Transport Policy Pathways for Autonomous Road Vehicles to Promote Sustainable Urban Development in the European Union: A Multicriteria Analysis

Abstract

The European Union’s policy aims for the wide-scale deployment of automated mobility by 2030, i.e., within the next programming period (2028–2034), with the deployment of autonomous road vehicles (AVs) in cities playing a key role. Researchers suggest that AV deployment will have complex impacts on urban development, which are difficult to quantify due to scarce real-life data. The present research aims to evaluate different policy pathways of AV deployment for sustainable urban development in the next EU programming period. A multicriteria analysis is conducted, combining AHP and VIKOR, with the participation of experts across Europe. Initially, the potential impacts on sustainable urban development are weighted as evaluation criteria. Then, different pathways are evaluated against these criteria, i.e., AV deployment as collective and/or private transport in specific areas and periods or in the whole Functional Urban Area (FUA) on a 24 h basis. An interesting finding is that the effect on the city’s spatial development, not thoroughly examined by literature, is highly ranked by experts. Regarding policy pathways, autonomous collective transport with 24 h service of the FUA emerged as the optimum alternative. The proposed methodology provides a tool for planners, researchers, and policy makers and a framework for an open debate with society.

1. Introduction

Urbanization has been rapidly increasing all over the world since the Industrial Revolution [1]. Nowadays, about 70% of people in the European Union (EU) live in urban areas, while the urban population is expected to further increase in the next years, with transport being a critical factor in the urbanization and urban growth rate [2]. Thus, the analysis of the interaction between transport and urban development can significantly contribute to formulating effective urban development policies within the framework of sustainability [3]. Towards this direction and identifying the potential of new technologies (such as autonomous mobility) in terms of sustainable urban development, it is important to assess the most appropriate policy pathways for the implementation of autonomous road vehicles.

Autonomous road vehicles (AVs), such as driverless cars, pods, and buses, can be defined as self-driving vehicles that are equipped with technologies that enable driving operations without requiring the driver’s intervention [4,5,6]. In the last decade, numerous autonomous driving trials have taken place in many cities worldwide, mainly along designated routes and in controlled areas, gradually progressing towards the capability of self-driving vehicles to operate under all traffic conditions and in complex urban environments. Among the projects conducted recently (in the last five years) [7,8,9,10], many pilots test autonomous public transport vehicles, such as the trials of autonomous shuttles in Gothenburg (Sweden), the SmartBus in Aalborg (Denmark), and the Driverless Shuttle Trial along a specified route in Newcastle NSW (Australia); in other projects, autonomous passenger cars are tested, as, for example, the projects: StreetWise (Leeds, UK), testing self-driving passenger cars with security drivers to provide 24 h service; Autonomoose (Waterloo, ON, Canada), with the scope of improving autonomous driving under all weather conditions; and Drive.ai (Mountain View, CA, USA), aiming to develop an on-demand self-driving car service on public roads.

According to the current estimations by policy, research, and industry stakeholders, the year 2030 is expected to be a milestone in the evolution of autonomous road vehicles. World leading industries and forerunner countries aim for the wide-scale deployment of driverless cars by 2030 [11,12]. The assumption of the availability of fully autonomous cars by the end of the decade is commonly used in relevant impact assessments [13,14,15].

Apart from technology readiness [16], the actual deployment of autonomous cars in cities depends on many factors, such as the following [17,18,19,20,21,22,23,24,25]:

• New business models and production processes for service providers, technology developers, and manufacturers.
• Integration of autonomous road vehicles to the multimodal transport system, including the interaction between vehicles of different levels of automation in traffic.
• Regulatory reforms to allow for the seamless and safe operation of driverless cars in the street.
• Impacts on social inclusiveness, mainly related to affordability and technology literacy.
• Cyber security and data governance, regarding the big data used by the autonomous system but also the data collected and generated by the autonomous system itself.
• Ethical and liability issues, mainly regarding decision making by the autonomous car especially in case of accidents and malfunctions.
• Employment implications referring to new jobs in specialized sectors, job losses in other sectors, such as professional drivers’ jobs, and upskilling potential.

It should be highlighted that the above factors are underpinned by the EU policy framework [26], which sets the year 2030 as a time horizon for coping with them and achieving the large-scale deployment of automated mobility [27]. At the urban level, the deployment of autonomous road vehicles in the next years is linked to unprecedented opportunities and unforeseen impacts. Moreover, the uncertainty of the long-term effects from the deployment of autonomous road vehicles is highlighted by literature [28,29].

