®xed and operational costs are steadily
decreasing.
The comprehensiv e and critical review of several existing driver-assistanc e
systems carried out in the framework of this work, allowed for the macroscopic
identi®cation of the impact of these systems to road safety and tra c e ciency.
The output of this critical review led to the formulation of an alternative
classi®cation approach of these systems based on the systems’ impact to road
safety and tra c e ciency, escaping from the traditional classi®cations, which
follow system or user oriented approaches . As driver-assistanc e systems reach
their maturity, the proposed system performance based classi®cation is more
suitable for their further development , coupling the existing work on systems
availability with work on systems usefulness. Besides, the appropriate develop-
ment process of any type of system should be based on the suitable balance
between system availability, attempting to provide answers to basic questions
concerning issues like whether the systems required are produced and whether the
systems produced are required.
The proposed classi®cation proposes a ranking of driver-assistance systems,
where at the top are found the systems presenting high values in the preselected road
safety and tra c-e ciency impact criteria. The systems providing real-time
information for the road surface systems as well as those related to adaptive cruise
control are ranked at the top impact levels followed by systems related to lane
change and merge collision avoidance as well as vision enhancement . From the
proposed classi®cation it appears that more systems present high safety impact than
high-e ciency impact, a result possibly attributed to the fact that historically driver-
assistance systems were primaril y aiming at road safety improvement.
This impact-oriente d classi®cation allows for another view of the particularities
of driver-assistance systems, leading thus to the identi®cation not only of the areas
where future research is needed, but also of the priorities for further system
development and promotion. It is obvious that developmen t and application of
driver-assistance systems should be carefully monitored and not left only to industry
and market forces, especially when road safety issues are addressed. Additionally,
this classi®cation should allow for the identi ®cation of the necessary legislation
actions for mandatory standards or procedures at internationa l level to ensure the
proper design of new intelligent transportation systems.
The results of this researc h make obvious the need for furthe r experimental
and laboratory research in the ®eld of quantifying road safety and tra c-
e ciency impact of driver-assistanc e systems. When more quanti®ed and model-
based impact analysis (at both micro and macro scale) is available, the proposed
systems classi®cation can be more detailed and consequently more useful for the
better identi®cation of the priorities for the development of driver-assistance
systems. Certainly, impact analysis should not be limited to microscale modelling
but also to network and transportatio n system modelling as well as to
Classification of driver-assistance systems 193
multicriteria analysis allowing for the identi®cation of the overall system impact
to safety, e ciency, environmen t and the economy.
Finally, this critical review of driver-assistanc e systems revealed that the
systems performanc e and impact is related to a number of developments
concerning to the concept of intelligent road, the human ± machine interface and
the systems implementation transitio n phases. Consequently, related further
research in these ®elds should be coupled with the corresponding research on the
driver-assistance systems so that an e cient and operative outcome is achieved.
References
ABE, A., HO RI, T. and OFUCHI, K., 1996, Observations of the eVect on drivers of lane guidance
function employing in-vehicle voice guidance and display units [CD-ROM]. In
Proceedings of the Third Annual World Congress on Intelligent Transport Systems,
Orlando, FL.
BANNATYNE, R. and WA RSHAWSKY, E., 1997, The changing world of automotive electronic
control units. In Proceedings of the Future Transportation Technology Conference
(Warrendale: Society of Automotive Engineers), paper 972638.
BENSON, J. L. and CLI MA, B. T., 1996, Implementing a MAYDAY system: institutional issues
and protocol involved in implementing vehicle MAYDAY technologies. In Third World
Congress on Intelligent Transport Systems, Orlando, FL.
BLYTHE, P. T., 1997, Implementing smart card technology to transform transport payment and
customer information. In Proceedings of the 4th World Congress on Intelligent Transport
Systems (Berlin: ICC).
