Policy and society related implications of automated driving: A review of literature and directions for future research

Abstract

In this paper, the potential effects of automated driving that are relevant to policy and society are explored, findings discussed in literature about those effects are reviewed and areas for future research are identified. The structure of our review is based on the ripple effect concept, which represents the implications of automated vehicles at three different stages: first-order (traffic, travel cost, and travel choices), second-order (vehicle ownership and sharing, location choices and land use and transport infrastructure) and third-order (energy consumption, air pollution, safety, social equity, economy and public health). Our review shows that first-order impacts on road capacity, fuel efficiency, emissions and accidents risk are expected to be beneficial. The magnitude of these benefits will likely increase with the level of automation and cooperation and with the penetration rate of these systems. The synergistic effects between vehicle automation, electrification and sharification can multiply these benefits. However, studies confirm that automated vehicles can induce additional travel demand because of more and longer vehicle trips. Potential land use changes have not been included in these estimations about excessive travel demand. Other third-order benefits on safety, economy, public health and social equity still remain unclear. Therefore, the balance between the short-term benefits and long-term impacts of vehicle automation remains an open question

Full text

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS
,VOL.,NO.,–
https://doi.org/./..
Policy and society related implications of automated driving: A review of literature
and directions for future research
Dimitris Milakis a, Bart van Arema,andBertvanWee
b
aDepartment of Transport and Planning, Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands;
bTransport and Logistics Group, Faculty of Technology, Policy and Management, Delft University of Technology, Delft, The Netherlands
ARTICLE HISTORY
Received  October 
Revised  September 
Accepted  January 
KEYWORDS
automated driving; first,
second, and third order
impacts; policy and societal
implications; ripple effect
ABSTRACT
In this paper, the potential eects of automated driving that are relevant to policy and society are
explored, ndings discussed in literature about those eects are reviewed and areasfor future research
are identied. The structure of our review is based on the ripple eect concept, which represents the
implications of automated vehicles at three dierent stages: rst-order (trac, travel cost, and travel
choices), second-order (vehicle ownership and sharing, location choices and land use, and transport
infrastructure), and third-order (energy consumption, air pollution, safety, social equity, economy, and
public health). Our review shows that rst-order impacts on road capacity, fuel eciency, emissions,
and accidents risk are expected to be benecial. The magnitude of these benets will likely increase
with the level of automation and cooperation and with the penetration rate of these systems. The
synergistic eects between vehicle automation, sharing, and electrication can multiply these bene-
ts. However, studies conrm that automated vehicles can induce additional travel demand because
of more and longer vehicle trips. Potential land use changes have not been included in these esti-
mations about excessive travel demand. Other third-order benets on safety, economy, public health
and social equity still remain unclear. Therefore, the balance between the short-term benets and
long-term impacts of vehicle automation remains an open question.
Introduction
Automated driving is considered to be one of those tech-
nologies that could signal an evolution toward a major
change in (car) mobility. Estimations about the extent of
this change can be inferred by answering the following
two questions: (a) what are the potential changes in mobil-
ityandtheimplicationsforsocietyassociatedwiththe
introduction of automated driving and, (b) to what extent
are these changes synchronized with broader concurrent
societal transformations that could enhance the radical
dynamicofsuchmobilitytechnology?Examplesofsocial
transformations could be the digital and sharing econ-
omy, the livability and environmental awareness move-
ment and the connectivity, networking, and personalized
consumption trends.
In this paper, the focus is on the rst question, aiming
to (a) explore the potential eects of automated driving
relevant to policy and society, (b) review ndings dis-
cussed in literature about these eects, and (c) identify
areas for future research. Thus far, scholarly eorts have
been mainly concentrated on the technological aspects of
vehicle automation (i.e. road environment perception and
CONTACT Dimitris Milakis Department of Transport and Planning, Faculty of Civil Engineering and Geosciences, Delft University of Technology,PO B ox ,
 GA Delft, The Netherlands.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gits.
motion planning) and on the implications for driver and
trac ow characteristics. Accordingly, review eorts
have focused on the development and operation of vehicle
automation systems and the associated technologies (see
Gerónimo, López, Sappa, & Graf, 2010; González, Pérez,
Milanés, & Nashashibi, 2016; Piao & McDonald, 2008;
Shladover, 2005; Shladover, 1995;Sun,Bebis,&Miller,
2006;Turner&Austin,2000;Vahidi&Eskandarian,
2003;Xiao&Gao,2010). Several review studies have also
focused on the rst-order impacts of vehicle automa-
tion with a special emphasis on trac ow eciency
(see Diakaki, Papageorgiou, Papamichail, & Nikolos,
2015;Hoogendoorn,vanArem,&Hoogendoorn,2014;
Hounsell, Shrestha, Piao, & McDonald, 2009;Scarinci
& Heydecker, 2014) and human factor aspects such as
behavioral adaptation, driver’s workload, and situation
awareness (see Brookhuis, de Waard, & Janssen, 2001;
de Winter, Happee, Martens, & Stanton, 2014;Stanton
&Young,1998). A partial overview of the wider impli-
cations of automated vehicles has been recently made by
Fagnant and Kockelman (2015)withtheaimtoprovide
an order-of-magnitude estimation about the possible
©  Dimitris Milakis, Bart van Arem, and Bert van Wee. Published with license by Taylor& Francis.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-
nd/./), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon
in any way.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 325
economic impacts of automated vehicles in the US
context.
The remainder of this paper is structured as follows.
Our methodology is rst described (Section 2) and then a
simplied concept, to represent the areas of possible pol-
icy and society related implications of automated vehi-
cles, is presented (Section 3). In Sections 4–6 the results of
our analysis about the rst, second, and third order impli-
cations of automated driving are presented, respectively.
Every sub-section in Sections 4–6 is structured in two
parts.Therstpartpresentstheanalysisaboutthepos-
sible implications of automated driving and their mecha-
nisms (assumptions) and the second part is the review of
the respective results found in existing literature (litera-
ture results). Section 7presents conclusions and summa-
rizes directions for future research.
Methodology
Our methodology involves two steps. First, a simplied
concept is developed in a structured and holistic way,
representing what the possible implications of automated
vehicles are. Then, (a) the impacts of automated driv-
ing and their respective mechanisms, (b) existing litera-
ture results about these implications, and (c) research gaps
between possible impacts and existing literature results
are identied.
The impacts of automated driving and their respective
mechanisms are explored, based on our own analytical
thinking. Then, the literature results about the implica-
tions of automated driving are reviewed based on Sco-
pus and Web of Science listed peer-reviewed journal arti-
cles.IncludedinourreviewwerearticlesdateduptoJan-
uary 2017 containing in the title, abstract, or keywords
any combination of the following keywords: advanced
driver assistance system(s), [cooperative (C)] adaptive
cruise control (ACC), vehicle automation, autonomous
vehicle(s), autonomous car(s), self-driving vehicle(s), self-
driving car(s), driverless vehicle(s), driverless car(s), auto-
mated vehicle(s), automated car(s), automated driving,
robocar(s), and the keywords appearing in Table 1 for
each area of implication. We primarily limited our review
to peer-reviewed academic literature for two reasons: (a)
the number of articles is already very high and (b) explicit
review is an indication of quality. This does not mean that
other literature does not have sucient quality. There-
fore,inthecaseofverylimitedornoresultsforspe-
cic implications of automated vehicles, our search was
expanded to Google and Google Scholar, aiming to iden-
tify any unpublished reports of systematic studies. We
did not include any policy reports on automated vehi-
cles produced by governments or other institutions in our
review.
Tab le . Keywords used to identify scholarly articles about the
implications of automated vehicles.
Implication Keyword
Travel cost Cost, travel time, comfort, value of time,
travel time reliability
Road capacity Capacity, congestion, traffic flow
Travel choices Travel choice(s), mode choice(s), travel
behavior, travel distance, vehicle
kilometers traveled, vehicle miles
traveled, modal shift
Vehicle ownership and
sharing
Vehicle ownership, car ownership, vehicle
sharing, car sharing, ride sharing,
shared vehicle(s)
Location choices and land
use
Location choice(s), land use(s),
accessibility, residential density, urban
form, urban structure, urban design
Transport infrastructure Road infrastructure(s), road planning, road
design, intersection design, parking
infrastructure(s), public transport
service(s), transit service(s), cycle
lane(s), cycle path(s), sidewalk(s),
pavement(s)
Energy consumption and
air pollution
Fuel, energy, emissions, pollution
Safety Safety, accident(s), crash(es), risk,
cyberattack(s)
Social equity Social equity, social impact(s), vulnerable
social group(s), social exclusion
Economy Economy, productivity, business(es)
Public health Public health, human health, morbidity,
mortality
This paper focuses on passenger transport and
employs the Society of Automotive Engineers (SAE)
International (2016) taxonomy, which denes ve levels
of vehicle automation. In level 1 (driver assistance) and
level 2 (partial driving automation), the human driver
monitors the driving environment and is assisted by a
driving automation system for execution of either the
lateralorlongitudinalmotioncontrol(level1)orboth
motion controls (level 2). In level 3 (conditional driving
automation), an automated driving system performs all
dynamic tasks of driving (monitoring of the environment
and motion control), but the human driver is expected
to be available for occasional control of the vehicle. In
level 4 (high driving automation) and level 5 (full driving
automation) an automated driving system performs all
dynamic tasks of driving, without any human interven-
tion at any time. In level 4, the automated driving system
controls the vehicle within a prescribed operational
domain (e.g. high-speed freeway cruising, closed campus
shuttle). In level 5, the automated driving system can
operate the vehicle under all on-road conditions with no
design-based restrictions.
The ripple eect of automated driving
The ripple model was used to conceptualize the sequen-
tial eects that automated driving might bring to several
aspects of mobility and society (see Milakis, van Arem,

326 D. MILAKIS ET AL.
Figure . The ripple effect of automated driving.
&vanWee,2015). The “ripple eect” has been widely
used to describe the sequentially spreading eects of
events in various elds including economics, psychology,
computer science, supply chain management, and biblio-
metric analysis of science (see e.g. Barsade, 2002;Black,
2001;Cooper,Orford,Webster,&Jones,2013;Frandsen&
Nicolaisen, 2013; Ivanov, Sokolov, & Dolgui, 2014;Meen,
1999). The ripple model of automated driving is presented
in Figure 1. Driving automation is placed in the center of
thegraphtoreectthesourceofthesequentialrst,sec-
ond, and third order eects in the outer ripples. The rst
ripple comprises the implications of automated driving on
trac, travel cost, and travel choices. The second ripple
includes implications of automated driving with respect to
vehicle ownership and sharing, location choices and land
use, and transport infrastructure. The third ripple con-
tains the wider societal implications (i.e. energy consump-
tion, air pollution, safety, social equity, economy, and
public health) of the introduction of automated vehicles.
The ripple model of automated driving does not hold
the exact same properties as the respective ripple model
in physics that describes the diusion of waves as a func-
tion of time and distance. Therefore, the ripple model of
automated driving should not be taken too strictly. Feed-
backscanoccurinourmodel.Forexample,changesin
travel cost (rst ripple) might inuence accessibility, then
subsequently location choices, land use planning, and real
estate investment decisions (second ripple), which in turn
could aect travel decisions (e.g. vehicle use) and trac
(rst ripple). Also, there might be no time lag between
sequential eects. For example, vehicle use changes will
immediately result in safety or air pollution changes.
Finally, it should be clear that eects on fuel consump-
tion, emissions and accidents risk can occur soon after the
introduction of automated vehicles, yet the wider (soci-
etal) impacts on energy consumption, air pollution, and
safety (third ripple) can be evaluated only after changes
in the rst two ripples are taken into account.
First-order implications of automated driving
In this section the rst-order implications of automated
driving on travel cost, road capacity, and travel choices are
explored (see also Table 2 for an overview of studies on
rst-order implications for automated vehicles).
Travel cost
Assumptions
Potential implications for both the xed (capital) cost
of owning an automated vehicle and the generalized
transport cost (GTC), which comprises eort, travel time,

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 327
Tab le  . Summary of literature review results.
Possible effect of
automated vehicles Effect Comments Source
First-order implications
Trav el co st
Fixed cost of automated
vehicles
+Current automated vehicle applications cost several times the
price of a conventional vehicle in the US, but the price could be
gradually reduced to $ or even lower with mass
production and the technological advances of automated
vehicles.
Fagnant & Kockelman, 
Travel comfort ? Comfort has been incorporated in trajectory planning and ACC
algorithms as the optimizing metric. Motion sickness, apparent
safety and natural human-like paths could be included in path
planning systems. Time headway between vehicles below
.–. seconds can influence comfort.
Dang,Wang,Li,&Li, Elbanhawi et al., ;
Glaser et al., ; Lewis-Evans et al., ;Li
et al., ;Luoetal.,; Moon et al., ;
Raimondi & Melluso, ; Siebert et al., ;
Bellem et al., ; Diels & Bos, ; Lefèvre
et al., 
Trav el time −Vehicle automation can reduce delays on highways, at
intersections and in contexts involving shared automated
vehicles.
Arnaout & Arnaout, ; Dresner & Stone, ;
Fajardo et al., ; Ilgin Guler et al., ;
International Transport Forum, ; Kesting
et al., ; Khondaker & Kattan, ;Levin
et al., ;Lietal.,; Ngoduy, ;Yang
et al., ;Zohdy&Rakha,
Value of time ? Automated vehicles (level  and higher) could reduce the value of
time. Yet, value of time could increase for users of automated
vehicles as egress mode to train trips. The ability to work on
the move is not perceived as a major advantage of an
automated vehicle.
Cyganski, Fraedrich, & Lenz, ; Milakis et al.,
; Yap et al., 
Road capacity
Highway capacity +The higher the level of automation, cooperation and penetration
rate, and the higher the positive impact on road capacity. A
% penetration rate of CACC appears to be a critical threshold
for realizing significant benefits on capacity (>%), while a
% penetration rate of CACC could theoretically double
capacity. Capacity impacts at level  or higher levels of vehicle
automation and more advanced levels of cooperation among
vehicles, but also between vehicles and infrastructure, could
well exceed this theoretical threshold. Capacity might be
affected by vehicle heterogeneity. Capacity could decrease in
entrance/exitofautomatedhighwaysystems.
Arnaout & Bowling, ; Arnaout & Arnaout,
; Delis, Nikolos, & Papageorgiou, ;
Fernandes, Nunes, & Member, ;Grumert,
Ma, & Tapani, ; Hoogendoorn, van Arem, &
Hoogendoorn, ; Huang, Ren, & Chan,
; Michael, Godbole, Lygeros, & Sengupta,
; Monteil, Nantes, Billot, Sau, & El Faouzi,
; Ngoduy, ; Rajamani & Shladover,
; Shladover, Su, & Lu, ; van Arem, van
Driel, & Visser, ; Yang, Liu, Sun, & Li, ;
Carbaugh et al., ;Halletal.,;LeVine
et al., ; Michael et al., ;Talebpour&
Mahmassani, ;Wangetal.,a,b;Xie
et al., ;Zhouetal.,)
Intersection capacity +Significant capacity benefits (more than %, under certain
conditions) are expected from automated intersection control
systems.
Clement, Taylor, & Yue,  Kamal et al., 
Travel choices
Vehicle miles traveled +Automated vehicles could induce an increase in travel demand of
between % and % due to changes in destination choice (i.e.
longer trips), mode choice (i.e. modal shift from public
transport and walking to car), and mobility (i.e. more trips,
especially from people currently experiencing travel
restrictions; e.g. elderly). Shared automated vehicles could
result in additional VMT because of their need to move or
relocate with no one in them to serve the next traveler. Extra
VMT are expected to be lower for dynamic ride-sharing
systems.
Childress, Nichols, & Coe,  Fagnant &
Kockelman, ,;Gucwa,;
International Transport Forum, ; Malokin
et al., ; Correia, de, & van Arem, ;
Fagnant & Kockelman, ; Lamondia et al.,
;Levin&Boyles,; Milak is et al., ;
Vogt et al., ;Zmudetal.,
Second-order implications
Vehicle ownership −Shared automated vehicles could replace from about % up to
over % of conventional vehicles delivering equal mobility
levels. The overall reduction of the conventional vehicle fleet
could vary according to the automated mode (vehicle-sharing,
ride-sharing, shared electric vehicle), the penetration rate of
shared automated vehicles and the presence or absence of
public transport.
Fagnant & Kockelman, ; International
Transport Forum, ; Spieser et al., ;
Boesch, Ciari, & Axhausen, ;Chenetal.,
; Fagnant & Kockelman, ; Zhang et al.,

