Content uploaded by Paolo Delle Site
Author content
All content in this area was uploaded by Paolo Delle Site on May 15, 2015
Content may be subject to copyright.
Available via license: CC BY-NC-ND 3.0
Content may be subject to copyright.
Transportation Research Procedia 3 ( 2014 ) 139 – 144
2352-1465 © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and peer-review under responsibility of the Scientific Committee of EWGT2014
doi: 10.1016/j.trpro.2014.10.099
Available online at www.sciencedirect.com
ScienceDirect
17th Meeting of the EURO Working Group on Transportation, EWGT2014, 2-4 July 2014,
Sevilla, Spain
Users’ preferences towards automated road public transport: results
from European surveys
Adriano Alessandrinia,b, Raffaele Alfonsia,b, Paolo Delle Sitea,b *, Daniele Stama,b
aDICEA, Department of Civil Architectural and Environmental Engineering, University of Rome La Sapienza, Via Eudossiana 18, 00184 Rome,
Italy
bCTL –Research Centre for Transport and Logistics, University of Rome La Sapienza, Via Eudossiana 18, 00184 Rome, Italy
Abstract
Collective automated road transport systems (ARTS) are the subject of current research in Europe. The paper reports on the
results of the investigations about users’ attitudes towards ARTS and conventional buses that have been carried out in twelve
cities where the implementation of an ARTS service is being planned within the Citymobil2 project. A common stated preference
questionnaire has been used. The econometric analysis has been based on the estimation of a logit model which has considered
the choice for two alternatives: ARTS and minibus. The observed attributes are: waiting time, riding time and fare. Of particular
interest, is the estimation of the alternative specific constant (ASC) of the ARTS, because this represents the mean of all the
unobserved attributes of the automated system that affect the choice. With a common specification of the systematic utilities of
ARTS and minibus, the observed attributes being the same, a positive value of the ASC is indicative of a relatively higher
preference for the ARTS. The results show a relatively higher preference for ARTS across the cities where the ARTS is
implemented inside a major facility. In other application contexts, commonalities in attitudes across cities are not found. The
impacts on attitudes of the socio-economic attributes of the users are heterogeneous across cities.
© 2014The Authors. Published by Elsevier B. V.
Selection and peer-review under responsibility of the Scientific Committee of EWGT2014.
Keywords: automated road public transport; stated preferences; discrete choice; random utility; logit
1. Introduction
Automation in collective public transport has been a reality for several years by now in the case of guided
systems with fully segregated right-of-way. Examples include automated metro and automated people movers in
airports. Research has been undertaken in Europe in recent years with the aim of implementing fully automated road
* Corresponding author. Tel.: +39-06-44585923; fax: +39-06-44585774.
E-mail address: paolo.dellesite@uniroma1.it
© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and peer-review under responsibility of the Scientific Committee of EWGT2014
140 Adriano Alessandrini et al. / Transportation Research Procedia 3 ( 2014 ) 139 – 144
collective transport systems (ARTS –Automated Road Transport Systems). The key advantage of ARTS is
identified in the potential for offering a higher frequency of service in the off-peaks, provided the operating costs are
lower than a conventional bus. Also, there is a potential for higher flexibility in adapting the supply to demand
because of the lack of drivers’ scheduling constraints. In this latter respect, it is worth mentioning that the on-
demand functionality of collective vehicles is limited by vehicle size constraints (Personal Rapid Transit, PRT, have
not been the subject of the research reported on here).
The technology is available and a few demonstrations of ARTS have taken place. The key barriers to
implementation are legal. The Rivium in Rotterdam is the only system currently operated on a permanent basis. The
system serves routes in peripheral areas of the city, with collective automated vehicles running along a dedicated
lane with priority at intersections (http://www.advancedtransit.org/advanced-transit/applications/rivium/).
Demonstrations of ARTS on routes in mixed traffic, low-speed, environments are currently being planned within the
Citymobil2 project of the Seventh Framework Programme of the European Commission
(http://www.citymobil2.eu/en/).
