Preferences of travellers for using automated vehicles as last mile public transport of multimodal train trips
Introduction
In the last years many developments took place regarding automated vehicles (AVs) technology. In fact automated vehicles are expected to become available on the market in the next 10–20 years (Shladover, 2015). Most studies on automated vehicles focus on the vehicle technology in relation to the effect on traffic flow characteristics, road capacity and traffic safety. The relation between different penetration rates of automated and cooperative transport systems and road capacity has been for example studied by Van Arem et al., 2006, Tampère et al., 2009, Arnaout and Bowling, 2011, Shladover et al., 2012, Hoogendoorn et al., 2014 and Schakel and Van Arem (2014). Kesting et al. (2005) also consider the effect of AVs on the capacity drop after congestion. The impact of AVs on traffic stability has, amongst others, been studied by Schakel et al. (2010). VanMiddlesworth et al. (2008) studied AVs in intersections management, whereas Van Driel and Van Arem (2010) considered the effect of AVs on both traffic flow efficiency and traffic safety.
It is however unknown to which extent the share of the existing transport modes will change as result of using AVs as a transit system (Correia et al., 2016). To the best of our knowledge this study is the first where traveller preferences for AVs are explored and compared to existing modes. Thereby its main objective is to position AVs in the transportation market and understand the sensitivity of travellers towards some of their travel attributes. Because there are no fully automated vehicles currently on the market we apply a stated preference (SP) experiment where the role of classic instrumental variables such as travel time and cost are explored. Moreover, due to the fact that these vehicles entail moving on the road network without a driver and entrusting that task to a computer, we expect that psychological factors translated through positive and negative attitudes play an important role in the choice to use automated vehicles. Therefore, in our SP experiment the role of attitudes in perceiving the utility of AVs is particularly explored in addition to these classical instrumental attributes.
Five different levels of automation are defined by Gasser and Westhoff (2012) and SAE International (2014). These 5 levels are driver support (level 1), partial automation (level 2), conditional automation (level 3), high automation (level 4) and full automation (level 5). A higher level of automation entails a less important role for the human driver in the driving task. Our study focuses on AVs which are able to operate according to level 5 automation, meaning that there is a full time performance of an automated driving system for all driving tasks, without any human intervention. In our SP experiment, we also explicitly assume that these vehicles are fully electrically powered, thus representing a lower environmental impact at least at the local level.
The scope of this paper is on studying the potential of AVs for the last mile trips between a train station and the travellers’ final destination. We realize that a modal shift from car to trains as main mode on medium-distance (20–40 km) trips is an important policy goal of the Dutch government (Ministry of Infrastructure and Environment, 2015). A higher train usage entails a higher level of sustainability in transportation and can also reduce congestion levels, with its related economic and environmental impacts. Currently, in the Netherlands there is a relatively high share of multimodal trips for medium-distance trips between urban areas. Between the most developed urban areas in the Netherlands, up to 17% of the trips are currently considered as being multimodal (a trip in which a traveller uses at least two different modes) (Van Nes et al., 2014). In multimodal train trips, a relatively high disutility is especially caused by the access and egress trip stage (Hoogendoorn-Lanser, 2005), hence it is hypothesized that providing AVs as egress mode may have the potential to improve the attractiveness of multimodal train trips and to realize a modal shift to the train + AV combination. AVs are thus considered as potential means to increase the attractiveness of the total door-to-door trip, by providing a last mile service which brings travellers from the train station to the front door of their final destination in a sustainable way.
The paper is structured as follows. Section 2 presents the applied methodology to investigate travellers’ preferences for using AVs. In Section 3, the survey and sample are shortly discussed. Section 4 shows and discusses the results of the final estimated model. At last, conclusions and recommendations for further research are presented in Section 5.
Section snippets
Alternatives and attributes
Public transport (PT) trips usually consist of three stages: access, main part and egress. We define a multimodal PT trip in this paper as a trip where more than one mode is used, with one or more public transport modes being used for the main part of the trip. For each stage different alternatives are available, such as walking, cycling, private car or bus, tram and metro (denoted as BTM) for access; train or BTM for the main stage; and walking, cycling, car-sharing or BTM for the egress part.
Survey and sample
We designed an online survey which was divided in several parts. At the start, automated vehicles and their role as last mile transport for PT trips were introduced in a brief and objective way. In the survey (which is in Dutch), AVs were introduced by using the italic text below which is translated from Dutch language, including two pictures of AVs and a schematic overview of the door-to-door trip with the relevant unimodal and multimodal alternatives:
“Over the last years, many developments
Results of the latent variable model
In Table 5 we show the results of the estimated latent variable model to determine the latent attitudinal factors relevant for the choice of AVs as last mile transport. After checking communalities between indicators and first checking whether a simple structure could be reached when performing a skewed rotation, a simple structure could be obtained by performing an orthogonal, varimax rotation. Indicators with a communality <0.25 or with all factor loads <0.50 were excluded from the
Conclusions and discussion
The aim of this study was to position automated vehicles in the public transport market and to understand the sensitivity of travellers towards instrumental travel attributes – like different travel time components and travel costs – socio-economic variables and attitudinal factors. Because there are no fully-automated vehicles currently on the market, we applied a SP experiment. Based on the results of this experiment, we can formulate several main conclusions. First, by travellers using first
Acknowledgements
The authors thank ProRail, the train infrastructure manager of the Netherlands, for financing the D2D100%EV project under which this study was carried out.
References (42)
- et al.
Travelers’ preferences in multimodal networks: design and results of a comprehensive series of choice experiments
Transp. Res. Part A
(2013) - et al.
Development of commuter and non-commuter mode choice models for the assessment of new public transport infrastructure projects: a case study
Transp. Res. Part A
(2007) - et al.
Intention to use a fully automated car: attitudes and a priori acceptability
Transp. Res. Part F
(2014) - et al.
Designing efficient SP experiments in the presence of reference alternatives
Transp. Res. Part B
(2008) - et al.
Generalized random utility model
Math. Social Sci.
(2002) - et al.
Towards reducing traffic congestion using cooperative adaptive cruise control on a freeway with a ramp
J. Indust. Eng. Manage.
(2011) BIOGEME: a free package for the estimation of discrete choice models
- et al.
Designing SP experiments: state-of-the-art
- et al.
Construction of experimental designs for mixed logit models allowing for correlation across choice observations
- et al.
Approximation of Bayesian efficiency in experimental choice designs
J. Choice Model.
(2008)
Modelling route-choice behavior in multi-modal transport networks
Transportation
A Study of Public Acceptance of Autonomous Cars
Ngene 1.1.1 User Manual & Reference Guide.
Vehicle automation for improving transport system performance: conceptual analysis, methods and impacts
Cited by (255)
Shared versus pooled automated vehicles: Understanding behavioral intentions towards adopting on-demand automated vehicles
2024, Travel Behaviour and SocietyExploring factors influencing consumer preferences for automated driving vehicles
2024, Transportation EngineeringDemographic and social differences in autonomous vehicle technology acceptance in Hungary
2024, Journal of Engineering and Technology Management - JET-MThe relevance of social factors in sharing a trip with strangers: Creating travel communities in the autonomous vehicles era
2024, Travel Behaviour and SocietyModeling the joint choice behavior of commuters’ travel mode and parking options for private autonomous vehicles
2024, Transportation Research Part C: Emerging Technologies