Location choices and land use ? Automated vehicles could enhance accessibility citywide,
especially in remote rural areas, triggering further urban
expansion. Automated vehicles could also have a positive
impact on the density of economic activity at the center of the
cities. Parking demand for automated vehicles could be shifted
to peripheral zones. Parking demand for shared automated
vehicles can be high in city centers, if empty cruising is not
allowed.
Childress et al., ; Zakharenko, ; Zhang
et al., 
(Continued on next page)

328 D. MILAKIS ET AL.
Tab le  . (Continued)
Possible effect of
automated vehicles Effect Comments Source
Transport infrastructure −Shared automated vehicles could significantly reduce parking
space requirements up to over %. The overall reduction of
parking spaces could vary according to the automated mode
(vehicle-sharing, ride-sharing, shared electric vehicle), the
penetration rate of shared automated vehicles and the
presence or absence of public transport. Less wheel wander
and increased capacity because of automated vehicles could
accelerate pavement-rutting damage. Increase in speed of
automated vehicles could compensate for such negative effect
by decreasing rut depth.
Fagnant & Kockelman, , ; International
Transport Forum, ; Boesch et al., ;
Chen et al., ;Chenetal.,; Spieser
et al., ; Zhang et al., 
Third-order implications
Energy consumption and air pollution
Fuel efficiency +Significant fuel savings can be achieved by various longitudinal,
lateral (up to %), and intersection control (up to %)
algorithms and optimization systems for automated vehicles.
Higher level of automation, cooperation, and penetration rate
could lead to higher fuel savings.
Asadi & Vahidi, ; Kamal et al., ;
Kamalanathsharma & Rakha, ; Khondaker
& Kattan, ;Lietal.,;Luoetal.,;
Manzie et al., ; Rios-torres & Malikopoulos,
; Vajedi & Azad, ;Wangetal.,;Wu
et al., ;Zohdy&Rakha,
Energy consumption (long
term)
? Battery electric shared automated vehicles are associated with
significant energy savings (–%) in the long term. The
energy gains are attributed to more efficient travel and
electrification. Several factors could lead to increased energy
use (e.g. longer travel distances and increased travel by
underserved populations such as youth, disabled, and elderly).
Thus, the net effect of vehicle automation on energy
consumption remains uncertain.
Brown et al., ; Greenblatt & Saxena, ;
Wadud et al., 
Emissions −Vehicle automation can lead to lower emissions of NOx, CO, and
CO. Higher level of automation, cooperation and penetration
rates could lead to even lower emissions. Shared use of
automated vehicles could further reduce emissions (VOC and
CO in particular) because of lower number of times vehicles
start.
Choi & Bae, ; Fagnant & Kockelman, ;
Grumert et al., ; Ioannou & Stefanovic,
;Wangetal.,; Bose & Ioannou, 
Air pollution (long term) ? Long-term impacts of battery electric shared automated vehicles
are associated with up to % less GHG. Yet, the net effect of
vehicle automation on GHG emissions remains uncertain.
Greenblatt & Saxena, ; Wadud et al., ;
Fagnant & Kockelman, 
Safety +Advanced driver assistance systems and higher levels of
automation (level  or higher) can enhance traffic safety.
Behavioral adaptation, cyberattacks, maliciously controlled
vehicles and software vulnerabilities can compromise traffic
safety benefits. Fully automated vehicles might not deliver
high safety benefits until high penetration rates of these
vehicles are realized.
Dresner & Stone, ; Ferguson, Howard, &
Likhachev, ;Hayashi,Isogai,
Raksincharoensak, & Nagai, ;Hou,Edara,&
Sun, ; Khondaker & Kattan, ;Kuwata
et al., ; Lee, Choi, Yi, Shin, & Ko, ; K.-R.
Li,Juang,&Lin,; Liebner, Klanner,
Baumann, Ruhhammer, & Stiller, ;
Martinez & Canudas-de-Wit, ;Shim,
Adireddy, & Yuan, ;M.Wang,
Hoogendoorn, Daamen, van Arem, & Happee,
;Carbaughetal.,;Spyropoulou,
Penttinen, Karlaftis, Vaa, & Golias, ;
Amoozadeh et al., ; Brookhuis et al., ;
Gerdes et al., ;Gouyetal.,;
Hoedemaeker & Brookhuis, ; Markvollrath
et al., ; Petit & Shladover, ;
Rudin-Brown & Parker, ;Strandetal.,;
Xiong et al., ; Young & Stanton, ; Dixit
et al., ;Gongetal.,; Naranjo et al.,

Social equity ? In-vehicle technologies can have positive effects (i.e. avoiding
crashes, enhancing easiness and comfort of driving, increasing
place, and temporal accessibility) for elderly. Automated
vehicles could induce up to % additional travel demand from
the non-driving, elderly, and people with travel-restrictive
medical conditions. Automated vehicles offer the opportunity
to incorporate social justice aspects in future traffic control
systems.
Harper, Hendrickson, Mangones, & Samaras,
;Ebyetal.,;Mladenovic&
McPherson, 
Economy ? Social benefits per automated vehicle per year could reach $
when there’s a % market share of automated vehicles. Jobs
in the transportation and logistics sectors have a high
probability of being replaced by computer automation within
the next two decades.
Fagnant & Kockelman, ;Frey&Osborne,
Public health ? No systematic studies were found about the implications of
automated vehicles for public health.
Note. Effects are described with the following symbols: ‘+’: positive/increase, ‘−’: negative/decrease, ‘?’: uncertain/limited evidence

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 329
and nancial costs of a trip, are explored. The xed costs
of automated vehicles will very likely be higher than for
conventional vehicles due to the advanced hardware and
software technology involved. The increased xed cost
could inuence the penetration rate and subsequently the
magnitude of the eects of automated vehicles. The GTC,
ontheotherhand,isexpectedtodecreasebecauseof
lowereort,time,andmoneyneededtotravel.First,more
travel comfort, enhanced travel safety, higher travel time
reliability, and the possibility to perform activities other
than driving (like working, meeting, eating, or sleeping)
while on the move will likely lead to lower values of
time. Second, less congestion delays because of increased
road capacity and reduced (or even eliminated) search
time for parking owing to self-parking capability, but also
increased use of shared vehicles, would possibly require
less travel time. Third, enhanced eciency of trac ow
along with more fuel-ecient vehicles because of their
lighter design (owing to less risk of having an accident)
couldalsoreducethemonetarycostoftravel.Dueto
shorter headways, air resistance will possibly decrease,
further reducing fuel use and costs. However, potential
increase of vehicle travel demand because of enhanced
road capacity, reduced GTC, and/or proliferation of vehi-
cle sharing systems and urban expansion in the longer
term, could compromise travel time and cost savings. The
counter eects of increased vehicle demand could include
increased congestion delays, longer trips, and more fuel
costs.
Literature results
Fagnant and Kockelman (2015) report estimations that
current automated vehicle applications cost several times
the price of a conventional vehicle in the US. However,
they estimate that this dierence in cost could be gradu-
ally reduced to $3000 or even lower with mass production
and the technological advances of automated vehicles.
Looking at the components of GTC, several studies have
incorporated comfort in terms of longitudinal and lateral
acceleration as the optimizing metric in their trajectory-
planning algorithms (see e.g. Glaser, Vanholme, Mam-
mar, Gruyer, & Nouvelière, 2010;Raimondi&Melluso,
2008). Moreover, multi-objective ACC algorithms usually
incorporate ride comfort (measured in terms of vehicle
acceleration) along with safety and fuel consumption as
system constraints (see e.g. Dang, Wang, Li, & Li, 2015;
Li, Li, Rajamani, & Wang, 2011;Luo,Chen,Zhang,&Li,
2015;Moon,Moon,&Yi,2009). Bellem, Schönenberg,
Krems, and Schrauf (2016) suggested several maneuver-
specic metrics such as acceleration, jerk, quickness, and
headway distance to assess comfort of automated driving
style. However, Elbanhawi, Simic, and Jazar (2015)argue
in their review paper that several factors of human com-
fort are largely ignored in research for autonomous path
planning systems [i.e. motion sickness, see also Diels &
Bos, 2016; apparent safety (the feeling of safe operation
of the automated vehicle); natural, human-like paths]. A
more recent study (Lefèvre, Carvalho, & Borrelli, 2016)
developed a learning-based approach for automated vehi-
cles with the aim to replicate human-like driving styles
(i.e. velocity control). Moreover, research has shown that
comfort is not only inuenced by vehicle acceleration
but also by the time headway when the driver is still in
the loop. Both Lewis-Evans, De Waard, and Brookhuis
(2010) and Siebert, Oehl, and Pster (2014) identied in
driver simulator experiments a critical threshold for time
headway in the area of 1.5–2.0 seconds below which a
driver’s perception of comfort reduces signicantly.
Limited evidence exists on the impacts of automated
vehicles on the travellers’ value of time. Yap, Correia, and
van Arem (2016)foundahighervalueoftimeforusing
fully automated (level 5) compared to manually driven
vehicles as egress mode of train trips in a stated preference
survey in the Netherlands. These researchers attributed
thisresulttothepossibleuncomfortablefeelingoftrav-
elers with the idea of riding an automated vehicle, the
lack of any real-life experience with automated vehicles,
and the fact that an egress trip is typically a short trip not
allowing the travelers to fully experience potential bene-
ts of automated vehicles such as travel safety. Cyganski
et al., (2015) reported that only a minor percentage
of the respondents in their questionnaire survey in
Germany declared as an advantage the ability to work on
the move in an automated vehicle (level 3 and higher). On
the contrary, most respondents agreed that activities that
they usually undertake while driving conventional vehi-
cles (e.g. gazing, conversing, or listening to music) would
continue to be important when riding an automated vehi-
cle. Respondents working in their current commute were
foundtobemorelikelytowishtoworkinanautomated
vehicle as well. Milakis, Snelder, van Arem, van Wee, and
Correia (2017) reported a possible decrease of the value of
time between 1% and 31% for users of automated vehicles
(level 3 and higher) in various scenarios of development
of automated vehicles in the Netherlands.
Several studies have reported results about travel time
and fuel savings based on simulation of various control
algorithms for automated car-following scenarios and
automated intersection management. Studies about fuel
savings are presented later in this article. Considering
travel time, Arnaout and Arnaout (2014)simulateda
four-lane highway involving several scenarios of pene-
tration rates for cars equipped with CACC and a xed
percentage for trucks (10%). They found that travel
time decreased substantially with the increase of CACC
penetration rate. Ngoduy (2012) reported that a 30% pen-
etration rate of ACC could signicantly reduce oscillation
waves and stabilize trac near a bottleneck, thus reducing

330 D. MILAKIS ET AL.
travel time by up to 35%. Kesting, Treiber, Schönhof, and
Helbing (2008) identied travel time improvements even
with relatively low ACC penetration rates. Also, Khon-
daker and Kattan (2015) showed that their proposed
variable speed limit control algorithm could reduce travel
time by up to 20% in a context of connected vehicles
compared to an uncontrolled scenario. However, travel
time improvements were lower when a 50% penetra-
tion rate of connected vehicles was simulated. Zohdy
and Rakha (2016) developed an intersection controller
that optimizes the movement of vehicles equipped with
CACC. Their simulation results showed that the average
intersection delay in their system (assuming 100% market
penetration of fully automated vehicles, level 4 or 5) was
signicantlylowercomparedtothetracsignalandall-
way-stop control scenarios. Similarly, Dresner and Stone
(2008) proposed a multi-agent, reservation-based con-
trol system for ecient management of fully automated
vehicles (level 4 or 5) in intersections that could widely
outperform current control systems like trac lights and
stop signs. According to these researchers, this system
could oer near-to optimal delays (up to 0.35 seconds);
about ten times lower than the delays observed in con-
ventional control systems. The eciency of reservation-
based intersection controls in reducing delays was also
demonstrated by Fajardo, Au, Waller, Stone, and Yang
(2012),Li,Chitturi,Zheng,Bill,andNoyce(2013)and
Levin, Fritz, and Boyles (2016). Yet, Levin, Boyles, and
Patel (2016) indicated some cases that optimized signals
can outperform reservation-based intersection controls
(e.g. in local road-arterial intersections) and thus, these
researchers recommended a network-based analysis
before any decision about replacement of trac sig-
nals is taken. Ilgin Guler, Menendez, and Meier (2014)
assumedthatonlyaportionofthevehicleswereequipped
with their intersection control algorithm and tested the
impactsondelaysfortwoone-way-streets.Theirsimu-
lations revealed a decrease by up to 60% in the average
delay per car when the penetration rate of the control
system-equipped vehicles increased by up to 60%. These
researchers reported further decrease of the delays by an
improved version of their intersection controller (Yang,
Guler, & Menendez, 2016). Chen, Bell, and Bogenberger
(2010) proposed a navigation algorithm for automated
vehicles that accounts not only for travel time but also
for travel time reliability. Thus, this algorithm can search
for the most reliable path within certain travel time
constraints using either dynamic or no trac informa-
tion. Finally, when considering the impacts of shared
automated vehicles on travel time, the International
Transport Forum (2015) reported a reduction of up to
37.9% compared to the current travel time of private cars
in Lisbon, Portugal, based on a simulation study.
Road capacity
Assumptions
Automated vehicles could have a positive inuence on free
ow capacity, the distribution of vehicles across lanes and
trac ow stability by providing recommendations (or
even determining in level 3 or higher levels of automa-
tion) about time gaps, speed and lane changes. Enhanced
free ow capacity and decreased capacity drops (i.e. fewer
episodes of reduced queue discharge rate) could increase
the road capacity and thus reduce congestion delays.
Nevertheless, benets in trac ow eciency will very
likely be highly dependent on the level of automation, the
connectivity between vehicles and their respective pen-
etration rates, the deployment path (e.g. dedicated lanes
versus integrated, mixed trac) as well as human factors
(i.e. behavioral adaptation). Moreover, increased vehi-
cle travel demand could have a negative impact on road
capacity owing to more congestion delays and subse-
quently increased capacity drops. Thus, although the ben-
ets of automated vehicles in the short term are expected
to be important, the long-term implications are uncertain
and highly dependent on the evolution of vehicle travel
demand.
Literature results
Hoogendoorn, van Arem, and Hoogendoorn (2014)
concluded in their review study that automated driving
might be able to reduce congestion by 50%, while this
reduction could go even higher with the help of vehicle-
to-vehicle and vehicle-to-infrastructure communication.
Several studies have explored the trac impacts of lon-
gitudinal automation (i.e. ACC and CACC), based on
simulations. Results suggest that ACC can only have a
slight impact on capacity (Arnaout & Arnaout, 2014).
CACC, on the other hand, showed positive impacts on
capacity (van Arem, van Driel, & Visser, 2006) but these
will probably only be important (e.g. >10%) if relatively
high penetration rates are realized (>40%) (Arnaout
&Bowling,2011; Shladover, Su, & Lu, 2012). A 100%
penetration rate of CACC could theoretically result in
double capacity compared to a scenario of all manually
driven vehicles (Shladover et al., 2012). Ngoduy (2013)
and Delis, Nikolos, and Papageorgiou (2015)havealso
conrmed that CACC performs better than ACC with
respect to both trac stability and capacity.
Several other studies have conrmed the benecial
eects of dierent types and levels of vehicle automa-
tion and cooperation on capacity in various trac sce-
narios (see e.g. Talebpour & Mahmassani, 2016). In par-
ticular, Fernandes, Nunes, and Member (2015)proposed
an algorithm for positioning and the cooperative behav-
ior of multiplatooning leaders in dedicated lanes. Their