BONSALL, P. and PARRY, T., 1991, Using an interactive route-choice simulator to investigate
drivers’ compliance with route guidance advice. Transportation Research Record,1306,
59 ± 68.
BOVERIE, S., DEVY, M., LEQUELLE C, JM., MENGEL, P. and ZITTL AU, D., 2000, 3D perception for
vehicle inner space monitoring. In AMAA Conference, Berlin, Germany, March.
BRA ND, C., DAVISON, P., LEWIS, A., MOON, D., SITE, P. D., GENTILE , C., FILIPP I, F., LANDOLFI, O.,
DOUGHERTY, M., KO RVER, W., HAR REL L, L., VAN TOORENBURG, J., AKER MAN, J., ERIK SSON,
E. A., HE RNA NDEZ, H., WEBER, M., DAUNERV, A., HECKELSMUELLER , J., HOFF MANN, A.,
LEISS, U., LINDE, E. and PET ZEL, W., 1997, Forecast of New Technologies with Major
Impacts. EU project FANTASIE, deliverable 9 November.
BUTSUE N, T., YOSHIOKA, T. and OKUDA, K., 1996, Introduction of the Mazda advanced safety
vehicle. In Proceedings of the 1996 IEEE Intelligent Vehicles Symposium, Tokyo, pp.
242 ± 247.
CAMPBELL, J. L., HOOEY, B. L., CAR NEY, C., HANOWSKI, R. J., GORE, B. F., KANTOWITZ, B. H. and
MITCHELL, E., 1996, Investigation of Alternative Displays for Side Collision Avoidance
Systems (Final Report) (Seattle: Battelle Research Center).
CARR EA, P., 1993, Integration between anticollision and AICC functions: the Alert Project. In
Proceedings of the Intelligent Vehicle s ’93 Symposium (Tokyo: IEEE).
CEN TC 278 WG 9, 2000, Dedicated Short-Range Communications in close cooperation with
ISO TC 204 WG 15 `DSRC for TICS Applications’, Road Transport and Tra c
Telematics ± Dedicated Short -Range Commun ications.
CHASSANT, Y., 2000, La vision de nuit. Auto Concept Magazine,March/April.
CREMO NA, P., KUNER T, M. and CA STINI E, F., 1994, Parametric spectrum analysis for target
characterization. In Proceedings of the Intelligent Vehicles ’94 Symposium (P aris: IEEE).
DAVI SON, P., BR AND, C., LEWIS, A., MOON, D., SITE , P. D., GEN TILE, C., FIL IPPI, F., LANDOLFI, O.,
DOUGHERTY, M., KORVER, W., HARRELL, L., VAN TOORENBURG, J., AKERMA N, J., DAUNER,
A., HECKELSMUELLER , J., LEISS, U., LINDE, E. and PE TZEL, E., 1997, A Structured State-of-
the-Art Survey and Review. EU project FANTASIE, deliverable 8 November 1997.
EADM, 1999, Electronic mirror merges views. European Automotive Design Magazine,June.
EUROPEAN COMMISSION, 1997, Comm unity Strategy and Framework for the D eployment of Road
Transport Telematics in Europe. COM (7) 223 ®nal (Brussels: EC).
GANCI, P., POTTS, S. and OKUR OWSK I, F., 1995, Forward-looking automotive radar sensor. In
Proceedings/SPIE ±The International Society for Optical Engineering, 2592, pp. 60 ± 65.
194 J. Golias et al.
GRA CE, R., 1999, A drowsy driver detection system for heavy vehicle. Paper presented to the
Conference on Ocular Measures of Driver alertness, Herndon, VA.
HAME T, P., 1999, Europe ± The Genera l D irection. TOLLtrans Supplement to Tra c
Technology International, October/November, 10 ± 11.
HAY WAR D, M., BECK ER, S., NILSSON, L., BROCKMANN, M. and SALA, G, 2000, TR1004, AC ±
ASSIST, Anti-Collision Autonomous Support and Safety Intervention System
(incorporating ROADSTER), submitted as project deliverable D3. 1: Report on Users’
Needs.