These impacts expand from the direct effect on mobility conditions and road safety to the contribution in shaping the digital and green transition as well as in the spatial development of cities, which is addressed to a lesser extent by literature [18,30,31]. Thus, planners and policy makers should explore the pathways to integrate self-driving road vehicles into the sustainable urban development context, considering the above fields of potential impacts [32]. Faced with the current scarcity of real-life experience and data from the implementation of autonomous driving in cities, researchers often turn to the analysis of stakeholder opinions to analyze the uncertainties of autonomous driving [33,34,35].

Halfway through the current EU programming period (2021–2027), the scope of the current research is the preliminary assessment of transport policy pathways for the integration of autonomous road vehicles in European cities in the first period of adoption and large-scale deployment. The purpose is the selection of the optimal pathway, in terms of sustainable urban development, in order to support policy makers in view of the next EU programming period (2028–2034). The main objectives of the research are the following: To provide an overview of the current EU transport policy for the implementation of autonomous road vehicles in European cities; discuss the potential spatial impacts and possible effects from the deployment of autonomous mobility on the sustainable urban development; perform a multicriteria analysis with the participation of research and innovation experts for the evaluation of the above-mentioned effects on the sustainable development of European cities and of the transport policy pathways for the implementation of autonomous mobility regarding each possible effect.

The research adopts the above-mentioned assumption that autonomous road vehicles will be fully available for implementation in European cities by 2030. Moreover, based on the EU transport policy targets [27], it is assumed that these vehicles will have zero tailpipe emissions. It should be noted that neither the life-cycle emissions nor the energy mix to produce the vehicle’s fuel/electricity are considered in the analysis, due to the different energy transition policies in European countries and cities. Furthermore, the current research analyses passenger mobility but not last-mile freight transportation and urban logistics.

The rest of the paper is structured as follows. The next section involves the overview of the main strategic priorities of the EU transport policy for the implementation of autonomous road vehicles in urban areas. The next section refers to the synthetic analysis of possible spatial impacts from the deployment of autonomous road vehicles on sustainable urban development, also in relation to the targets of the United Nation’s Sustainable Development Goal 11. Then, the methodology of the current research’s multicriteria analysis is presented. For the needs of the current research, the Analytic Hierarchy Process (AHP) is combined with the VIKOR approach. The fourth section refers to the implementation of the methodology and the fifth section presents the discussion of results. Conclusive remarks and prospects for follow-up research are presented in the last section of the paper.

2. EU Transport Policy Priorities for Autonomous Road Vehicles

The main strategic document of the European Union’s (EU) policy framework in the current programming period is the European Green Deal, which aims to implement the United Nations’ (UN) 2030 Agenda and the corresponding Sustainable Development Goals (SDGs) [36]. The European Green Deal outlines a series of transformative policies in all sectors of the EU economy. In the transport sector, the main objective is the acceleration of the shift towards sustainable and smart mobility. For cities, “new mobility services”, such as Mobility as a Service (MaaS), and connected and automated mobility are depicted as drivers for the reduction of congestion and pollution [37]. It should be noted that in the EU policy framework, the terms: Cooperative, Connected and Automated Mobility (CCAM) and Connected and Automated Driving (CAD) are also used, referring similarly to autonomous and connected (via telecommunications) vehicles, which can guide themselves without human intervention [38,39].

In order to achieve the European Green Deal’s target for a 90% reduction in transport emissions by 2050, the EU Sustainable and Smart Mobility Strategy, which describes the current European transport policy priorities, sets a series of targets with a short-term (2030), mid-term (2035), and long-term (2050) time-horizon [27]. As already mentioned, one of the strategy’s short-term targets for the green and digital transition of European cities is the large-scale deployment of connected and automated mobility (CAM), the promotion of active transport and micromobility, and the multimodal cooperation through Mobility as a Service (MaaS). The strategy also highlights the need to rethink the goals of urban transport planning, especially for alleviating congestion in the urban nodes of the Trans-European Transport (TEN-T) Network to ensure the effective operation of the entire TEN-T and to support the EU’s mission for 100 Climate-Neutral and Smart Cities [7].

Specific reference to the priorities for the implementation of autonomous road vehicles in urban areas is made in the New EU Urban Mobility Framework [37]. The strategic framework aims to support cities towards the transition to “safe, accessible, inclusive, smart, resilient and zero emission urban mobility”. It is suggested that digitalization and automation are the tools to shape an interconnected multimodal transport system with public transport being the backbone of collective mobility in urban areas by 2030. Specifically, the following goals are set:

• Automation in rail and road public transport systems is expected to improve the quality of urban mobility services while reducing operating costs.
• Combining connected and automated transport with zero emission technologies will be one of the enablers for climate neutrality in European cities.
• Autonomous transport systems will enhance connectivity within the wider urban area and between the cities and rural areas.
• Automation will improve traffic and mobility management in congested urban areas.