The attitudes of the users towards ARTS, i.e. collective automated road transport systems, are still largely
unexplored. Most of the studies for which literature is available investigated the potential demand for ARTS services
on specific routes (Bekhor and Zvirin, 2004; CyberMove Consortium, 2004; NETMOBIL Consortium, 2005). Only
a few studies have tackled the research objective of assessing the relative preferences of the users towards ARTS
vis-à-vis a conventional bus. One took place in Leeds (Shires and Ibañez, 2008), one in Rome (Delle Site et al.,
2011). Both were based on stated preference (SP) survey methodology (Louviere et al., 2000; Hensher et al., 2005),
which is nowadays common practice in transportation planning and economic studies. Other studies have considered
personal rapid transit (among these: Minderhoud and van Zuylen, 2005; Shires and Ibañez, 2008; Cirillo and
Hetrakul, 2010; Cirillo and Xu, 2010).
The Citymobil2 project gave the opportunity to investigate the attitudes of the users towards automation in
collective road transport in a variety of application contexts. Technical and non-technical feasibility studies
concerning the implementation of a small scale ARTS service have been conducted in twelve cities across Europe.
The cities have selected the route for the proposed demonstration. The routes include a range of applications: within
city centre, within a major facility (such as a technology park or a university), from public transport node to a major
facility, and from public transport node to a residential area.
The paper reports on the investigation which has been carried out with the potential users of the ARTS service
that has been planned in each city. A common SP questionnaire across the twelve cities has been used, with the aim
of investigating the relative preferences towards ARTS and conventional systems. Questionnaire responses have
been analysed using logit models (Ben-Akiva and Lerman, 1985).
2. Methodology
2.1. Stated preference questionnaire and surveys
An SP questionnaire common across cities has been used. First, the route of the public transport service under
planning is described. A brief description of two vehicle options, a conventional minibus and an ARTS, is provided.
It is specified that the two vehicles are equal in terms of propulsion, and of total and seating capacity. Both will run
in mixed traffic. The difference is the presence/absence of the driver. A picture of both is shown.
In the second part of the questionnaire, respondents are asked to choose between a minibus and an ARTS in
different supply scenarios for a trip of a given length; the supply scenarios are defined according to different levels
of the waiting time, riding time and fare of the minibus and the ARTS. The third part relates to the personal
characteristics of the respondents: gender, age, income, education, occupation, car availability in the household,
ownership of a public transport monthly ticket.
The attributes and corresponding levels of the SP design are in Table 1. The number of combinations in the full
factorial design (8 combinations) has been reduced to 4 combinations using a within-alternative orthogonal design
technique.
Samples are composed by potential users of the ARTS services under planning in each city. Data collection
activities took place in Spring 2013. The majority of surveys used face-to-face interviews, in a minority of cases on-
141
Adriano Alessandrini et al. / Transportation Research Procedia 3 ( 2014 ) 139 – 144
line questionnaires have been used. In a few cities the level of the extra-fare was reduced to 1 EUR per return
journey to take into account local conditions.
Table 1. Stated preference design –attributes and levels
Alternative attribute number of levels levels
minibus/ARTS waiting time 23/8 minutes
riding time 25/10 minutes
fare 2as other public transport means
in the city/extra-fare of 2 EUR
per return journey
2.2. The econometric model
A binomial logit model is considered. We denote by 1 the minibus alternative, by 2 the ARTS alternative. The
basic model has the following specifications of systematic utilities:
(1)
where WT is waiting time, RT is riding time, FA is fare,
1
E
,
2
E
,
3
E
are the coefficients, ASC is the alternative
specific constant of ARTS.
For fare, “effects coding” (-1/1) has been used instead of “dummy coding” (0/1) to eliminate confounding with
the ASC: the code -1 represents the case where an extra-fare is paid; the code +1 represents the case where the same
fare as other public transport is paid.
In the basic specification, the coefficients, which are representative of the marginal utilities of the attributes, are
common to the two alternatives. A specification with alternative-specific coefficients has also been considered.
Other specifications include socio-economic variables of the users (gender, age, income, education, occupation,
car availability in the household, ownership of a public transport monthly ticket) in the systematic utility of ARTS.