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 331
simulations showed that the proposed platooning sys-
tem can achieve high trac capacity (up to 7200 vehi-
cles/hour) and outperform bus and light rail in terms of
capacity and travel time. Huang, Ren, and Chan (2000)
designed a controller for automated vehicles that requires
information only from vehicle sensors. Their simulations
in mixed trac conditions that involved both automated
and human controlled vehicles showed that peak ow
could reach 5000 vehicles/hour when 70% of the vehicles
are automated. Moreover, Michael, Godbole, Lygeros, and
Sengupta (1998) showed, via the simulation of a single
lane automated highway system, that capacity increases
as the level of cooperation between vehicles and pla-
toon length increases. Several other studies have not only
reported enhanced trac ow eciency because of coop-
eration and exchange of information between vehicles
(e.g.Monteil,Nantes,Billot,Sau,&ElFaouzi,2014;Wang,
Daamen, Hoogendoorn, & van Arem, 2016b;Xie,Zhang,
Gartner, & Arsava, 2016;Yang,Liu,Sun,&Li,2013;
Zhou, Qu, & Jin, 2016)butalsobetweenvehiclesand
infrastructure (e.g. variable speed limits, see Grumert,
Ma, & Tapani, 2015; Wang, Daamen, Hoogendoorn, &
van Arem, 2016a). Rajamani and Shladover (2001)com-
pared the performance of autonomous control systems
and cooperative longitudinal control systems (with and
without inter vehicle communication respectively). These
researchersshowedanalyticallythattheautonomous
control system could indeed deliver capacity benets
reaching a theoretical maximum trac ow of 3000
vehicles/hour. However, a cooperative system comprising
10-vehicle platoons with a distance between the vehicles
of 6.5 m was far more ecient, achieving a theoretical traf-
c ow of 6400 vehicle/hour. Theoretical trac ow of the
cooperative system could increase to 8400 vehicles/hour
if the distance between the vehicles in the platoons was
further reduced to 2 m.
Another group of studies identify signicant capacity
benets from using automated intersection control sys-
tems.Clement,Taylor,andYue(2004)proposedoneof
these conceptual systems whereby vehicles can move in
closely spaced platoons when the lights turn to green at
signalized intersections. These researchers showed ana-
lytically that this system could increase throughput by
163% compared to current road intersections even when
they used quite conservative values for vehicle spacing
in the platoons (i.e. 7.2 m). Kamal, Imura, Hayakawa,
Ohata, and Aihara (2015)developedacontrolsystem
which coordinates connected vehicles so they can safely
and smoothly cross an intersection with no trac lights.
Both their estimations and simulations showed an almost
100% increase in capacity compared to the performance
of a traditional signalized intersection. It should be noted
that both Clement, Taylor, and Yue (2004)andKamal,
Imura, Hayakawa, Ohata, and Aihara (2015) assumed
in their studies 100% market penetration of fully auto-
mated vehicles (level 4 or 5), no other road users (bicy-
clists or pedestrians), and perfect control performance
(no errors).
However, some studies have identied possible trade-
os between increases in capacity and various aspects of
automated vehicles. Le Vine, Zolfaghari, and Polak (2015)
identied a possible trade-o between comfort level and
intersection capacity. These researchers showed that if the
passengers of automated vehicles were to enjoy comfort
levels similar to light rail or high-speed rail (in terms of
longitudinal and lateral acceleration/deceleration), inter-
section capacity reduction could reach 53% and delays
could increase by up to 1924%. Van den Berg and
Verhoe f ( 2016) showed that automated vehicles could
have both positive and negative externalities through
increases in capacity and parallel decreases in the value of
time, although net positive externalities seem more likely
according to their analysis. Moreover, Carbaugh, God-
bole, and Sengupta (1998) showed that the probability of
rear-end crashes in automated highway system platoons
(level 4) increases as capacity increases, especially when
intra-platoon spacing becomes very small (e.g. 1 m). Yet,
collision severity tends to decrease because speed dier-
ences associated with crashes become smaller in higher
capacity. The results of this study refer to the rst rear-
end crash between two vehicles and not to secondary
crashesinaplatoonofvehicles.Also,Hall,Nowroozi,
and Tsao (2001)pointedtopossiblecapacityreductions
in entrance/exit of automated highway systems relative to
the ideal ‘pipeline’ capacity without any entrances or exits,
whileMichael,Godbole,Lygeros,andSengupta(1998)
showed that capacity in automated highway systems could
decrease compared with passengers cars, when trucks and
buses are added.
Travel choices
Assumptions
In the short term, the increase of road capacity, the
subsequent congestion relief and the decrease in GTC
couldleadtoanincreaseofvehicletraveldemand.
However, vehicle travel demand might also increase
because of transfers, pick-ups, drop-os, and repositions
of ride-sharing and vehicle-sharing vehicles. Moreover,
the decrease of GTC could enhance the accessibility of
moredistantlocations,thusallowingpeopletochoose
such destinations to live, work, shop, recreate, and sub-
sequently increase the amount of their daily vehicle use.
Theincreaseinvehicleusemightalsobetheresultof
a modal shift from conventional public transport. For
example, buses could be gradually replaced by more

332 D. MILAKIS ET AL.
exible, less costly, and easier to operate automated ride-
sharing and vehicle-sharing services. The use of high
capacity public transport systems, such as trains, metro,
andlightrailmightalsodropaftertheintroductionof
automated vehicles, if ride-sharing or vehicle-sharing
could adequately serve high-demand corridors. Finally,
the increase of ride-sharing and vehicle-sharing systems
might negatively inuence the use of active modes, since
automated shared vehicles could eectively serve short
distance trips or feeder trips to public transportation.
Also, further diusion of the activities across the city
might deter walking and bicycle use. However, the pos-
sibility that people still prefer active modes for short
and medium distances for exercise and health reasons
or simply because they like cycling or because cycling is
cheaper, cannot be excluded. Moreover, enhanced road
safety might also improve (the perception of) the safety
of bicycling and subsequently positively inuence cycle
use, especially among the more vulnerable cycling groups
(e.g. the elderly, children, and women; see Xing, Handy,
&Mokhtarian,2010;Milakis,2015).
Literature results
Fagnant and Kockelman (2015) estimated a 26% increase
of system-wide vehicle miles traveled (VMT) using a
90% market penetration rate of automated vehicles. This
estimation was based on a comparison with induced
travel demand caused by enhancement of road capac-
ity after the expansion of road infrastructures. Milakis
et al. (2017) reported a possible VMT increase between
3% and 27% for various scenarios of development of
automated vehicles in the Netherlands. Higher VMT
levels because of automated vehicles were identied by
Vogt, Wang, Gregor, and Bettinardi (2015)througha
fuzzy cognitive mapping approach that accounted for
interactions among several factors including emerging
mobility concepts (e.g. demand responsive services and
intelligent infrastructure). Also, Gucwa (2014) reported
an increase in VMT between 4% and 8% using dierent
scenariosofroadcapacityandvalueoftimechanges
through the introduction of automated vehicles. His sce-
nario simulations in the San Francisco Bay area involved
increases in road capacity of between 10% and 100% and
decreases in value of time to the level of a high quality
trainortohalfthecurrent(in-vehicle)valueoftime.In
the extreme scenario of zero time cost for traveling in
an automated vehicle the increase of VMT was 14.5%.
Additional vehicle travel demand in this study was due
to changes in destination and mode choices. Correia,
and van Arem (2016) reported an increase of 17% in
VMT after replacing all private conventional vehicles by
automated ones in simulations of the city of Delft, The
Netherlands. Increase in VMT was the result of more
automated vehicle trips either occupied (shifted from
public transport) or unoccupied (moving vehicles to nd
parking places with lower cost). Another study showed
that a modal shift of up to 1%, mainly from local public
transport (bus, light rail, subway) and bicycle, to drive-
aloneandshared-ridemodescouldbepossiblebecauseof
the ability to multitask in automated vehicles (Malokin,
Circella, & Mokhtarian, 2015). Levin and Boyles (2015)
conrmed the possibility of increased modal shift from
public transport to automated vehicles especially when
these vehicles become widely available to travellers with
lower value of time. Lamondia, Fagnant, Qu, Barrett, and
Kockelman (2016) focused on possible modal shift from
personal vehicles and airlines to automated vehicles for
long distance travel using Michigan State as case study.
These researchers found a modal shift of up to 36.7% and
34.9% from personal vehicles and airlines respectively to
automated vehicles for trips less than 500 miles. For trips
longer than 500 miles, automated vehicles appeared to
draw mainly from personal vehicles (at a rate of about
20%) and much less from airlines. Childress, Nichols, and
Coe (2015) used the Seattle region’s activity-based travel
model to explore the impacts of automated vehicles on
travel demand. They simulated four dierent scenarios
withrespecttotheAVpenetrationrateandchangesin
capacity, value of time, parking and operation costs. They
concluded that an increase of VMT between 4% and
20% is likely in the rst three scenarios that assumed
capacity increases of 30%. Additional VMT was the result
of both more and longer trips and also because of a
modal shift from public transport and walking to car.
Congestion delays appeared in only one of the rst three
scenarios that assumed a universal decline of value of
time by 35% along with reduced parking costs. In the
other two scenarios (with no or limited impact on the
value of time), capacity increases oset additional travel
demand, oering higher network speeds. In the fourth
and nal scenario, a shared autonomous vehicles-based
transportation system with users bearing all costs of
driving was assumed. Simulation results in this case
showed that VMT could be reduced by 35% with less
congestion delays. Signicantly higher user costs per mile
(up to about 11 times) induced shorter trip lengths, lower
single-occupant vehicle share and an increase of public
transport use and walking by 140% and 50%, respectively.
Fagnant and Kockelman (2014), on the other hand,
indicated in their agent-based simulation study that auto-
mated vehicle-sharing schemes could result in 10% more
VMT compared to conventional vehicles. The reason is
that shared automated vehicles will need to move or relo-
cate with no one in them to serve the next traveler. Yet,
extra VMT was found to be around 4.5% when dynamic
ride-sharing services were included in the simulation

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 333
(Fagnant & Kockelman, 2016). Extra VMT was even lower
when the ride-matching parameter (i.e. max time from
initial request to nal drop o at destination) for ride
sharing travelers was increased. Also, in their simula-
tion study for Lisbon, Portugal, the International Trans-
port Forum (2015)reportedanincreaseinVMTover
thecourseofadaythatcouldvarybetween6.4%and
90.9% depending on the mode (vehicle-sharing or ride-
sharing automated vehicles), the penetration rate, and the
availability of high-capacity public transport. It should
be noted that these studies did not take into account
any potential changes in travel demand because of the
introduction of automated vehicles. For example, Harper,
Hendrickson, Mangones, and Samaras (2016)estimated
that light-duty VMT could increase by up to 14% in the
US, only through the additional travel demand of the
non-driving, elderly, and people with travel-restrictive
medical conditions because of automated vehicles.
Finally, Zmud, Sener, and Wagner (2016)explored
impacts of automated vehicles on travel behavior using
face-to-face interviews with 44 respondents from Austin,
Texas. Contrary to the above modeling estimates, most
of the participants (66%) stated that their annual VMT
would remain the same if they would use an automated
vehicle,becausetheywouldnotchangetheirroutines,
routes, activities, or housing location. Twenty-ve percent
of the participants responded that they would increase
their annual VMT adding more long-distance, leisure,
and local trips to their existing travel patterns.
Second-order implications of automated driving
In this section the second-order implications of auto-
mated driving for vehicle ownership and sharing, loca-
tion choices and land use, and transport infrastructure are
explored (see also Table 2 for an overview of studies on
second-order implications of automated vehicles).
Vehicle ownership and sharing
Assumptions
The introduction of automated vehicles could facilitate
the development of ride-sharing and vehicle-sharing ser-
vices. Automated vehicles could signicantly reduce oper-
ational costs (e.g. no driver costs) for ride-sharing and
vehicle-sharing services. Such schemes could eectively
meet individuals’ travel demand needs with lower cost
and higher exibility compared to what today’s bus and
taxi systems oer to passengers. Subsequently, urban resi-
dents could decide to reduce the number of cars they own
or even live car-free, avoiding the xed costs associated
with car ownership as well. However, shared automated
vehicles might be utilized more intensively (e.g. additional
travel to access travellers or to relocate) than conventional
cars. We may thus expect shared automated vehicles to
wear out faster and to be replaced more frequently.
Literature results
Several studies have simulated transport systems to
explore the possibility of automated vehicles substituting
conventional vehicles. Fagnant and Kockelman (2014;
2016) simulated the operation of shared automated vehi-
cles (automated vehicles oering vehicle-sharing and
dynamic ride-sharing services) in an idealized mid-size
grid-based urban area and in Austin, Texas’ coded net-
work. These researchers reported that each shared auto-
mated vehicle could replace around eleven conventional
vehicles. This rate dropped to around nine in a scenario of
signicantly increased peak hour demand. Also, Zhang,
Guhathakurta, Fang, and Zhang (2015) and Boesch, Ciari,
and Axhausen (2016) indicated in hypothetical and real
city simulations (Zurich, Switzerland) that every shared
automated vehicle could replace around ten and fourteen
conventional vehicles, respectively. However, accord-
ing to Chen, Kockelman, and Hanna (2016)ifvehicle
charging is also taken into account in the case of shared,
electric, automated vehicles then the replacement rate of
privately owned vehicles drops between 3.7 and 6.8. The
International Transport Forum (2015)simulateddierent
scenarios of automated modes (automated vehicles for
ride-sharing and vehicle-sharing services), penetration
rates, and availability of high-capacity public transport.
This report indicated that shared automated vehicles
could replace all conventional vehicles, delivering equal
mobility levels with up to 89.6% (65% at peak-hours)
less vehicles in the streets (scenario of automated ride-
sharing services with high capacity public transport).
Another conclusion of this study is that less automated
ride-sharing than vehicle-sharing vehicles could replace
all conventional vehicles. The reductions in eet size were
much lower (varying between 18% and 21.8%) when
the penetration rate of shared automated vehicles was
assumedata50%levelandhigh-capacitypublictransport
was also available. Finally, Spieser et al. (2014)estimated
that only one third of the total number of passenger
vehicles would be needed to meet travel demand needs
if all modes of personal transportation vehicles were
replaced by shared automated vehicles (automated vehi-
cles oering vehicle-sharing services). These researchers
used analytical techniques and actual transportation data
in the case of Singapore for their study.
Location choices and land use
Assumptions
Automated vehicles could have an impact on both
the macro (regional) and micro (local) spatial scale.
At regional level, automated vehicles could enhance