HEDDEBAUT, M. and RIOULT, J, 1998, Appel d’urgence, localisation d’us agers en detresse et
alerte rapide des conducteurs. Analyse technique des solutions en cours de de
Âploiement.
TEC journal, 150, 11 ± 19.
HEIJER, T., OEI, H. L., WIETHOFF, M., BOVERIE, S., PENT TINEN, M., SCHIROKOFF, A., KULMAL A,
R., HEINR ICH, J., ERN ST, A. C., SNEEK, N., HEER EN, H., STEVE NS, A., BEKIARIS, A. and
DAMIANI, S., 2000, Problem identi®cation, user needs and inventory of ADAS,
ADVISORS project deliverable D1/2.1.
HERNANDEZ-GRESS, N., ESTE
ÁVE, D. and BEKIARIS, A., 1998, IMU: integrated monitoring unit of
the SAVE diagnostic system. In Intelligent Component for Vehicles Workshop (ICV ’98),
Seville, Spain.
HOFM EISTE R, A., MATTO S, P., NICOLAI, J. and KASPARIAN, M., 1997, The Driver Information
System for the Rest of Us. Paper 970171 (Warrendale: Society of Autom otive
Engineers).
KAMI YA, H., FUJITA, Y., TSURGA, T., NAKUMR A, Y., MA TSUDA, S. and EN OMO TO, K., 1996,
Intelligent technologies of Honda ASV. In Proceedings of the 1996 IEEE Intelligent
Vehicles Symposium, Tokyo, pp. 236 ± 241.
KANT OWIT Z, B. H., HANOWSKI, R. J. and KANTOWITZ, S. C., 1997, Driver acceptance of
unreliable tra c information in familiar and unfamiliar settings. Human Factors,39,
164 ± 176.
LLOY D, M. M., BITTNER, JR, A. C. and PIERO WIC Z, J. A., 1996, Driver-vehicle interface (DVI)
design issues of an intersection collision avoidance (ICA) system [CD-ROM]. In
Proceedings of the Third Annual World Congress on Intelligent Transport Systems,
Orlando, FL.
LUEDEKE , J., 1997, Highway Railroad Grade-Crossing Warning Device and Barrier System
Technologies for High-Speed Applications, vol. III (Draft Report) (Columb us: Battelle
Memorial Institute).
MAHACH, K., KNOBLAUC H, R. and SIMMONS, C., 1997, A Preliminary Field Evaluation Test of
Ultraviolet-Activated Fluorescent Roadway Delineation (FHWA-RD-97-082) (Washing-
ton, DC: FHWA).
MARTIN, P., 1993, Autonomou s Intelligent Cruise Contro l Incorporating Automatic Braking.
Technical Paper 930510 (Warre ndale: Society of Automotive Engineers).
MAZZAE, E. N., GARROTT, W. R. and FLIC K, M. A., 1995, Human Factors Evaluation of Existing
Side Collision Avoidance System Driver Interfaces. Tech nical Paper 952659 (Warrendale:
Society of Automotive Engineers).
MICHON, J., 1985, A critical view of driver behavior models: what do we know, what should we
do? In L. Evans and R. C. Schwing (eds), Human Behavior and Tra c Safety (New
York: Plenum), pp. 485 ± 524.
NANIOPOULOS, A., 2000, Advanced driver-assistance systems and tra c safety. In Proceedings
of Workshop on `The role of Advanced Driver Assistance Systems on tra c safety and
e ciency’, organized by NTUA and AUTh, Athens, 18 October.
OEI HWAY-LIEM, 1998a, Advanced Cruise Control (ACC): A Literature Study (Leidschendam:
SWOV). [in Dutch]
OEI HWAY-LIEM, 1998b, Intelligent Speed Adaptation (ISA). A Literature Study (Leidschendam:
SWOV). [in Dutch]
PAPAGEORGIOU , C., EVGENIOU, TH. and POGGIO, T, 1998, A trainable pedestrian detection system.