Of particular interest, is the estimation of the ASC of the ARTS, because this represents the mean of all the
unobserved attributes that affect the choice: with a common specification of the systematic utilities of ARTS and
minibus, the observed attributes being the same, a positive value of the ASC is indicative of a relatively higher
preference for the ARTS than for the minibus, because of the resulting higher choice probability.
Also of interest, is the estimation of the effect on the ASC of the socio-economic attributes of the users, because
this is indicative of the effect of the socio-economic attribute on the relative preference for ARTS (e.g. whether
preference for ARTS increases with age or viceversa).
3. Application cases
According to the routes proposed for the ARTS demonstrations by the twelve cities, the following four
application cases are included:
A1 -within city centre: La Rochelle (France), Oristano (Italy), Reggio Calabria (Italy), Trikala (Greece);
A2 -within major facility: CERN (Switzerland), Lausanne (Switzerland), San Sebastian (Spain), Sophia Antipolis
(France);
A3 -from public transport node to major facility: Brussels (Belgium), León (Spain), Milan (Italy);
A4 -from public transport node to residential area: Vantaa (Finland).
The demonstrations that are planned to be actually implemented within the CityMobil2 project may be different
from the scenarios of the SP questionnaire. In the planned demonstrations, segregated lanes or mixed traffic lanes
will be used, the ARTS service will be free of charge.The infrastructure at stops will tend to be minimized because
ASCFARTWTV
FARTWTV
3212
3211
EEE
EEE
142 Adriano Alessandrini et al. / Transportation Research Procedia 3 ( 2014 ) 139 – 144
the system will be operated on a non-permanent basis; no elevated platforms are required; vehicles will have
unfolding ramps for wheelchairs.
3.1. Within City Centre
In La Rochelle the system links the railway station and the central park, with a length of 2.6 km. In Oristano the
service connects the seafront of Torre Grande and the touristic port, with a length of about 6 km. In Reggio Calabria
the Arts covers 4.2 km linking the Railway Central Station and the Executive Administrative Centre. In Trikala the
system serves the areas of Varussi and Central Square, covering a length of 1.8 km.
3.2. Within major facility
In the CERN campus the system is thought to serve the main building area and the Restaurant #2, with a length of
2 km. In Lausanne the system is deployed in the EPFL campus where connects the north with the south area with
length of 1.4 km. In San Sebastian the Arts runs inside the Miramón Technology park, linking the entrance of the
park and the head offices of the different companies for a 2.1 km length. In Sophia Antipolis the system runs for 3.8
km linking the areas of Trois-Mulins and Saint-Philippe.
3.3. From public transport node to major facility
In Brussels the ARTS route is 2 km long and links the Saint-Luc Hospital with the Kraainem metro station. In
León the system links the FEVE train station “La Asunción” and the University Campus, with an itinerary of 1.5 km.
In Milan the system connects the “Molino –Dorino” metro station with the south gate of the EXPO 2015 (Milan
Universal Exhibition).
3.4. From public transport node to residential area
In Vantaa the ARTS links the Kivistö railway station and its residential area with a length of 2 km.
4. Results
4.1. Econometric analysis
The results of the estimation of the model with the basic specification of the systematic utilities for the twelve
samples are in Table 2. All the time and fare coefficients have the right sign. On the basis of the sign of the ASC, a
relatively higher preference for ARTS is found in all cities of the application case “within major facility”; in three
out of four of these cases the ASC of ARTS is statistically significant (null hypothesis of a zero coefficient, 10%
significance level, two-tailed). In the other application cases, the relative preference is in some cities for the ARTS,
in other for the minibus.
Estimation with specifications of the utility of ARTS including socio-economic characteristics of the users (not
shown) have provided the following results. In ten cities out of the twelve the attribute “male” induces a relatively
higher preference for the ARTS, while the attribute “female” induces a relatively lower preference for the ARTS.
This occurs, with statistical significance (null hypothesis of a zero coefficient, 10% significance level, two-tailed), in
all the four cities of the “within major facility” application case.