334 D. MILAKIS ET AL.
accessibility by aecting its transportation, individual
and temporal components (see Geurs & van Wee, 2004
for an analysis of the accessibility components). Less
travel eort, travel time, and cost and thus lower GTC
could have an impact on the transportation component
of accessibility. People without access to a car (not owning
a car or not being able to drive) may travel to activities
using(shared)automatedvehicles,thusinuencingthe
individual component of accessibility. Moreover, (fully)
automated vehicles could perform certain activities
themselves (e.g. pick up the children from school or the
groceriesfromthesupermarket).Thiscouldovercome
any constraints resulting from the temporal availability
of opportunities (e.g. stores opening/closing times) and
individuals’ available time. Enhanced regional accessibil-
ity might allow people to compensate lower travel costs
with living, working, shopping, or recreating further
away. Thus, an ex-urbanization wave to rural areas of
former inner city and suburban residents could be pos-
sible,subjecttolandavailabilityandlandusepolicies.
Enhanced accessibility may also aect the development of
new centers. For example, former suburban employment
centers could evolve into signicant peripheral growth
poles, serving the increased demand for employment and
consumption of new ex-urban residents. The possibility
to eliminate extensive parking lots in these kinds of
centers because of the self-parking capability of (fully)
automated vehicles could further enhance the potential
ofmixed-usegrowthintheseareas.Atthelocallevel,
automated vehicles could trigger changes in streetscape,
building landscape design and land uses. First, the capa-
bility of self-parking and the opportunity of increased
vehicle-sharing services because of automated vehicles
could reduce demand for on-street and o-street park-
ing, respectively. Subsequently, parking lanes could be
converted into high occupancy vehicle lanes, bus lanes,
and cycle lanes or to new public spaces (e.g. green spaces
or wider sidewalks). A reduction of o-street parking
requirements could bring changes in land use (inll
residential or commercial development) and in building
design (i.e. access lanes, landscaping). Moreover, surface
parking lots and multi-story parking garages in central
areas could be signicantly reduced, enhancing inll
development potential for people-friendly land use.
Literature results
Childressetal.(2015) identied potential changes in
households’ accessibility patterns in Seattle, WA, in a sce-
nario where the transportation system of this region is
entirely based on automated vehicles. This scenario not
only assumed that driving is easier and more enjoyable
(increased capacity by 30% and decreased value of time
by 35%), but also cheaper because of lower parking costs.
An analysis was performed on an activity-based model for
a typical household type, using aggregate logsums to mea-
sure accessibility changes compared to a 2010 baseline
scenario. Results showed that the perceived accessibility
was universally enhanced across the whole region. The
highest increase in accessibility was observed for house-
holds living in more remote rural areas. Changes to acces-
sibility were also associated with an average increase of
20% in total VMT. The increase in travel demand was
far higher (up to 30.6%) in outlying areas. Zakharenko
(2016) analyzed the eects of fully automated vehicles
on urban form from an urban economics perspective.
This researcher developed a model of a monocentric two-
dimensional city of half-circular shape that was calibrated
to a representative US city. He assumed that workers could
chooseamongnocommute,traditionalvehiclecommut-
ing and commuting by an automated vehicle taking into
account variable, parking and xed costs of each choice.
According to the results, about 97% of the daily parking
demand would be shifted to a “dedicated parking zone” in
the periphery of the city center. This in turn would have
a positive impact on the density of economic activity at
the center of the city driving land rents 34% higher. On
the other hand, reduced transportation costs because of
automatedvehicleswouldcausethecitytoexpandand
land rents to decline about 40% outside the city center.
Finally, Zhang et al. (2015) showed in their agent-based
simulation of a hypothetical city that the longer the empty
cruising of shared automated vehicles the more evenly
distributedtheparkingdemandofthesevehicleswouldbe
throughout the study area. If no empty cruising is allowed
then parking demand of shared automated vehicles
tended to be concentrated in the center of the study area.
Transport infrastructure
Assumptions
Increased road capacity because of automated vehicles
could reduce future needs for new roads. However,
induced travel demand resulting from enhanced road
capacity, reduced GTC, and/or the proliferation of vehi-
cle sharing systems and urban expansion may reduce or
even cancel out or more than oset initial road capacity
benets.Inthelastcase(morethanoset),additional
road capacity may be required to accommodate new
travel demand. Automated vehicles will also be likely to
reduce demand for parking, thus, probably, fewer parking
infrastructures—either on-street or o-street—will be
required. Moreover, a reduced need for public transport
services in some areas (especially those with low and
medium densities) could also lead to public transport
service cuts. Pedestrians and cyclists could benet from
morespaceaftertheintroductionofautomatedvehicles

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 335
as a result of road capacity improvements. Finally, changes
in ownership, organizational structure and operation of
transport infrastructures might appear when fully auto-
mated vehicles (level 4 or 5) increase considerably their
share in the vehicle eet. According to Van Arem and
Smits (1997) these changes could include a segmentation
of the road network, operation and maintenance by
private organizations and the emergence of transporta-
tion providers that could guarantee trip quality, regardless
of the travel mode.
Literature results
International Transport Forum (2015) reported that
both on-street and o-street parking spaces could be
signicantly reduced (between 84% and 94%) in all sim-
ulated scenarios that assumed a 100% shared automated
vehicle eet in the city of Lisbon, Portugal. Yet, the reduc-
tion was only incremental or even non-existent when
these researchers tested scenarios with a 50% mix between
shared automated and conventional vehicles. Also, Chen,
Balieu, and Kringos (2016),Boesch,Ciari,andAxhausen
(2016) Fagnant and Kockelman (2014,2016), Zhang et al.
(2015)andSpieseretal.(2014) oered estimations about a
replacement rate of conventional vehicles by shared auto-
mated vehicles that varies between three and fourteen.
Thus, parking demand could be reduced from about 67%
up to over 90%.
Concerning the impact of automated vehicles on the
long-term service performance of road infrastructures,
Chen, Balieu, and Kringos (2016) showed that less wheel
wander and increased capacity could accelerate pavement
rutting damage, but potential increase in speed of auto-
mated vehicles could compensate for such negative eect
by decreasing rut depth.
Third-order implications of automated driving
In this section the third-order implications of automated
driving on energy consumption and air pollution, safety,
social equity, economy, and public health are explored
(see also Table 2 for an overview of studies on third-order
implications of automated vehicles).
Energy consumption and air pollution
Assumptions
The introduction of automated vehicles might result in
energy and emission benets because of reduced con-
gestion, more homogeneous trac ows, reduced air
resistance due to shorter headways, lighter vehicles (a
result of enhanced safety), and less idling (a result of
less congestion delays). Also, automated vehicles may
require less powerful engines because high speeds and
very rapid acceleration will not be needed for a large
share of the eet (e.g. shared automated vehicles). This
could further improve the fuel eciency and limit emis-
sions. Yet, privately owned automated vehicles could
still oer the possibility of mimicking dierent human
driving styles (e.g. fast, slow, and aggressive). Moreover,
the possibility that automated vehicles will be larger
than conventional vehicles, serving passengers’ needs to
perform various activities while on the move, cannot be
excluded. For example, extra space might be needed to
facilitate oce-like work (table and docking stations),
face-to-face discussions (meeting table) or sleeping, and
relaxing (couch, bed). Larger vehicles may limit fuel e-
ciency gains in this case. Shorter search time for parking
and reduced needs for construction and maintenance of
parking infrastructures can also lead to environmental
benets. However, shared automated vehicles may be pro-
grammed to drive continuously until the next call rather
than try to nd parking in a downtown area, generating
more emissions. Additionally, an automated vehicle may
be programmed to drive itself outside of the downtown
center to an area where parking is cheaper or free, thus
consuming more energy, producing more emissions and
creating more trac congestion. Finally, a smaller eet
size could be associated with lower energy and emissions
for car manufacturing and road infrastructure develop-
ment. Nevertheless, the potential environmental benets
of automated vehicles could be signicantly mitigated by
increased travel demand in the long term.
Literature results
Several studies have reported fuel savings from vehicle
automation systems. Wu, Zhao, and Ou (2011) demon-
strated a fuel economy optimization system that provides
human drivers or automated systems with advice about
optimal acceleration/deceleration values, taking into
account vehicle speed and acceleration, but also current
speed limit, headway spacing, trac lights, and signs.
Their driving simulator experiment in urban conditions
with signalized intersections revealed a decrease in fuel
consumption of up to 31% for the drivers who used
the system. Khondaker and Kattan (2015) reported fuel
savings of up to 16% for their proposed variable speed
limit control algorithm compared with an uncontrolled
scenario. Their control system incorporated real-time
information about individual driver behavior (i.e. accel-
eration/deceleration and level of compliance with the
set speed limit) in a context of 100% connected vehicles.
Yet, fuel savings were lower when the penetration rate of
connected vehicles was assumed at a 50% level. Also, Li,
Peng, Li, and Wang (2012) showed that the application
of a Pulse-and-Gliding (PnG) controller could result in
fuelsavingsofupto20%comparedtoalinearquadratic

336 D. MILAKIS ET AL.
(LQ)-based controller in automated car-following sce-
narios. Other studies have also reported signicant fuel
consumption savings in eld and simulation tests of their
ACC and CACC control algorithms (see e.g. Eben Li,
Li, & Wang, 2013; Kamal, Taguchi, & Yoshimura, 2016;
Luo, Liu, Li, & Wang, 2010; Rios-torres & Malikopoulos,
2016; Wang, Hoogendoorn, Daamen, & van Arem, 2014)
including controllers for hybrid electric vehicles (Luo,
Chen,Zhang,&Li,2015; Vajedi & Azad, 2015)
In a context where there are intersections, the con-
troller proposed by Zohdy and Rakha (2016)provides
advice about the optimum course of vehicles equipped
with CACC. These researchers reported fuel savings
of, on average, 33%, 45%, and 11% for their system
compared with the conventional intersection control
approaches of a trac signal, all-way-stop, and round-
about, respectively. Moreover, Ala, Yang, and Rakha
(2016), Kamalanathsharma and Rakha (2016)andAsadi
and Vahidi (2011) reported fuel savings up to 19%,
30%, and 47%, respectively, for their cooperative adap-
tive cruise controller that uses vehicle-to-infrastructure
(trac signal in this case) communication to optimize
a vehicle’s trajectory in the vicinity of signalized inter-
sections. Finally, Manzie, Watson, and Halgamuge (2007)
showed that vehicles exchanging trac ow information
through sensors and inter-vehicle communication could
achieve the same (i.e. 15–25%) or even more (i.e. up to
33%, depending on the amount of trac information they
can process) reductions in fuel consumption compared to
hybrid-electric vehicles.
Looking at the implications of vehicle automation for
air pollution, Grumert et al. (2015) reported a reduction
in NOx and Hydrocarbon (HC) emissions from the appli-
cation of a cooperative variable speed limit system that
uses infrastructure-to-vehicle communication to attach
individualized speed limits to each vehicle. Emissions
were found to decrease with higher penetration rates with
this system. Wang, Chen, Ouyang, and Li (2015)also
found that a higher penetration rate of intelligent vehi-
cles (i.e. vehicles equipped with their proposed longitu-
dinal controller) in a congested platoon was associated
with lower emissions of NOx. Moreover, Bose and Ioan-
nou (2001) found, through using simulation and eld
experiments, that emissions could be reduced from 1.5%
(NOx) to 60.6% (CO and CO2) during rapid acceleration
transients with the presence of 10% ACC equipped vehi-
cles. Choi and Bae (2013)comparedCO
2emissions for
lane changing of connected and manual vehicles. They
found that connected vehicles can emit up to 7.1% less
CO2through changing from a faster to a slower lane and
up to 11.8% less CO2through changing from a slower
to a faster lane. Environmental benets from the smooth
reaction of ACC vehicles in trac disturbances caused
by high-acceleration maneuvers, lane cut-ins, and lane
exiting were also conrmed by Ioannou and Stefanovic
(2005).
In a larger scale agent-based study, Fagnant and Kock-
elman (2014) simulated a scenario of a mid-sized city
where about 3.5% of the trips in day are served by shared
automated vehicles. These researchers reported that envi-
ronmental benets of shared automated vehicles could
be very important in all of the pollutant indicators exam-
ined (i.e. SO2, CO, NOx, Volatile organic compounds
[VOC] PM10,andGHG[Greenhousegas]).VOCand
CO showed the highest reductions, mainly because of
the signicantly less number of times a vehicle starts,
while the impact on Particulate matter with eective
diameter under 10 µm(PM
10), and GHG was relatively
small,mainlybecauseoftheadditionaltravelshared
vehicles have to undertake in order to access travelers or
to relocate. It should be noted that this simulation study
assumed that shared automated vehicle users would
not make more or longer trips and that the eet (both
automated and conventional vehicles) would not be elec-
tric, hybrid-electric or using alternative fuels. Finally, in
another study focusing on the long-term eects of auto-
matedvehicles,GreenblattandSaxena(2015)estimated
that autonomous taxis (i.e. battery electric shared auto-
mated vehicles) in 2030 could reduce GHG emissions per
vehicle per mile (a) by 87–94% compared to the emissions
of internal combustion conventional vehicles in 2014 and
(b) by 63–82% compared to the estimated emissions
for hybrid-electric vehicles in 2030. According to these
researchers, a signicant increase in travel demand for
autonomous taxis makes battery electric vehicle technol-
ogy more cost-ecient compared to internal combustion
or hybrid-electric vehicle technologies. Lower GHG
intensity of electricity and smaller vehicle sizes explain
the signicant reductions of GHG for (battery) electric
autonomous taxis. Furthermore, these researchers indi-
cated that autonomous taxis could oer almost 100%
reductioninoilconsumptionpermilecomparedto
conventional vehicles because oil provides less than 1% of
electricity generation in the US. Large energy savings of
up to 91% per automated vehicle in 2030 were also esti-
mated by Brown, Gonder, and Repac (2014)inascenario
that accounted for maximum impact of factors that could
lead to energy savings (e.g. ecient travel, lighter vehicles,
and electrication) and increased energy use (e.g. longer
travel distances and increased travel by underserved pop-
ulations such as youth, disabled, and elderly). However,
it remains uncertain which of these factors and to what
extent will they be realized in the future. Therefore, the
balance between energy savings and increased energy use
from automated vehicles could vary signicantly. Similar
uncertainty about the net eect of vehicle automation

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 337
on emissions and energy consumption was reported by
Wadud, MacKenzie, and Leiby (2016).
Safety
Assumptions
Over 90% of crashes are attributed to human driver
(National Highway Trac Safety Administration, 2008;
data for the US context). Typical reasons include, in
descending order, errors of recognition (e.g. inatten-
tion), decision (e.g. driving aggressively), performance
(e.g. improper directional control), and non-performance
(e.g. sleep). The advent of automated vehicles could
signicantly reduce trac accidents attributed to the
human driver by gradually removing the control from the
driver’s hands. This can be achieved through advanced
technologies applied to automated vehicles with respect
to perception of the environment and motion planning,
identication and avoidance of moving obstacles, longi-
tudinal, lateral and intersection control, and automatic
parking systems, for example. However, any unexpected
behavioral changes by a driver because of vehicle automa-
tion systems, human limitation in monitoring automation
or in taking control when necessary, along with possi-
ble cyberattacks, maliciously controlled vehicles and soft-
ware vulnerabilities might compromise the safety levels of
automated vehicles.
Literature results
A signicant amount of studies have proposed a wide
variety of advanced driver assistance systems that could
enhance trac safety levels. These systems include
collision avoidance (see e.g. Hayashi, Isogai, Raksin-
charoensak, & Nagai, 2012;Li,Juang,&Lin,2014;Shim,
Adireddy, & Yuan, 2012; Naranjo, Jiménez, Anaya, Talav-
era, & Gómez, 2016), lane keeping (see e.g. Lee, Choi, Yi,
Shin, & Ko, 2014) and lane change assistance (see e.g. Hou,
Edara, & Sun, 2015;Luo,Xiang,Cao,&Li,2016), longitu-
dinal speed assistance (see e.g. Martinez & Canudas-de-
Wit, 2007), and intersection assistance (see e.g. Liebner,
Klanner, Baumann, Ruhhammer, & Stiller, 2013). Several
other studies suggested that greater levels of safety could
besecuredbyadvancedlongitudinalorlateralmulti-
objective optimization controllers (see e.g. Gong, Shen,
&Du,2016;Khondaker&Kattan,2015; Wang, Hoogen-
doorn, Daamen, van Arem, & Happee, 2015), intersec-
tion controllers (see e.g. Dresner & Stone, 2008)and
path planning algorithms (see e.g. Ferguson, Howard, &
Likhachev, 2008;Kuwataetal.,2009)withspecicsafety
requirements.
Although advanced driver assistance systems can
reduce accident exposure and improve driver behavior
(see Spyropoulou, Penttinen, Karlaftis, Vaa, & Golias,
2008), adaptive behavior (i.e. the adoption of riskier
behavior because of over-reliance on the system) may
have adverse eects on trac safety (see Brookhuis et al.,
2001). For example, Hoedemaeker and Brookhuis (1998)
showed that the use of ACC may induce the adoption of
higher speed, smaller minimum time headway and larger
brake force. Rudin-Brown and Parker (2004)indicated
lower performance in brake light reaction time and
lane keeping for ACC users. Markvollrath, Schleicher,
and Gelau (2011) reported delayed reactions (i.e. speed
reduction) for ACC users when approaching curves or
entering fog, while Dixit, Chand, and Nair (2016)showed
that reaction times in taking control of the vehicle after
disengagement of the autonomous mode increases with
vehicle miles travelled. Xiong, Boyle, Moeckli, Dow, and
Brown (2012) showed that drivers’ adaptive behavior—
and therefore the safety implications of ACC—is related
to trust in automation, driving styles, understanding
of system operations and the driver’s personality. Fur-
thermore, safety levels might not substantially increase
(or even decrease) until high penetration rates of fully
automated vehicles are realized. For example, human
driving performance could degrade in level 3 of automa-
tionbecausepeoplehavelimitationswhenmonitoring
automation and taking over control when required (see
e.g. Strand, Nilsson, Karlsson, & Nilsson, 2014;Young
&Stanton,2007). Moreover, automated vehicles might
negatively inuence a driver’s behavior when using con-
ventional vehicles in mixed trac situations by making
themadoptunsafetimeheadways(contagioneect;see
Gouy, Wiedemann, Stevens, Brunett, & Reed, 2014).
Cyberattacks could also be an important threat for traf-
c safety. According to Petit and Shladover (2015), global
navigation satellite systems (GNSS) spoong and injec-
tion of fake messages into the communication between
vehicles are the two most likely and most severe attacks
for vehicle automation. Amoozadeh et al. (2015)simu-
lated message falsication and radio jamming attacks in
a CACC vehicle stream, inuencing the vehicles’ accel-
eration and space gap, respectively. These researchers
showed that security attacks could compromise trac
safety, causing stream instability and rear-end collisions.
Also,Gerdes,Winstead,andHeaslip(2013)showedthat
the energy expenditure of a platooning system could
increase by up to 300% through the attack of a malicious
vehicle, inuencing the motion (braking and accelera-
tion) of surrounding vehicles.
Social equity
Assumptions
Thesocialimpactsanddistributioneectsofatransport
system can be signicant. Vulnerable social groups, such

