Understanding the impacts of automated mobility: User experiences stakeholder perspectives and vehicle performance across Europe

Automated vehicles are an innovation that have the potential to not only revolutionize urban transport,
but also transform cities. However full commercial deployment remains far from reality in Europe. Pilot
projects in Europe have multiplied over the years, with a focus on shared autonomous mobility services,
but questions remain about the actual performance of these services, and how autonomous mobility is
perceived by users and other stakeholders. As such, the potential implications of autonomous mobility
are uncertain. Thus, we aimed in this thesis to investigate the implications of different forms of automated
mobility at different levels and from multiple perspectives, answering the following question: What are
the current perceptions and expectations of users and stakeholders on automated mobility and how
do the vehicles perform in real deployments?
Starting with privately owned automated vehicles, we captured and analyzed the expectations of po-
tential future users in terms of time use and activity scheduling using a survey in chapter 2. The re-
spondents had no exposure to automated vehicles within the context of this study, and thus the responses
were based on the information they were provided and their own knowledge and imagination. Through
our analysis, we identified heterogeneity in the expectations of the effect automated vehicles would have
on activity scheduling. This effect is not the same across all participants, as some would use the newly
available time for leisure purposes, while others would be more interested in the time saving benefits.
This is not only influenced by personal characteristics, as well as commitment to the survey, but also by
how useful they perceive AVs to be, and how they would make use of that utility for their own needs. For
some, the main attraction of an AV is its utilitarian benefits and potential to allow schedule optimization.
For others who may face strong time pressure, the most attractive feature of an AV might be the level of
comfort which would allow the space and time to relax and unwind, which they may have not had time
for previously. In the end, not everyone perceived AVs in the same way, and assuming that the ability to
engage in activities on-board will be available and attractive to all is too simplistic and does not consider
variability in individual perception and interests.
Moving on to impacts at a wider scale, and understanding the limitations of such a speculative ap-
proach, we explored in the next chapter virtual reality as a tool to provide a realistic view of automated
vehicles within a system and to support the evaluation of user acceptance. In chapter 3, we not only
measure acceptance using a model based on the UTAUT theory, we also do so within a pre-test post-test
design, before and after a virtual experience in which the visual appearance of Brussels was transformed
with the assumption that shared automated vehicles became the dominant transport mode. As such, we
could compare the responses and evaluate the effect of the virtual experience in shaping the perception
of shared automated mobility. We found that the virtual reality experience itself had a significant effect
on acceptance, observed not only through increases of the ratings of different items and constructs of
acceptance, but also through changes in the constructs predicting behavioural intention. The perception
of the urban form in the virtual environment became a significant factor associated with acceptance after
the experience, suggesting that the perception of AVs may not be solely limited to how people perceive
the vehicles themselves and the utility from which they can benefit, but it could also be related to the
indirect consequences they could bring.
However, the perspectives of users are not the only ones of interest, as there are many other stake-
holders who have vested interests in automated mobility. With this, we expanded in chapter 4 our scope
of analysis to other stakeholders (authorities, transport operators, mobility service providers etc.) and
to multiple scenarios of shared automated mobility (automated mass transit, feeder services, shared rob-
otaxis, door-to-door automated shuttles). Applying a multi-actor multi-criteria approach in workshops,
we captured stakeholders’ evaluation of the aforementioned scenarios across 6 European countries and
identified the critical factors in their assessment. We identified heterogeneity in the perspectives of pub-
lic actors, like transport operators, who had a clear preference for automated modes with higher sharing
capacity than robotaxis, and private actors, like mobility service providers, who valued the flexibility of
robotaxis. We also identify road safety, closely followed by user acceptance, as critical factors for stake-
holders when evaluating the different deployment scenarios, as they considered them a pre-requisite that
must be at a satisfactory level before deploying a service. The two factors are also interlinked, as the
safety perception does affect acceptance (as observed in chapter 3). Alternatively, employment was not
considered a priority now in the assessment of shared automated mobility services.
Finally, the last layer of our evaluation covered real life performance of shared automated vehicles
deployed in 11 European cities in small scale pilots. We evaluate in chapter 5 the traffic, safety, and
user acceptance performance of the pilot sites, distinguished into 3 scenarios, feeder shuttles in open
environments, shuttles/buses in closed environments, and shared robotaxis, using multi-criteria analysis
method TOPSIS. The feeder shuttles in open environments scenario performed significantly worse than
the others, as more conflicts and instances of illegal overtaking were reported in sites with that scenario.
We note that no more than 1 accident was reported in any of the sites. Overall, the safety performance
of most scenarios is positive, but some potential risks, mostly in sites with more interactions with other
vehicles, arise. As for traffic efficiency, pilot sites with the shared robotaxis scenario performed the best
on average, ahead of cities with the feeder shuttles scenario. The safety and traffic efficiency ranking
of sites showed no differences and no trade off between the impact areas. As such, we note that there
might not necessarily be an improvement in safety by reducing traffic efficiency, driving at lower speeds
for instance. As for the user acceptance evaluation, we found that it was not necessarily in line with
the safety and traffic efficiency performance, as sites that ranked low in the latter two ranked highly
in the former. There was considerable variation between sites with the same scenario in terms of user
experience and acceptance, indicating that the experience was not necessarily uniform despite similarities
in the implementation scenario. Through these evaluations, the usefulness of the services and the safety
and reliability of the vehicles emerged as important factors for users, even above speed.
This thesis provides a multi-perspective view of not only user and stakeholder expectations of both
private and shared automated mobility, but also a data-driven evaluation of the current state of performance of shared automated vehicles in Europe.