A commonality across cities in the sign of the effect on preferences of variables such as age, income, education,
occupation, availability of car and ownership of public transport monthly ticket, is not found; the effects of these
variables are heterogeneous across cities. In ten cities out of the twelve,the attribute “male” induces a relatively
higher preference for the ARTS while the attribute “female” induces a relatively lower preference for the ARTS; this
occurs, with statistical significance, in all the four cities of the “within major facility” application case. In three out
of four cities of the “within city centre” application case, a higher education level induces a statistically significant
relatively higher preference for ARTS.
143
Adriano Alessandrini et al. / Transportation Research Procedia 3 ( 2014 ) 139 – 144
The econometric analysis has also shown that alternative-specific coefficients for the time and cost attributes are
not statistically significant, with the exception of the riding time in one city only (Brussels). This means that the
marginal disutility of travel time (in particular, the travel time spent on-board) is found to be the same, irrespective
of which of the two systems, ARTS or minibus, are used. In the case of Brussels only, the disutility of the travel time
spent on-board the ARTS is found to be lower than the minibus.
Table 2. Model estimation in the twelve cases (basic specification –coefficient and t-statistic).
variable
A1 A2 A3 A4
La Rochelle (FR)
Oristano (IT)
Reggio Calabria
(IT)
Trikala (GR)
CERN (CH)
Lausanne (CH)
San Sebastian
(ES)
Sophia Antipolis
(FR)
Brussels (BE)
Leon (ES)
Milan (IT)
Vantaa (FI)
waiting
time
-.0452
-1.11
-.32052
-6.22
-.1925
-3.58
-.0960
-2.28
-.38293
-13.24
-.3778
-15.22
-.2003
-4.39
-.3131
-8.42
-.2094
-4.92
-.0372
-.53
-.23900
-5.30
-.1562
-2.18
riding
time
-.0563
-1.95
-.29100
-8.67
-.2471
-6.89
-.0872
3.02
-.34816
-17.28
-.3976
-23.18
-.2078
-6.72
-.2807
-10.75
-.1989
-6.45
-.2414
-4.83
-.19362
-6.35
-.0067
-.14
fare .50548
4.03
1.56972
9.08
.78382
4.41
.46828
3.57
.99446
11.62
.55397
8.19
.83580
5.69
.73213
6.08
.49727
3.64
.35450
1.60
1.11167
7.64
.45181
1.89
ASA
(ARTS)
.56365
3.90
1.19353
7.11
-.4746
-2.65
-.0136
-.09
1.22331
12.14
.86415
10.07
.28531
1.85
.16204
1.24
-.2753
-1.85
-1.870
-7.48
.86800
5.69
-1.568
-6.34
sample
size
200 200 209 208 482 742 200 290 201 227 200 167
Key:
A1 = within city centre
A2 = within major facility
A3 = from public transport node to major facility
A4 =from public transport node to residential area
4.2. Comparison with previous studies
Shires and Ibañez (2008) and Delle Site et al. (2011) are relevant references because they have used the same
methodology. i.e. logit models calibrated on SP data. The former has considered the relative preference between a
conventional bus service and an ARTS in the context of an existing public transport line in Leeds (UK). The latter
has considered the relative preference between a minibus and an ARTS for a short connection with a parking area in
the new trade fair district in Rome (Italy).
Shires and Ibañez (2008) have found a relatively higher preference for the existing bus service compared with an
ARTS, a result which, according to the authors, can be ascribed to an inbuilt bias towards the currently used
transport mode. In the cities of the surveys reported on here new transport services (i.e. not currently existing) are
considered, except the case of Reggio Calabria. Therefore, results are affected to a lower extent by the currently used
mode bias, a circumstance that is confirmed by the relatively higher preference for the ARTS found in some of the
“within city centre” cases (namely La Rochelle and Oristano). The Reggio Calabria case shows a statistically
significant relatively higher preference for the minibus, a result which is similar to the one of Leeds.
Delle Site et al. (2011) have found a relatively higher preference for the ARTS in the case of the trade fair district.
This result is similar to those found in the “within major facility” cases (namely CERN, Lausanne, San Sebastian,
Sophia Antipolis), and in the case of Milan (connection between a metro station and the exhibition), which all show
a relatively higher preference for the ARTS.