338 D. MILAKIS ET AL.
as the poorest people, children, younger, older, and dis-
abledpeoplecansuermorefromtheseimpacts,resulting
in their limited participation in society and potentially,
in social exclusion (Lucas & Jones, 2012). The intro-
duction of automated vehicles could have both positive
and negative implications for social equity. Automated
vehicles could oer the social groups that are currently
unable to own or drive a car (e.g. younger, older and dis-
abled people) the opportunity to overcome their current
accessibility limitations. For example, not only people
with physical and sensory (vision, hearing) disabilities,
but also younger and older people, could use automated
(shared) on demand services to reach their destinations.
However, the rst automated vehicles in the market are
likely to be quite expensive, thus limiting these benets
to only the wealthier members of these groups for certain
time. Safety benets might also be unevenly distributed
among dierent social groups. Owners of automated
vehicles will probably enjoy higher levels of travel safety
compared to drivers of conventional vehicles. Moreover,
potential spread of urban activities and possible reduc-
tion of public transport services (especially buses) might
further limit access to activities for poorer social groups.
On the other hand, potential conversion of redundant
road space to bicycle and pedestrian infrastructures
(especially infrastructures that connect with high capac-
ity public transport) could oer accessibility benets
to vulnerable population groups. Finally, the increase
of vehicle-sharing services and the subsequent possible
decrease of the requirements for construction of o-street
parking spaces could increase housing aordability.
Literature results
Eby et al. (2016) reported, in their review paper, a posi-
tive eect (i.e. avoiding crashes, enhancing easiness and
comfort of driving, increasing place, and temporal acces-
sibility) of many in-vehicle technologies (e.g. lane depar-
ture warning, forward collision warning/mitigation, blind
spot warning, parking assist systems, navigation assis-
tance,andACC)forolderdrivers.Suchimprovements
could allow older adults to drive for more years despite
declining of their functional abilities. Harper et al. (2016)
estimated the extent to which total travel demand could
increase in the US because of an increase in travel
demand by the non-driving, elderly, and people with
travel-restrictive medical conditions. They assumed that
in a fully automated vehicle context, people currently fac-
ing mobility restrictions would travel just as much as nor-
mal drivers within each age group and gender. They found
that the combined increase in travel demand from dier-
ent social groups could result in a 14% increase in annual
light-duty VMT for the US population. Finally, Mlade-
novic and McPherson (2016)analyzedtheopportuni-
ties arising from vehicle automation to incorporate social
justice in future trac control systems in terms of e-
ciency and equal access.
Economy
Assumptions
Automated vehicles could bring signicant economic
benets to individuals, society and businesses, but they
may also induce restructuring and possible losses in some
industries as well. The eects on GTC are distinguished
from other eects that are relevant for the economy. Look-
ing at the GTC eects, improved trac safety could pre-
vent accidents, thus avoiding the costs to society of acci-
dents, such as human capital losses, medical expenses,
lost productivity and quality of life, property damage,
insurance, and crash prevention costs. A reduction in
congestion delays would mean less travel costs for individ-
uals and reduced direct production costs for businesses.
Moreover, less congestion delays, along with increased
potential for performing other activities (e.g. working or
meeting) while on the move, could result in productivity
gains. Finally, an increase in shared automated vehicle ser-
vices would save individuals signicant (xed) costs asso-
ciated with car ownership without compromising their
mobility needs.
Other eects are now discussed. The reduction of o-
street parking requirements (ground oor level parking,
parking lots or multi-story parking garages) could allow
the development of more economically productive activ-
ities (e.g. residential, commercial or recreational). How-
ever, a possible massive reduction in car ownership levels
might have a critical negative impact on the automotive
industry. New business models in this industry are likely
to emerge, reecting the convergence of dierent tech-
nologies in automated vehicles, while car-related indus-
tries might experience losses (e.g. motor vehicle parts,
primary and fabricated metal, and plastics and rub-
ber products). Also, jobs in professional and technical
services, administration, wholesale and retail trade,
warehousing, nance and insurance, and management
of automotive companies could be negatively aected
by the reduction of turnover in the automotive industry.
Full vehicle automation could also directly lead to job
losses for various professions such as taxi, delivery, and
truck drivers. On the other hand, new jobs in hardware
and software technology for automated vehicles might
be generated. It is likely that such job related changes
will vary between countries and regions. Finally, overall
household expenditures can change because of auto-
mated vehicles (either increase or decrease). This could
subsequently inuence expenditures on other goods or
services (assuming constant saving rates). Such changes
in households’ expenditures could create or reduce jobs
in various sectors.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 339
Literature results
A rst systematic attempt to provide an order-of-
magnitude estimate about both the social and private
economic impacts of automated vehicles in the US con-
text was made by Fagnant and Kockelman (2015). Their
estimation took into account the safety, congestion,
parking, travel demand and vehicle ownership impacts
andwasbasedonseveralassumptionsaboutmarket
share,thenumberofautomatedvehicles,fuelsaving,
delay reduction, crash reduction, and VMT, among other
things. Their results showed that social benets per auto-
mated vehicle per year could reach $2960 (10% market
share) and increase up to $3900 (90% market share) if the
comprehensive costs of crashes, in the context of pain,
suering and the full value of a statistical life, are taken
into account. These estimations were based on the
assumption that crash and injury rates would be reduced
by 50% and 90% for 10% and 90% market penetration
rate of automated vehicles, respectively. The main rea-
son behind such signicant reductions in crash rates is
assumedtobethenear-eliminationofcrashescaused
byhumanerrorbecauseofthevehicleautomationtech-
nology. These researchers also showed that benets
for individuals are likely to be small, assuming current
technology costs at $100,000. Yet, an investment in this
technology when purchase price drops to $10,000 seems
to generate a positive return rate for many individu-
als, even with quite low values of time. Another study
examined the susceptibility of 702 occupations to tech-
nological developments (Frey & Osborne, 2017). This
study concluded that about 47% of total US employment
acrossallsectorsoftheeconomy,includingoccupa-
tions in the transportation and logistics sector (e.g. taxi,
ambulance, transit, delivery services, heavy truck drivers,
chaueurs, parking lot attendants, and trac techni-
cians) has a very high risk (probability of 0.7 or higher)
of being replaced by computer automation within next
two decades. This study assumed that not only routine,
but also non-routine cognitive and manual tasks would
be increasingly susceptible to automation because of
the expansion of computation capabilities (i.e. machine
learning and mobile robotics) and the decrease of the
market price of computing in the future. Yet, it was also
assumed that non-routine tasks involving perception and
manipulation,creative,andsocialintelligencewouldstill
be extremely dicult to automate in the near future.
Public health
Assumptions
Public health benets might result from reduced conges-
tion, lower trac noise, increased trac safety, and lower
emissions from automated vehicles. Literature has shown
a clear positive association between morbidity outcomes,
premature mortality rates, stress, and trac congestion
(see Hennessy & Wiesenthal, 1997; Levy, Buonocore, &
von Stackelberg, 2010;Miedema,2007). Furthermore, the
enhancement of road capacity, along with the reduction
of on-street parking demand, might allow conversion
of redundant road space into bicycle and pedestrian
infrastructures. Several studies have indicated that the
provision of such infrastructures is associated with higher
levelsofuseofactivemodes(Dill&Carr,2003;Buehler
&Pucher,2012) and subsequently with important public
health benets (e.g. obesity and diabetes; see Pucher,
Buehler, Bassett, & Dannenberg, 2010; Oja et al., 2011).
However, an increase in vehicle use because of automated
vehicles(eithermoreorlongervehicletrips)couldalso
have a negative impact on public health, since levels of
physical activity is likely to decrease.
Literature results
No systematic studies were found about the implications
of automated vehicles for public health.
Conclusions and directions for future research
So far, current literature has mainly explored the techno-
logical aspects of vehicle automation and its impacts on
driver and trac ow characteristics. However, interest
in the wider implications of automated vehicles is con-
stantly growing as this technology evolves. In this paper,
the eects of automated driving that are relevant to policy
and society were explored, literature results about these
eects were reviewed and areas for future research were
identied. This review is structured, based on the ripple
eect concept, which represents the implications of auto-
mated vehicles at three stages: rst-order (trac, travel
cost, and travel choices), second-order (vehicle ownership
and sharing, location choices and land use, and trans-
port infrastructure), and third-order (energy consump-
tion, air pollution, safety, social equity, economy, and
public health). General conclusions are presented below
and more specic ones for rst, second and third order
impacts, along with suggestions for the future research are
presented in subsequent sections.
Literature about the policy and society related implica-
tions of automated driving is rapidly evolving. Most stud-
ies in this review are dated after 2010. This does not mean
that research on development of automated vehicle sys-
tems and their implications has only been done in the last
7years.Bender(1991) oers a comprehensive overview
of the historic development of automated highway sys-
tems from the late 1950’s up to about 1990 (e.g. the Gen-
eral Motors’ systems). Moreover, several explorative or in-
depth modeling studies examined a wide arrange of the
impacts of automated highway systems several decades
earlier (see e.g. congestion, travel speed, vehicle hours

340 D. MILAKIS ET AL.
of delay, Benjamin, 1973;Miller,Bresnock,Shladover,&
Lechner, 1997;emissionrates,Barth,1997)orinitiated
discussions about the implications of these systems for
safety and driver convenience (Ward, 1994), infrastruc-
ture and urban form, (see Miller, 1997)andsocialequity
(see Stevens, 1997). These studies can oer valuable infor-
mation about the historical evolution of automated sys-
tems and the initial estimations of their impacts.
The majority of the studies in our literature review have
explored impacts on capacity, fuel eciency, and emis-
sions. Research on wider impacts and travel demand in
particular has started to pick up during the last 3 years.
The implications of automated vehicles for the economy,
public health, and social equity are still heavily under-
researched (see Table 2).
The policy and societal implications of automated
vehicles involve multiple complex dynamic interac-
tions. The magnitude of these implications is expected
to increase with the level of vehicle automation (espe-
cially for level 3 or higher), the level of cooperation
(vehicle-to-vehicle and vehicle-to-infrastructure), and
the penetration rate of vehicle automation systems. The
synergistic eects between vehicle automation, sharing,
and electrication can strengthen the potential impacts
of vehicle automation. Yet, the balance between the
short-term benets and the long-term impacts of vehicle
automation remains an open question.
Further research in a number of areas, as indicated
in following sections, could reduce this uncertainty. A
holistic evaluation of the costs and benets of automated
vehicles could help with urban and transport policies,
ensuring a smooth and sustainable integration of this new
transport technology into our transportation systems.
First-order implications of automated driving
Conclusions
First-order implications of automated vehicles comprised
travelcost,roadcapacity,andtravelchoices.Thexedcost
of automated vehicles is likely to reduce over time. GTC
will possibly be lower while both road capacity and travel
demand will probably increase in the short term.
Vehicle automation can result in travel time savings.
Simulations have explored this assumption on highways,
intersections and contexts involving shared automated
vehicles. Intersections appear to have more room for
travel time optimization compared to highways, while
a higher penetration rate of vehicle automation systems
seems to result in more travel time savings. Literature
results also suggest that vehicle automation systems
could result in lower fuel consumption and subsequently
reducedtravelcostintheshortterm.Researchonvarious
aspects of the third component of GTC (travel eort) is
rather limited. Moreover, there is still very little to read
in existing literature about the impact of vehicle automa-
tion on the values of time, leaving a striking gap in the
literature on this subject. Most, studies have focused on
incorporating comfort (in terms of acceleration and jerk)
as the optimizing metric in path-planning algorithms.
Yet, human comfort is inuenced by many other factors
(e.g. time headway), some of which remain unexplored
(e.g. motion sickness, apparent safety, and natural paths).
Therefore, there is no conclusion about the balance of
allcomfortrelatedeects.Also,studiesaboutvehicle
automation impacts on travel time reliability and on
utilization of travel time while on the move are scarce.
Researchresultsshowthatautomatedvehiclescould
have a clear positive impact on road capacity in the short
term.Themagnitudeofthisimpactisrelatedtothelevel
of automation, cooperation between vehicles and the
respective penetration rates. A 40% penetration rate of
CACC appears to be a critical threshold for realizing
signicant benets on capacity (>10%), while a 100%
penetration rate of CACC could theoretically double
capacity. Capacity impacts at level 3 or higher levels of
vehicle automation and more advanced levels of coop-
eration among vehicles but also between vehicles and
infrastructure (e.g. multi-platooning leaders, intersection
control systems, and variable speed limits) could well
exceed this theoretical threshold.
Most studies show that automated vehicles could
induce an increase of travel demand between 3% and 27%,
due to changes in destination choice (i.e. longer trips),
mode choice (i.e. modal shift from public transport and
walking to car), and mobility (i.e. more trips). Additional
increases in VMT are possible for shared automated vehi-
clesbecauseofemptyvehiclestravelingtothenextcus-
tomer or repositioning. However, one study (Childress
et al., 2015) indicated that if user costs per mile are very
high in a shared automated vehicle based transportation
system,VMTmayactuallybereduced.Thesamestudy
attained mixed, non-conclusive results about the trade-o
between increased travel demand, capacity increases and
congestion delays. No study took into account the poten-
tialchangesinlandusepatterns,whichmayalsoinuence
future travel demand.
Directions for future research
There is still a critical knowledge gap about the impact of
vehicle automation on individual components of travel
eort (i.e. comfort, travel time reliability and utilization
of travel time while on the move). For example, how
can factors such as motion sickness and perceived safety
aectthetravelcomfortofautomatedvehicles?Towhat
extent can vehicle automation systems reduce travel time
variability? How will people utilize available time in