MIT, CBCL and AI Laboratory Cambridge, USA. In 1998 IEEE International
Conference on Intelligent Vehicles.
PLINE J., 1992, Tra c Engineering Handbook, 4th edn (Englewood CliVs: Prentice Hall for the
Institute of Transportation Engineers).
Classification of driver-assistance systems 195
POLK, A. E., 1997, ITS applications to railroad crossing safety: a summary of U.S. activities
[CD-ROM]. In Proceedings of the ITS America Seventh Annual Meeting (Washington,
DC: ITS America).
POMERLEAU, D., THOR PE, C. and EM ERY, L., 1997, Performance speci®cation development for
roadway departure collision avoidance systems. In Proceedings of the 4th World
Congress on Intelligent Transport Systems (Berlin: ICC).
RUMAR , K, CARSTEN, O., FLEYRY, D., HEI JER, T., KILDEBOGAARD , J., KULMALA, R., LIND, G.,
MACHA TA, K., MA URO, V., ZACKOR, H., BERRY, J., BREEN, J. and WAR D, M., 1999,
Intelligent Transportation Systems and Road Safety (Brussels: European Transport
Safety Council).
SABEY, B. E. and TAYLOR, H., 1980, The Known Risks We Run: The Highway. Report SR 567
(Crowthorne: TRRL).
SALA, G. and MUSSONE, L., 2000, The potential impact on tra c safety of lateral support
systems. In Proceedings of the 7th World Congress on ITS, Turin, Italy.
SOBOLEWSKI, M. and DEETER , D., 1997, Multi-Jurisdictional Mayday (MJM) ± adding value to
existing operational tests [CD-ROM]. In Proceedings of the ITS America Seventh Annual
Meeting (Washington, DC: ITS America).
SPELT, P., 1997, Development and evaluation of an in-vehicle information system [CD-ROM].
In Proceedings of the ITS America Seventh Annual Meeting (Washington, DC: ITS
America).
SRINIVASAN, R. and JOVANIS, P., 1997, EVect of selected in-vehicle route guidance systems on
driver reaction times. Human Factors, 39, 200 ± 215.
STERNBERG, S., 1969, The discovery of processing stages. In W. C. Koster (ed.), Attention and
Performance, vol. 2 (Amsterdam: North Holland).
SUGASAWA, F., UENO , H., KANE DA, M., KOREISHI, J., SHIRATO, R. and FUKUHARA, N., 1996,
Development of Nissan’s ASV. In Proceedings of the 1996 IEEE Intelligent Vehicles
Symposium, Tokyo, pp. 254 ± 259.
VENAB LE, D. L., MACHEMEHL, R. B. and EURITT, M. A., 1995, Electronic Toll Collection Systems
(Austin: Center for Transportation Research, Bureau of Engineering Research,
University of Texas).
WIECK, M., 1997, FAST-TRAC± a success for dynamic rou te guidance [CD-ROM]. In
Proceedings of the ITS America Seventh Annual Meeting (Washington, DC: ITS
America).
WINNER, H., WITTE, S., UHLER, W. and LICHTENBERG, B., 1996, Adaptive Cruise Control System
Aspects and Development Trends. Technical Paper 961010 (Warrendale, PA: Society of
Automotive Engineers).
WOLL, J., 1995, Monopulse Doppler radar for vehicle applications. In Proceedings of the
Intelligent Vehicles ’95 S ymposium (Detroit: IEEE).
YOUNG , S., 1995, Collision avoidance system performance speci®cation for lane-change,
merge, and backing: phase I results and future plans. In Proceedings of the Peer Review
Workshop of the National Highway Tra c Safety Administration Program (Washington,
DC: US DOT, NHTSA).
196 J. Golias et al.