144 Adriano Alessandrini et al. / Transportation Research Procedia 3 ( 2014 ) 139 – 144
5. Conclusions
The surveys have portrayed, based on twelve cross-European samples, the attitudes towards automation in
collective road transport systems. The samples relate to the potential users of ARTS for specific planned routes in a
variety of application contexts in twelve different European cities. The results are representative of the attitudes of
users who have been exposed to minimal information about ARTS, and who have had no experience of these
innovative systems. Information and experience are, together with emotions, the key determinants of preference
formation according to the standard behavioural choice model (McFadden, 2014). This is an important issue with
implications for future research on the stability of attitudes with information and experience (the impact of
experience on attitudes towards an electric and a conventional vehicle has been tackled recently; Jensen et al., 2013).
The cases investigated (except one city) relate to new services on routes where public transport is currently not
existing. The results show that automation is, on average, not necessarily perceived as valuable, if the travel time and
fare of the ARTS are the same as those of a conventional bus. A common trait of the results across the cases is the
relatively higher preference for ARTS when this is implemented within a major facility. Common traits relating to
the impacts on preferences of socio-economic characteristics of the users have not been found.
The econometric models that have been estimated are useful tools to obtain the shifts in preferences that would
result from an increase of the relative frequency of the ARTS on that of a bus, or to compute the increase in the
relative frequency of the ARTS that is needed to achieve a relatively higher preference. Also, they implicitly provide
an assessment of the willingness-to-pay of the users for automation, since they can provide estimates of the shifts in
preferences that would follow different fare policies.
Acknowledgements
The survey activities have been carried out by the partners of the Citymobil2 project involved in the feasibility
studies in the twelve cities.
References
Bekhor S. and Zvirin Y. (2004) Estimating the potential use of cybernetic cars for a university campus. Proceedings of the 10th WCTR World
Conference on Transport Research, Istanbul, Turkey.
Ben-Akiva M, Lerman S.R. (1985) Discrete Choice Analysis. Theory and Application to Travel Demand. The MIT Press, Cambridge MA.
Cirillo C., Hetrakul P. (2010) Continuous random coefficient logit models: a comparison of parametric and non-parametric methods to estimate
individual prefer ences over Cybernetic Transportation Systems. Paper prepared for presentation at the 89th Annual Meeting of the
Transportation Research Board, Washington, D.C.
Cirillo C., Xu R. (2010) Forecasting cybercar use for airport ground access: case study at Baltimore Washington International Airport. Journal of
Urban Planning and Development 136(3), 186-194.
CyberMove Consortium (2004) Ex-ante Evaluation. Deliverable D2.3a&6.2b of the CyberMove (Cybernetic transport systems for the cities of
tomorrow) project. Fifth Framework Programme, European Commission.
Delle Site P., Filippi F., Giustiniani G. (2011) Users’ preferences towards innovative and conventional public transport. Procedia -Social and
Behavioural Sciences 20, 906-915.
Hensher, D.A., Rose J.M., Greene W.H. (2005) Applied Choice Analysis. A Primer. Cambridge University Press, Cambridge, UK.
Jensen A.F., Cherchi E., Mabit S.L. (2013) On the stability of preferences and attitudes before and after experiencing an electric vehicle.
Transportation Research Part D 25, 24-32.
Louviere J., Hensher D.A., Swait J. (2000) Stated Choice Methods. Analysis and Applications. Cambridge University Press, Cambridge, UK.
McFadden D. (2014) The New Science of Pleasure. Consumer Choice Behavior and the Measurement of Well-Being. To appear in: Hess S., Daly
A. (eds) Handbook of Choice Modelling. Edward Elgar.
Minderhoud M. M. and van Zuylen H. J. (2005) Willingness-to-pay for personal rapid transit in the city of Almelo. Proceedings of the 10th APM
Automated People Movers Conference, Orlando, Florida.
NETMOBIL Consortium (2005) EU Potential for Innovative Personal Urban Mobility. Deliverable D7 of the NETMOB IL (New transport
system concepts for enhanced and sustainable personal urban mobility) project. Fifth Framework Programme, European Commission.
Shires J.D., Ibañez N. (2008) CityMobil and DISTILLATE. Stated Preference and Ranking Surveys. Final Report. ITS, Institute for Transport
Studies, University of Leeds.