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 341
automated vehicles? Also, what is the collective impact of
the dierent components of travel eort on values of time
for dierent socioeconomic groups and trip purposes?
Evidence about these individual factors—and subse-
quently GTC—can oer valuable input to a multitude
of other related areas of research, such as the impacts
on travel choices, accessibility and land uses, energy
consumption, and air pollution.
Additional research on travel demand impacts is criti-
cal as well. Possible travel demand changes will to a large
extentdeterminethemagnitudeofseveraloftheother
impacts of automated vehicles. Future studies should fur-
ther explore travel demand implications not only because
of changes in destination choice, mode choice and relo-
cation of (shared) automated vehicles but also because
of possible changes in land uses, parking demand and
latent demand from social groups with travel-restrictive
conditions.
Furthermore, although rst-order impacts of vehi-
cle automation on capacity are well-researched, potential
trade-os between additional capacity and GTC associ-
ated factors such as travel comfort, safety, and travel time
reliability remain relatively unexplored.
Second-order implications of automated driving
Conclusions
Second-order implications of automated vehicles com-
prised vehicle ownership and sharing, location choices,
land use and transport infrastructure. Literature results
suggest that shared automated vehicles could replace a sig-
nicantnumberofconventionalvehicles(fromabout67%
up to over 90%) delivering equal mobility levels. The over-
all reduction of the conventional vehicle eet could vary
according to the automated mode (vehicle-sharing, ride-
sharing, shared electric vehicle), the penetration rate of
shared automated vehicles and the presence or absence
of public transport. For example, a wide penetration of
shared automated vehicles supported by a high capacity
public transport system would be expected to result in
the highest reduction of conventional vehicle eet. Few
studies have explored the impact of automated vehicles on
location choices and land use. According to their results
automated vehicles could enhance accessibility citywide,
especially in remote rural areas, triggering further urban
expansion. Automated vehicles could also have a positive
impact on the density of economic activity at the center of
the cities. Parking demand for automated vehicles could
be shifted to peripheral zones, but could also remain high
in city centers, if empty cruising of shared automated vehi-
cles is not allowed. Moreover, several studies showed that
shared automated vehicles can signicantly reduce park-
ing space requirements up to over 90%. Finally, less wheel
wander and increased capacity because of automated
vehicles could accelerate pavement-rutting damage. Yet,
increase in speed of automated vehicles could compensate
for such negative eect by decreasing rut depth.
Directions for future research
A critical research priority is the exploration of the
implications that automated vehicles have for accessibil-
ity and, subsequently, for land uses. Results from this
kind of research will give some input into the assessment
of many other longer-term impacts of automated vehi-
cles, including energy consumption, air pollution, and
social equity. A comprehensive assessment of accessibil-
ity changes should focus on all components of accessibil-
ity (transportation, land use, individual, and temporal).
The impacts of automation on vehicle ownership could
be further explored. Thus far, research has discovered how
many shared automated vehicles can substitute conven-
tional vehicles to serve (part of ) current mobility demand.
Yet, a more important question is: what will the size of
vehicle eet reduction be if possible changes in travel
demand and the willingness of people to own or use
shared automated vehicles are taken into account?
Possible changes in urban streetscape and building
landscape because of vehicle automation also oer an area
for design research and experimentation. To what extent
will vehicle automation aect the level and geographical
distribution of parking demand? What will be the poten-
tial changes in the geometrical characteristics of roads and
intersections because of capacity enhancement, motion
stability of automated vehicles and automated intersec-
tion management? How will potential new urban space
be redistributed among dierent land uses (e.g. between
open space and new buildings) and users (e.g. vehicles,
cyclists, and pedestrians)?
Third-order implications of automated driving
Conclusions
Third-order implications of automated vehicles com-
prised energy consumption and air pollution, safety,
social equity, the economy, and public health. First-order
impacts on fuel eciency, emissions and accident risk
were also included in this section of our analysis for con-
sistency reasons. Literature results suggest that the use of
automated vehicles can result in fuel savings and lower
emissions in the short term. The net eect of vehicle
automation on energy consumption and GHG emissions
in the long term remains uncertain. Trac safety can
improveintheshorttermbutbehavioraladaptationand
low penetration rates of vehicle automation might com-
promise these benets. Few studies on the economic and
social equity impacts exist, while no systematic studies
were found for public health implications of automated
vehicles.

342 D. MILAKIS ET AL.
Various longitudinal, lateral and intersection control
algorithms and optimization systems can oer signicant
fuel savings and lower emissions of NOx, CO, and CO2.
Studies reviewed in this paper reported fuel savings up
to 31% for longitudinal and lateral movement controllers
andupto45%forintersectioncontrollers.Bothfuel
economy and emission reductions are reported higher
as the penetration rate of vehicle automation systems
increases. Furthermore, shared use of automated vehicles
is associated with reduced emissions (VOC and CO in
particular) because of the lower number of times a vehi-
cles starts. One study (Greenblatt & Saxena, 2015)associ-
ated the long-term impacts of battery electric shared auto-
mated vehicles with up to 94% less GHG and nearly 100%
less oil consumption per mile, compared to conventional
internal combustion vehicles. Yet, several factors could
lead to increased energy use (e.g. longer travel distances
andincreasedtravelbyunderservedpopulationssuchas
youth, disabled, and elderly). Thus, the net eect of vehi-
cle automation on energy consumption and GHG emis-
sions remains uncertain.
As for trac safety, literature results suggest that
advanced driver assistance systems can reduce expo-
sure to accidents. Level 3 or higher levels of automa-
tion can further enhance trac safety. However, as
long as the human driver remains in-the-loop, behav-
ioral adaptation—namely the adoption of riskier behav-
ior because of over-reliance on the system—can compro-
mise safety benets. Moreover, fully automated vehicles
might not deliver high safety benets until high penetra-
tion rates of these vehicles are realized. Cyberattacks, such
as message falsication and radio jamming, can compro-
mise trac safety as well.
Finally, research on the impacts of vehicle automa-
tion on the economy, social equity and public health
is almost non-existent. Automated vehicles could have
signicant impacts on all three areas. Results from one
study (Fagnant & Kockelman, 2015)indicatethatsocial
benets per automated vehicle per year could reach $3900
where there is a 90% market share of automated vehicles,
while a positive return rate for individuals should not be
expected before the additional cost for vehicle automa-
tion drops to $10,000. Another study (Frey & Osborne,
2017) concluded that occupations in the transportation
and logistics sectors (e.g. taxi, ambulance, transit, deliv-
ery services, heavy truck drivers, chaueurs, parking lot
attendants, and trac technicians) have a high probability
(>0.7) of being replaced by computer automation within
the next two decades. In-vehicle technologies can have
positive eects (i.e. avoiding crashes, enhancing easiness
and comfort of driving, increasing place, and tempo-
ral accessibility) for elderly. Such improvements could
extend driving life expectancy for older adults. One study
estimated that automated vehicles could induce up to 14%
additional travel demand from the non-driving, elderly,
andpeoplewithtravel-restrictivemedicalconditions.
Directions for future research
The emission and fuel eciency eects of vehicle automa-
tion are well researched. However, the magnitude of the
eect at dierent levels of automation and penetration
rates could be further tested. A clear research priority
is the exploration of the long-term eects of automated
vehicles on energy consumption and emissions, taking
into account potential travel demand changes but also the
additional synergistic eects between vehicle automation,
sharing, and electrication and possible changes in vehi-
clesize.Resultsfromthiskindofresearchwillallowus
to better assess the balance between the short-term ben-
ets and the long-term impacts of automated vehicles on
energy consumption and emissions.
Another critical research priority concerns safety
implications in the transitional contexts of fully auto-
mated and conventional vehicles. To what extent will
vehicleautomationandhumandriversofconventional
vehicles compromise the performance of each other in
mixed trac situations? A better understanding of the
types of cyberattacks and their potential impacts on
trac safety is critical too.
A comprehensive assessment of economic, public
health and social impacts is also missing from current lit-
erature. For example, what could be the scale of job losses
(or gains) due to full vehicle automation? Which sec-
tors and which countries and/or regions would be most
aected?Andwhatcouldbethestrategiestomitigate
the economic impacts of expected job losses? The impact
of vehicle automation on public health is also an impor-
tant area for further research. To what extent will vehi-
cle automation induce lower levels of physical activity and
what will the possible impacts be on activity-related pub-
lic health issues, such as obesity and diabetes? The explo-
ration of social impacts and distribution eects through
the analysis of potential accessibility changes would also
contribute to a better understanding of the social implica-
tionsofautomatedvehicles.Towhatextentcould(shared)
automated vehicles inuence the ability of vulnerable
social groups (e.g. people with physical, sensory and men-
taldisabilities,youngerorolderpeople,andsinglepar-
ents) to access economic and social opportunities? How
benets stemming from vehicle automation will be dis-
tributed among dierent social groups?
Methodological challenges in exploring the
implications of automated vehicles
To further explore the implications of automated vehicles,
we will have to face several methodological challenges.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 343
One critical issue is that this technology (especially
at level 3 or higher levels of automation) is still in its
infancy. Thus, no adequate empirical data about the use
of automated vehicles exist yet. Therefore, studies have
mainly made use of micro and macro trac simula-
tion, driving simulators, eld experiments and analytical
methods to explore rst-order implications of automated
vehicles on travel time, capacity, fuel eciency, emis-
sions, and safety. More empirical studies about rst-order
implications of vehicle automation are a clear priority
as this technology evolves. For second and third-order
implications, the armory of methods needs to expand
to capture the behavioral aspects, underlying potential
changes due to vehicle automation. Thus, for example,
qualitative methods, such as focus groups or in-depth
interviews, in combination with quantitative methods,
like stated choice experiments, could be used for explor-
ing questions about the impacts of vehicle automation
on travel comfort, utilization of travel time while on the
move, value of time, travel, and location choices. Yet,
people may have diculties in envisioning automated
vehicles, so stated choice experiments could suer from
hypothetical bias (see Fifer, Rose, & Greaves, 2014). More
creative techniques such as virtual reality or serious gam-
ing would be useful in behavioral experiments about the
impacts of automated vehicles. Another approach may
be to investigate similar systems that are essentially auto-
mated. For example, investigate the value of time for train
commuters who both live and work near stations, as for
them a train trip is essentially automated. Travel behavior
changes because of ICT (e.g. telecommuting) could oer
insights into possible behavioral changes because of
vehicle automation. Expert opinion research (e.g. Delphi
technique) could also be an alternative method.
Agent-based and activity-based models could then be
used to simulate possible changes in travel demand, vehi-
cle ownership and other environmental indicators, such
as energy consumption and emissions. The connection
of travel models with land use models (in so-called Land
Use—Transport Interaction, or LUTI models) would also
allow potential long-term land use impacts on travel
demand to be captured. Alternative approaches could
involve empirical models for the analysis of comparable
systems and their potential impacts on land use (e.g. valet
parking, car-free neighborhoods, and high speed train).
Finally, accessibility metrics and measures of inequality
couldbeusedintheanalysisofthesocialequityimpacts
of automated vehicles.
Acknowledgments
The authors would like to thank three anonymous reviewers for
their constructive comments on an earlier draft of this paper.
Funding
This research was supported by the Delft Infrastructures and
Mobility Initiative (grant no. C61U01).
ORCID
Dimitris Milakis http://orcid.org/0000-0001-5220-4206
References
Ala, M. V., Yang, H., & Rakha, H. (2016). Modeling eval-
uation of Eco—Cooperative adaptive cruise control in
vicinity of signalized intersections. Transportation Research
Record: Journal of the Transportation Research Board,2559,
108–119.
Amoozadeh,M.,Raghuramu,A.,Chuah,C.,Ghosal,D.,Zhang,
H. M., Rowe, J., & Levitt, K. (2015). Security vulnerabili-
ties of connected vehicle streams and their impact on coop-
erative driving. IEEE Communications Magazine,52(6),
126–132.
Arnaout, G., & Arnaout, J.-P. (2014). Exploring the eects of
cooperative adaptive cruise control on highway trac ow
using microscopic trac simulation. Transportation Plan-
ning and Technology,37(2), 186–199.
Arnaout, G., & Bowling, S. (2011). Towards reducing trac con-
gestion using cooperative adaptive cruise control on a free-
way with a ramp. Journal of Industrial Engineering and Man-
agement,4(4), 699–717.
Asadi, B., & Vahidi, A. (2011). Predictive cruise control: Utiliz-
ing upcoming trac signal information for improving fuel
economy and reducing trip time. IEEE Transactions on Con-
trol Systems Technology,19(3), 707–714.
Barsade, S. (2002). The ripple eect: Emotional contagion
and its inuence on group behavior. Administrative Science
Quarterly,47, 644–675.
Barth, M. J. (1997). The eects of AHS on the environment. In
P. A . I o a n n o u ( E d . ) , Automated highway systems (pp. 265–
291). New York: Plenum Press.
Bellem, H., Schönenberg, T., Krems, J. F., & Schrauf, M. (2016).
Objective metrics of comfort: Developing a driving style for
highly automated vehicles. Transportation Research Part F:
Psychology and Behaviour,41, 45–54.
Bender, J. G. (1991). An overview of systems studies of auto-
mated highway systems. IEEE Transactions on Vehicular
Tech n o l o g y ,40(1), 82–99.
Benjamin, P. (1973). Analysis of urban dual mode trans-
portation. In Personal Rapid Transit II: Progress, problems,
potential (pp. 95–106). Minneapolis, MN: University of
Minnesota.
Black, S. (2001). Computing ripple eect for software main-
tenance. Journal of Software Maintenance and Evolution:
Research and Practice,13(4), 263–279.
Boesch, P. M., Ciari, F., & Axhausen, K. W. (2016). Autonomous
vehicle eet sizes required to serve dierent levels of
demand. Transportation Research Record: Journal of the
Transportation Research Board,2542, 111–119.
Bose, A., & Ioannou, P. (2001). Evaluation of the environmental
eects of intelligent cruise control vehicles. Transportation
Research Record,1774, 90–97.

344 D. MILAKIS ET AL.
Brookhuis, K., de Waard, D., & Janssen, W. (2001). Behavioural
impacts of advanced driver assistance systems—an
overview. European Journal of Transport and Infrastructure
Research,1(3), 245–253.
Brown, A., Gonder, J., & Repac, B. (2014). An analysis of pos-
sible energy impacts of automated vehicle. In G. Meyer &
S. Beiker (Eds.), Road vehicle automation (pp. 137–153).
Berlin: Springer.
Buehler, R., & Pucher, J. (2012). Cycling to work in 90 large
American cities: new evidence on the role of bike paths and
lanes. Transportation,39(2), 409–432.
Carbaugh, J., Godbole, D. N., & Sengupta, R. (1998). Safety and
capacity analysis of automated and manual highway sys-
tems. Transportation Research Part C: Emerging Technolo-
gies,6(1–2), 69–99.
Chen, F., Balieu, R., & Kringos, N. (2016). Potential inu-
ences on Long-Term service performance of road infras-
tructure by automated vehicles. Transportation Research
Record: Journal of the Transportation Research Board,2550,
72–79.
Chen, Y., Bell, M. G. H., & Bogenberger, K. (2010). Risk-Averse
autonomous route guidance by a constrained A∗search.
Journal of Intelligent Transportation Systems: Technology,
Planning, and Operations,14(3), 188–196.
Chen, T. D., Kockelman, K. M., & Hanna, J. P. (2016). Opera-
tions of a shared, autonomous, electric vehicle  eet: Implica-
tions of vehicle & charging infrastructure decisions. Trans-
portation Research Part A,94, 243–254.
Childress, S., Nichols, B., & Coe, S. (2015). Using an activity-
based model to explore possilbe impacts of automated vehi-
cles. Transportation Research Record: Journal of the Trans-
portation Research Board, 2493, 99–106.
Choi, J.-E., & Bae, S.-H. (2013). Development of a methodol-
ogy to demonstrate the environmental impact of connected
vehicles under lane-changing conditions. Simulation,89(8),
964–976.
Clement, S. J., Taylor, M. A. P., & Yue, W. L. (2004). Simple pla-
toon advancement: A model of automated vehicle move-
ment at signalised intersections. Transportation Research
Part C: Emerging Technologies,12(3–4), 293–320.
Cooper, C., Orford, S., Webster, C., & Jones, C. B. (2013).
Exploring the ripple eect and spatial volatility in house
prices in England and Wales: regressing interaction domain
cross-correlations against reactive statistics. Environment
and Planning B: Planning and Design,40(5), 763–782.
Correia, G. H., de, A., & van Arem, B. (2016). Solving the User
Optimum Privately Owned Automated Vehicles Assign-
ment Problem (UO-POAVAP): A model to explore the
impacts of self-driving vehicles on urban mobility. Trans-
portation Research Part B: Methodological,87, 64–88.
Cyganski, R., Fraedrich, E., & Lenz, B. (2015). Travel time valua-
tion for automated driving: a use-case-driven study. In 94th
annual meeting of the transportation research board.Wash-
ington, DC: Transportation Research Board.
Dang, R., Wang, J., Li, S. E., & Li, K. (2015). Coordinated
adaptive cruise control system with Lane-Change assis-
tance. IEEE Transactions on Intelligent Transportation Sys-
tems,16(5), 2373–2383.
de Winter, J., Happee, R., Martens, M. H., & Stanton, N. A.
(2014). Eects of adaptive cruise control and highly auto-
mated driving on workload and situation awareness: A
review of the empirical evidence. Transportation Research
Part F: Trac Psychology and Behaviour,27, 196–217.
Delis, A. I., Nikolos, I. K., & Papageorgiou, M. (2015). Macro-
scopic trac ow modeling with adaptive cruise con-
trol: Development and numerical solution. Computers and
Mathematics with Applications, 70(8), 1921–1947.
Diakaki, C., Papageorgiou, M., Papamichail, I., & Nikolos, I.
(2015). Overview and analysis of vehicle automation and
communication systems from a motorway trac manage-
ment perspective. Transportation Research Part A: Policy
and Practice,75, 147–165.
Diels, C., & Bos, J. E. (2016). Self-driving carsickness. 53, 374–
382.
Dill, J., & Carr, T. (2003). Bicycle commuting and facilities in
majorUScitiesifyoubuildthem,commuterswillusethem.
Transportation Research Record: Journal of the Transporta-
tion Research Board,1828(1), 116–123.
Dixit, V. V, Chand, S., & Nair, D. J. (2016). Autonomous vehi-
cles: Disengagements, accidents and reaction times. PLoS
ONE,11(12), e0168054.
Dresner, K., & Stone, P. (2008). A multiagent approach to
autonomous intersection management. Journal of Articial
Intelligence Research,31, 591–656.
Eben Li, S., Li, K., & Wang, J. (2013). Economy-oriented
vehicle adaptive cruise control with coordinating multi-
ple objectives function. Veh i cle Sy s t e m D y n ami c s ,51(1), 1–
17.
Eby, D. W., Molnar, L. J., Zhang, L., Louis, R. M. S., Zanier, N.,
Kostyniuk, L. P., & Stanciu, S. (2016). Use, perceptions, and
benetsofautomotivetechnologiesamongagingdrivers.
Injury Epidemiology, 3, 1–20.
Elbanhawi, M., Simic, M., & Jazar, R. (2015). In the passenger
seat: Investigating ride comfort measures in autonomous
cars. IEEE Intelligent Transportation Systems Magazine,Fall,
4–17.
Fagnant, D. J., & Kockelman, K. M. (2014). The travel and
environmental implications of shared autonomous vehicles,
using agent-based model scenarios. Transportation Research
Part C: Emerging Technologies,40, 1–13.
Fagnant, D. J., & Kockelman, K. M. (2015). Preparing a nation
for autonomous vehicles: Opportunities, barriers and policy
recommendations for capitalizing on Self-Driven vehicles.
Transportation Research Part A: Policy and Practice,77,1–
20.
Fagnant, J. D., & Kockelman, K. M. (2016). Dynamic ride-
sharing and eet sizing for a system of shared autonomous
vehicles in Austin, Texas. Transportation. Advanced online
publication. https://doi.org/10.1007/s11116-016-9729-z
Fajardo, D., Au, T. C., Waller, T., Stone, P., & Yang, D. (2012).
Automated intersection control: Performance of a future
innovation versus current trac signal control. Trans por t a-
tion Research Record: Journal of the Transportation Research
Board,2259, 223–232.
Ferguson, D., Howard, T. M., & Likhachev, M. (2008). Motion
planning in urban environments. Journal of Field Robotics,
25(11–12), 939–960.
Fernandes, P., Nunes, U., & Member, S. (2015). Multiplatoon-
ing leaders positioning and cooperative behaviora lgorithms
of communicant automated vehicles for high trac capac-
ity. IEEE Transactions on Intelligent Transportation Systems,
16(3), 1172–1187.
Fifer, S., Rose, J., & Greaves, S. (2014). Hypothetical bias in
Stated Choice Experiments: Is it a problem? And if so, how
do we deal with it?. Transportation Research Part A: Policy
and Practice,61, 164–177.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 345
Frandsen, T., & Nicolaisen, J. (2013). The ripple eect: Citation
chain reactions of a nobel prize. Journal of the American
Society for Information Science and Technology,64(3), 437–
447.
Frey, B. C., & Osborne, M. A. (2017). The Future of Employ-
ment: How Susceptible Are Jobs To Computerisation. Tech-
nologicalForecasting&SocialChange,114, 254–280.
Gerdes, R. M., Winstead, C., & Heaslip, K. (2013). CPS: An
Eciency-motivated attack against autonomous vehicular
transportation. Proceedings of the 29th Annual Computer
Security Applications Conference (ACSAC), 99–108.
Gerónimo, D., López, A. M., Sappa, A. D., & Graf, T. (2010). Sur-
vey of Pedestrian detection for advanced driver assistance
systems. IEEE Transactions on Pattern Analysis and Machine
Intelligence,32(7), 1239–1258.
Geurs, K. T., & van Wee, B. (2004). Accessibility evaluation
of land-use and transport strategies: Review and research
directions. Journal of Transport Geography,12(2), 127–140.
Glaser, S., Vanholme, B., Mammar, S., Gruyer, D., & Nouvelière,
L. (2010). Maneuver-based trajectory planning for highly
autonomous vehicles on real road with trac and driver
interaction. IEEE Transactions on Intelligent Transportation
Systems,11(3), 589–606.
Gong, S., Shen, J., & Du, L. (2016). Constrained optimization
and distributed computation based car following control of
a connected and autonomous vehicle platoon. Tran spo r ta-
tion Research Part B: Methodological,94, 314–334.
González, D., Pérez, J., Milanés, V., & Nashashibi, F. (2016). A
review of motion planning techniques for automated vehi-
cles. IEEE Transactions on Intelligent Transportation Sys-
tems,17(4), 1135–1145.
Gouy, M., Wiedemann, K., Stevens, A., Brunett, G., & Reed, N.
(2014). Driving next to automated vehicle platoons: How do
short time headways inuence non-platoon drivers’ longi-
tudinal control?. Transportation Research Part F: Trac Psy-
chology and Behaviour,27, 264–273.
Greenblatt, J. B., & Saxena, S. (2015). Autonomous taxis
could greatly reduce greenhouse-gas emissions of US
light-duty vehicles. Nature Climate Change,5(September),
860–863.
Grumert, E., Ma, X., & Tapani, A. (2015). Eects of a coopera-
tive variable speed limit system on trac performance and
exhaust emissions. Transportation Research Part C: Emerg-
ing Technologies,52, 173–186.
Gucwa, M. (2014). Mobility and energy impacts of automated
cars (PhD Thesis). Stanford, CA: Department of Manage-
ment Science and Engineering, Stanford University.
Hall, R. W., Nowroozi, A., & Tsao, J. (2001). Entrance capac-
ity of an automated highway system. Transportation Science,
35(1), 19–36.
Harper, C., Hendrickson, C. T., Mangones, S., & Samaras,
C. (2016). Estimating potential increases in travel with
autonomous vehicles for the non-driving, elderly and peo-
ple with travel-restrictive medical conditions. Tra nsp orta -
tion Research Part C: Emerging Technologies,72, 1–9.
Hayashi, R., Isogai, J., Raksincharoensak, P., & Nagai, M. (2012).
Autonomous collision avoidance system by combined con-
trol of steering and braking using geometrically optimised
vehicular trajectory. Vehicl e S y stem Dynam i c s,50, 151–168.
Hennessy, D. A., & Wiesenthal, D. L. (1997). The relationship
between trac congestion, driver stress and direct versus
indirect coping behaviours. Ergonomics,40(3), 348–361.
Hoedemaeker, M., & Brookhuis, K. A. (1998). Behavioural
adaptation to driving with an adaptive c ruise control (ACC).
Transportation Research Part F: Trac Psychology and
Behaviour,1(2), 95–106.
Hoogendoorn, R., van Arem, B., & Hoogendoorn, S. (2014).
Automated driving, trac ow eciency and human fac-
tors: A literature review. Transportation Research Record,
2422, 113–120.
Hou, Y., Edara, P., & Sun, C. (2015). Situation assessment and
decision making for lane change assistance using ensemble
learning methods. ExpertSystemswithApplications,42(8),
3875–3882.
Hounsell, N. B., Shrestha, B. P., Piao, J., & McDonald, M. (2009).
Review of urban trac management and the impacts of new
vehicle technologies. IET Intelligent Transport Systems,3(4),
419–428.
Huang, S., Ren, W., & Chan, S. C. (2000). Design and per-
formance evaluation of mixed manual and automated-
control trac. IEEE Transactions on Systems, Man, and
Cybernetics—Part A: Systems and Humans,30(6), 661–673.
Ilgin Guler, S., Menendez, M., & Meier, L. (2014). Using con-
nected vehicle technology to improve the eciency of inter-
sections. Transportation Research Part C: Emerging Tech-
nologies,46, 121–131.
International Transport Forum. (2015). Urban mobility: System
upgrade.Paris:OECD/InternationalTransportForum.
Ioannou, P., & Stefanovic, M. (2005). Evaluation of the ACC
vehicles in mixed trac: Lane change eects and sensitivity
analysis petros ioannou, margareta stefanovic. IEEE Trans-
actions on Intelligent Transportation Systems,6(1), 79–89.
Ivanov, D., Sokolov, B., & Dolgui, A. (2014). The Ripple eect in
supply chains: trade-o “eciency-exibility-resilience” in
disruption management. International Journal of Production
Research,52(7), 2154–2172.
Kamal, A. S., Imura, J., Hayakawa, T., Ohata, A., & Aihara,
K. (2015). A Vehicle-Intersection coordination scheme for
smooth ows of trac without using trac lights. IEEE
Transactions on Intelligent Transportation Systems,16(3),
1136–1147.
Kamal, A. S., Taguchi, S., & Yoshimura, T. (2016). Ecient driv-
ing on multilane roads under a connected vehicle environ-
ment. IEEE Transactions on Intelligent Transportation Sys-
tems,17(9), 2541–2551.
Kamalanathsharma, R. K., & Rakha, H. A. (2016). Leverag-
ing connected vehicle technolog y and telematics to enhance
vehicle fuel eciency in the vicinity of signalized intersec-
tions. JournalofIntelligentTransportationSystems,20, 33–
44.
Kesting, A., Treiber, M., Schönhof, M., & Helbing, D. (2008).
Adaptive cruise control design for active congestion avoid-
ance. Transportation Research Part C: Emerging Technolo-
gies,16(6), 668–683.
Khondaker, B., & Kattan, L. (2015). Variable speed limit: A
microscopic analysis in a connected vehicle environment.
Transportation Research Part C: Emerging Technologies,58,
146–159.
Kuwata, Y., Teo, J., Fiore, G., Karaman, S., Frazzoli, E., & How,
J. P. (2009). Real-time motion planning with applications to
autonomous urban driving. IEEE Transactions on Control
Systems Technology,17(5), 1105–1118.
Lamondia,J.J.,Fagnant,D.J.,Qu,H.,Barrett,J.,&Kockel-
man, K. (2016). Shifts in Long-Distance travel mode due to

346 D. MILAKIS ET AL.
automated vehicles and travel survey analysis. Tran spo r ta-
tion Research Record: Journal of the Transportation Research
Board,2566, 1–10.
Le Vine, S., Zolfaghari, A., & Polak, J. (2015). Autonomous cars:
The tension between occupant experience and intersection
capacity. Transportation Research Part C: Emerging Tech-
nologies,52, 1–14.
Lee, J., Choi, J., Yi, K., Shin, M., & Ko, B. (2014). Lane-keeping
assistance control algorithm using dierential braking to
prevent unintended lane departures. Control Engineering
Practice,23(1), 1–13.
Lefèvre, S., Carvalho, A., & Borrelli, F. (2016). A Learning-
Based framework for velocity control in autonomous driv-
ing. IEEE Transactions on Automation Science and Engineer-
ing,13(1), 32–42.
Levin, M. W., & B oyles, S. D. (2015). Eects of autonomous vehi-
cle ownership on trip, mode, and route choice. Transp orta -
tion Research Record: Journal of the Transportation Research
Board,2493, 29–38.
Levin, M. W., Boyles, S. D., & Patel, R. (2016). Paradoxes of
reservation-based intersection controls in trac networks.
Transportation Research Part A,90, 14–25.
Levin, M. W., Fritz, H., & Boyles, S. D. (2016). On optimizing
Reservation-Based intersection controls. IEEE Transactions
on Intelligent Transportation Systems.Advanceonlinepub-
lication. https://doi.org/10.1109/TITS.2016.2574948
Levy, J. I., Buonocore, J. J., & von Stackelberg, K. (2010). Eval-
uation of the public health impacts of trac congestion:
a health risk assessment. Environmental Health: A Global
Access Science Source,9, 65.
Lewis-Evans, B., De Waard, D., & Brookhuis, K. A. (2010). That’s
close enough-A threshold eect of time headway on the
experience of risk, task diculty, eort, and comfort. Acci-
dent Analysis and Prevention,42(6), 1926–1933.
Li,Z.,Chitturi,M.V,Zheng,D.,Bill,A.R.,&Noyce,D.A.
(2013). Modeling Reservation-Based autonomous intersec-
tion control in VISSIM. Transportation Research Record:
Journal of the Transportation Research Board,2381, 81–90.
Li, K.-R., Juang, J.-C., & Lin, C.-F. (2014). Active collision avoid-
ance system for steering control of autonomous vehicles.
IET Intelligent Transport Systems,8(6), 550–557.
Li, S., Li, K., Rajamani, R., & Wang, J. (2011). Model predic-
tive multi-objective vehicular adaptive cruise control. IEEE
Transa ctions on C ont rol Sys tem s Technol ogy,19(3), 556–566.
Li, S. E., Peng, H., Li, K., & Wang, J. (2012). Minimum fuel
control strategy in automated car-following scenarios. IEEE
Transactions on Vehicular Technology,61(3), 998–1007.
Liebner, M., Klanner, F., Baumann, M., Ruhhammer, C., &
Stiller, C. (2013). Velocity-based driver intent inference at
urban intersections in the presence of preceding vehicles.
IEEE Intelligent Transportation Systems Magazine,5(2), 10–
21.
Lucas, K., & Jones, P. (2012). Social impacts and equity issues in
transport: An introduction. Journal of Transport Geography,
21, 1–3.
Luo, Y., Chen, T., Zhang, S., & Li, K. (2015). Intelligent hybrid
electric vehicle ACC with coordinated control of tracking
ability, fuel economy, and ride comfort. IEEE Transactions
on Intelligent Transportation Systems,16(4), 2303–2308.
Luo, L., Liu, H., Li, P., & Wang, H. (2010). Model predictive
control for adaptive cruise control with multi-objectives:
comfort, fuel-economy, safety and car-following. Journal of
Zhejiang University SCIENCE A,11(3), 191–201.
Luo, Y., Xiang, Y., Cao, K., & Li, K. (2016). A dynamic auto-
mated lane change maneuver based on vehicle-to-vehicle
communication. Transportation Research Part C: Emerging
Tech n o l o g i es,62, 87–102.
Malokin, A., Circella, G., & Mokhtarian, P. L. (2015). How
Do activities conducted while commuting inuence mode
choice? Testing public transportation advantage and
autonomous vehicle scenarios. In 94th Annual Meeting of
the Transportation Research Board.
Manzie, C., Watson, H., & Halgamuge, S. (2007). Fuel economy
improvements for urban driving: Hybrid vs. intelligent vehi-
cles. Transportation Research Part C: Emerging Technologies,
15(1), 1–16.
Markvollrath, Schleicher, S., & Gelau, C. (2011). The inu-
ence of cruise control and adaptive cruise control on driv-
ing behaviour—A driving simulator study. Accident Analy-
sis and Prevention,43(3), 1134–1139.
Martinez, J.-J., & Canudas-de-Wit, C. (2007). A safe longitu-
dinal control for adaptive cruise control and Stop-and-Go
scenarios. IEEE Transactions on Control Systems Technology,
15(2), 246–258.
Meen, G. (1999). Regional house prices and the ripple eect: A
new interpretation. Housing Studies,14(6), 733–753.
Michael, J. B., Godbole, D. N., Lygeros, J., & Sengupta, R. (1998).
Capacity analysis of trac ow over a Single-Lane auto-
mated highway system. ITS Journal—Intelligent Transporta-
tion Systems Journal,4(1–2), 49–80.
Miedema, H. M. E. (2007). Adverse eects of trac noise.
In T. Garling & L. Steg (Eds.), Threats from car trac to
the quality of urban life. Problems, causes, and solutions
(pp. 53–77). Oxford, UK: Elsevier.
Milakis, D. (2015). Will Greeks Cycle? Exploring intention and
attitudes in the case of the new bicycle network of Patras.
International Journal of Sustainable Transportation,9(5),
321–334.
Milakis, D., Snelder, M., van Arem, B., van Wee, B., & Cor-
reia, G. (2017). Development and transport implications of
automated vehicles in the Netherlands: scenarios for 2030
and 2050. European Journal of Transport and Infrastructure
Research,17(1), 63–85.
Milakis, D., van Arem, B., & van Wee, B. (2015). The ripple
eect of automated driving. In 2015 BIVEC-GIBET Trans-
port Research Day. Eindhoven, The Netherlands: BIVEC-
GIBET.
Miller, M. A. (1997). Institutional and societal issues associated
with atuomated highway systems. An environmental per-
spective.InP.A.Ioannou(Ed.),Automated highway systems
(pp. 313–324). New York: Plenum Press.
Miller, M. A., Bresnock, A., Shladover, S., & Lechner, E. H.
(1997). Regional mobility impacts assessment of highway
automation.InP.A.Ioannou(Ed.),Automated highway sys-
tems (pp. 293–311). New York: Plenum Press.
Mladenovic, M. N., & McPherson, T. (2016). Engineering social
justice into trac control for Self-Driving vehicles?. Science
and Engineering Ethics, 22, 1131–1149.
Monteil, J., Nantes, A., Billot, R., Sau, J., & El Faouzi, N. (2014).
Microscopic cooperative trac ow: calibration and simu-
lation based on a next generation simulation dataset. IET
Intelligent Transport Systems,8(6), 519–525.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS 347
Moon, S., Moon, I., & Yi, K. (2009). Design, tuning, and evalua-
tion of a full-range adaptive cruise control system with col-
lision avoidance. Control Engineering Practice,17(4), 442–
455.
Naranjo, J. E., Jiménez, F., Anaya, J. J., Talavera, E., &
Gómez, O. (2016). Application of vehicle to another entity
(V2X) communications for motorcycle crash avoidance.
Journal of Intelligent Transportation Systems: Technology,
Planning, and Operations. Advance online publication.
https://doi.org/10.1080/15472450.2016.1247703
National Highway Trac Safety Administration. (2008).
National motor vehicle crash causation survey. Report DOT
HS 811 059.Washington,DC:NHTSA.
Ngoduy, D. (2012). Application of gas-kinetic theory to mod-
elling mixed trac of manual and ACC vehicles. Tra nsp ort-
metrica,8(1), 43–60.
Ngoduy, D. (2013). Instability of cooperative adaptive cruise
control trac ow: A macroscopic approach. Communica-
tions in Nonlinear Science and Numerical Simulation,18(10),
2838–2851.
Oja,P.,Titze,S.,Bauman,A.,deGeus,B.,Krenn,P.,Reger-Nash,
B., & Kohlberger, T. (2011). Health benets of cycling: A sys-
tematic review. Scandinavian Journal of Medicine and Sci-
ence in Sports,21(4), 496–509.
Petit, J., & Shladover, S. E. (2015). Potential cyberattacks on
automated vehicles. IEEE Transactions on Intelligent Trans-
portation Systems,16(2), 546–556.
Piao, J., & McDonald, M. (2008). Advanced driver assistance
systems from autonomous to cooperative approach. 28(5),
659–684.
Pucher, J., Buehler, R., Bassett, D. R., & Dannenberg, A. L.
(2010). Walking and cycling to health: A comparative anal-
ysis of city, state, and international data. American Journal
of Public Health,100(10), 1986–1992.
Raimondi, F. M., & Melluso, M. (2008). Fuzzy motion control
strategy for cooperation of multiple automated vehicles with
passengers comfort. Automatica,44(11), 2804–2816.
Rajamani, R., & Shladover, S. E. (2001). Experimental compara-
tive study of autonomous and co-operative vehicle-follower
control systems. Transportation Research Part C: Emerging
Tech n o l o g i es,9(1), 15–31.
Rios-torres, J., & Malikopoulos, A. A. (2016). Auto-
mated and cooperative vehicle merging at high-
way On-Ramps. IEEE Transactions on Intelligent
Transportation Systems. Advance online publication.
https://doi.org/10.1109/TITS.2016.2587582
Rudin-Brown, C. M., & Parker, H. A. (2004). Behavioural adap-
tation to adaptive cruise control (ACC): Implications for
preventive strategies. Transportation Research Part F: Trac
Psychology and Behaviour,7(2), 59–76.
SAE International. (2016). Taxonomy and denitions for terms
related to driving automation systems for On-Road motor
vehicles.Warrendale,PA:SAEInternational.
Scarinci, R., & Heydecker, B. (2014). Control concepts for facil-
itating motorway On-ramp merging using intelligent vehi-
cles. Transp ort Re v iew s ,34(6), 775–797.
Shim, T., Adireddy, G., & Yuan, H. (2012). Autonomous vehicle
collision avoidance system using path planning and model-
predictive-control-based active front steering and wheel
torque control. Proceedings of the Institution of Mechani-
cal Engineers, Part D: Journal of Automobile Engineering,
226(6), 767–778.
Shladover, S. E. (1995). Review of the state of development of
Advanced Vehicle Control Systems (AVCS). Vehicl e S y stem
Dynamics,24(6–7), 551–595.
Shladover, S. E. (2005). Automated vehicles for highway opera-
tions (automated highway systems). Proceedings of the Insti-
tution of Mechanical Engineers, Part I: Journal of Systems and
Control Engineering,219(1), 53–75.
Shladover, S. E., Su, D., & Lu, X.-Y. (2012). Impacts of coopera-
tive adaptive cruise control on freeway trac ow. Trans-
portation Research Record: Journal of the Transportation
Research Board,2324, 63–70.
Siebert, F. W., Oehl, M., & Pster, H. R. (2014). The inuence of
timeheadwayonsubjectivedriverstatesinadaptivecruise
control. Transportation Research Part F: Trac Psychology
and Behaviour,25(PART A), 65–73.
Spieser, K., Treleaven, K., Zhang, R., Frazzoli, E., Morton, D.,
& Pavone, M. (2014). Toward a systematic approach to the
design and evaluation of automated Mobility-on-Demand
systems: A case study in Singapore. In G. Meyer & S. Beiker
(Eds.), Road vehicle automation (pp. 229–245). Switzerland:
Springer International Publishing.
Spyropoulou, I., Penttinen, M., Karlaftis, M., Vaa, T., &
Golias, J. (2008). ITS Solutions And Accident Risks:
Prospective and limitations. Transpo rt Rev i ews,28(5), 549–
572.
Stanton, N. A., & Young, M. S. (1998). Vehicle automation and
driving performance. Ergonomics,41(7), 1014–1028.
Stevens, W. B. (1997). Societal and institutional aspects of AHS
deployment. In P. A. Ioannou (Ed.), Automated highway sys-
tems (pp. 335–348). New York: Plenum Press.
Strand, N., Nilsson, J., Karlsson, I. C. M., & Nilsson, L. (2014).
Semi-automated versus highly automated driving in criti-
cal situations caused by automation failures. Transportation
Research Part F: Trac Psychology and Behaviour,27, 218–
228.
Sun, Z., Bebis, G., & Miller, R. (2006). On-Road vehicle detec-
tion: A review. IEEE Transactions on Pattern Analysis and
Machine Intelligence,28(5), 694–711.
Talebpour, A., & Mahmassani, H. S. (2016). Inuence of con-
nected and autonomous vehicles on trac ow stability and
throughput. Transportation Research Part C: Emerging Tech-
nologies,71, 143–163.
Turner, J. D., & Austin, L. (2000). A review of current sensor
technologies and applications within automotive and traf-
c control systems. Proceedings of the Institution of Mechan-
ical Engineers, Part D: Journal of Automobile Engineering,
214(6), 589–614.
Vahidi, A., & Eskandarian, A. (2003). Research advances in
intelligent collision avoidance and adaptive cruise control.
IEEE Transactions on Intelligent Transportation Systems,
4(3), 143–153.
Vajedi, M., & Azad, N. L. (2015). Ecological adaptive cruise con-
troller for Plug-In hybrid electric vehicles using nonlinear
model predictive control. IEEE Transactions on Intelligent
Transportation Systems,17(1), 113–122.
van Arem, B., & Smits, C. (1997). An exploration of the devel-
opment of automated vehicle guidance systems.Delft,The
Netherlands: TNO.
van Arem, B., van Driel, C., & Visser, R. (2006). The impact of
Co-operative Adaptive Cruise Control on trac ow char-
acteristics. IEEE Transactions on Intelligent Transportation
Systems,7(4), 429–436.

348 D. MILAKIS ET AL.
van den Berg, V., & Verhoef, E. T. (2016). Autonomous cars
and dynamic bottleneck congestion: The eects on capacity,
value of time and preference heterogeneity. Transportation
Research Part B,94, 43–60.
Vogt, R., Wang, H., Gregor, B., & Bettinardi, A. (2015).
Potential changes to travel behaviors & patterns: a fuzzy
cognitive map modeling approach. Transportation,42(6),
967–984.
Wadud, Z., MacKenzie, D., & Leiby, P. (2016). Help or hin-
drance? The travel, energy and carbon impacts of highly
automated vehicles. Transportation Research Part A: Policy
and Practice,86, 1–18.
Wang, Z., Chen, X. M., Ouyang, Y., & Li, M. (2015). Emission
mitigation via longitudinal control of intelligent vehicles in
aCongestedPlatoon.Computer-Aided Civil and Infrastruc-
ture Engineering,30, 490–506.
Wang, M., Daamen, W., Hoogendoorn, S. P., & van Arem,
B. (2016a). Connected variable speed limits control and
car-following controlwith vehicle-infrastructure communi-
cation to resolve stop-and-go waves. Journal of Intelligent
Transportation Systems: Technology, Planning, and Opera-
tions.
Wang, M., Daamen, W., Hoogendoorn, S. P., & van Arem, B.
(2016b). Cooperative Car-Following control: Distributed
algorithm and impact on moving jam features. IEEE Trans-
actions on Intelligent Transportation Systems,17(5), 1459–
1471.
Wang, M., Hoogendoorn, S., Daamen, W., & van Arem, B.
(2014). Potential impacts of ecological adaptive cruise con-
trol systems on trac and environment. IET Intelligent
Transp ort Syst ems,8(2), 77–86.
Wang, M., Hoogendoorn, S. P., Daamen, W., van Arem, B.,
& Happee, R. (2015). Game theoretic approach for pre-
dictive lane-changing and car-following control. Trans-
portation Research Part C: Emerging Technologies,58, 73–
92.
Ward, J. D. (1994). An hypothesized evolution of an automated
highway system. Seal Beach, CA: Rockwell International
Corporation.
Wu, C., Zhao, G., & Ou, B. (2011). A fuel economy optimization
system with applications in vehicles with human drivers
and autonomous vehicles. Transportation Research Part D:
Transp ort and En vir onm ent ,16(7), 515–524.
Xiao, L., & Gao, F. (2010). A comprehensive review of the devel-
opment of adaptive cruise control systems. Vehicle S y stem
Dynamics,48(10), 1167–1192.
Xie, Y., Zhang, H., Gartner, N. H., & Arsava, T. (2016). Collab-
orative merging strategy for freeway ramp operations in a
connected and autonomous vehicles environment. Journal
of Intelligent Transportation Systems: Technology, Planning,
and Operations.
Xing, Y., Handy, S. L., & Mokhtarian, P. L. (2010). Factors associ-
ated with proportions and miles of bicycling for transporta-
tion and recreation in six small US cities. Transportation
Research Part D: Transport and Environment,15(2), 73–81.
Xiong,H.,Boyle,L.N.,Moeckli,J.,Dow,B.R.,&Brown,T.L.
(2012). Use patterns among early adopters of adaptive cruise
control. Human Factors: The Journal of the Human Factors
and Ergonomics Society,54(5), 722–733.
Yang, K., Guler, S. I., & Menendez, M. (2016). Isolated intersec-
tion control for various levels of vehicle technology: Con-
ventional, connected, and automated vehicles. Transp orta -
tion Research Part C,72, 109–129.
Yang, S., Liu, W., Sun, D., & Li, C. (2013). A new extended
multiple Car-Following Model considering the Backward-
Looking eect on trac ow. Journal of Computational and
Nonlinear Dynamics,8(1), 11016.
Yap, M. D., Correia, G., & van Arem, B. (2016). Preferences of
travellers for using automated vehicles as last mile public
transport of multimodal train trips. Transportation Research
Part A: Policy and Practice,94, 1–16.
Young, M. S., & Stanton, N. A. (2007). Back to the future:
brake reaction times for manual and automated vehicles.
Ergonomics,50(1), 46–58.
Zakharenko, R. (2016). Self-Driving cars will change cities.
Regional Science and Urban Economics,61, 26–37.
Zhang, W., Guhathakurta, S., Fang, J., & Zhang, G. (2015).
Exploring the impact of shared autonomous vehicles
on urban parking demand: An agent-based simulation
approach. Sustainable Cities and Society,19, 34–45.
Zhou, M., Qu, X., & Jin, S. (2016). On the impact of cooper-
ative autonomous vehicles in improving freeway merging:
A modied intelligent driver Model-Based approach. IEEE
Intelligent Transportation Systems Magazine.
Zmud, J., Sener, I. N., & Wagner, J. (2016). Self-Driving vehicles
determinants of adoption and conditions of usage. Trans-
portation Research Record: Journal of the Transportation
Research Board,2665, 57–64.
Zohdy, I. A., & Rakha, H. A. (2016). Intersection manage-
ment via vehicle connectivity: The intersection cooperative
adaptive cruise control system concept. Journal of Intelligent
Transportation Systems,20(1), 17–32.