Moreover, the document highlights that Connected, Cooperative and Automated Mobility (CCAM) is a competitive field for the development of research and innovation in Europe. On the other hand, the impact of automation and digitalization on urban public transport’s workers, and especially on drivers, is discussed, highlighting the need for appropriate policies for the reskilling and upskilling of the work force. Finally, the framework stresses the need to integrate CCAM into the cities’ transport policy and, namely, the Sustainable Urban Mobility Plans (SUMPs). It is worth mentioning that a recent Practitioner Briefing on the Sustainable Urban Mobility Plan Guidelines focuses on the challenges and potential approaches for developing a transport strategy that includes CCAM, faced with the absence of empirical data and appropriate planning tools and models for the detailed assessment of CCAM impacts [40].

3. Potential Spatial Impacts from AV Deployment on Sustainable Urban Development

Sustainable urban development is the main goal of the UN SDG11: “Make cities and human settlements inclusive, safe, resilient and sustainable” [41]. SDG 11 includes a topic dedicated to sustainable transport, with reference to safety, comfort, effectiveness, efficiency, and inclusiveness, as well as to the impacts of transport systems on air quality, Green House Gas (GHG) emissions, and public health. Moreover, the need to find sustainable pathways for urbanization is highlighted by the specific SDG, targeting enhanced accessibility to public and green space and connectivity throughout the urban area.

Autonomous road transport is a mobility option, which may contribute to the above priorities. By removing human driving, a decrease in accidents caused by human error and an increase in the ability of non-driving travelers to use road vehicles are expected. The use of AVs will change mobility choices and patterns, affecting traffic conditions, with an impact on the following: i. Energy consumption and emissions; ii. Accessibility opportunities, including access to public and green spaces; and iii. Location choices, with an effect on the patterns of urban development and the connectivity between urban and non-urban areas.

Inspired by the above-mentioned links between the targets of SDG11 and the features of autonomous road transport, the current research discusses the possible effects of AVs on sustainable urban development by analyzing the potential impacts on urban space and, namely, the following: 1. Accessibility to activity locations; 2. Location choices and spatial development of cities; 3. Availability of public space, and; 4. Quality of urban environment, in terms of mobility and environmental conditions [42]. It should be highlighted that these effects depend significantly on the use of autonomous road vehicles as privately owned or collective (shared and/or public) transport.

3.1. Accessibility to Activity Locations

AVs can be operated by people without a driving license and, thus, can be accessed by a greater share of potential users compared to human-driven cars [43]. This way, they can enhance the accessibility of people with disabilities, younger or older travelers, and, in general, people with reduced mobility [44]. Through improved connectivity and automation, on-demand travel solutions, following the MaaS concept, may increase accessibility for all and the full coverage of the city’s land use network [45]. Provided the appropriate mix of policies, MaaS can upgrade the mobility conditions of people who do not own a car [46].

Furthermore, the accessibility of underserved and remote urban areas may be improved by using shared, autonomous mobility solutions [47]. The enhanced driving capability of autonomous cars, enabling them to move and maneuver in narrower spaces than conventional cars, allows them to access roads of poor geometry, as in the case of some historical centers or residential areas in Europe.

However, the potential of AVs to improve accessibility for all depends on their acceptance, adoption, and trust by different social groups, which is linked to the disruptive nature of autonomous mobility and the high dependence on machine-made decisions [48,49,50,51]. Another significant parameter refers to the wider notion of social inclusiveness, not only in terms of inclusive services towards people with reduced mobility but also in terms of affordable services for people of lower income and less access to technology [52]. Moreover, the contribution of AVs to the service of people with reduced mobility relies on the ability of the city’s transport system to ensure seamless door-to-door mobility through appropriate physical and digital infrastructure, not only while in the vehicle but also from the trip origin to the AV pick-up location and from the drop-off location to the trip’s final destination.

3.2. Location Choices and Spatial Development of Cities

Self-driving cars are expected to allow all travelers to engage in other activities (social or recreational activities, work etc.), while on the move. The fact that the time spent in traffic can be “exploited” by the driver is expected to affect the value of time, especially for commuters, professional drivers, and professionals who use driving as part of their daily work (e.g., service vehicle drivers, couriers, etc.) [53]. Consequently, the availability of private and shared autonomous cars may affect travel choices leading to the increase of the duration, distance, and frequency of trips. This, in turn, is expected to affect the location choices of households and firms through the interaction of the transport and land-use systems, as described by Wegener & Fürst [54].

The impact from the changes in location choices due to autonomous mobility on spatial development is uncertain, ranging from urban sprawl to more compact city patterns [55]. In the first case, it is possible that households will be more willing to locate at a longer distance from the city center in search of lower land values and/or better living conditions [56]. The experience from the rise of the private car in the second half of the 20th century shows that the capability for more flexible door-to-door mobility favored urban sprawl [57].

On the other hand, automation would enable the development of completely new, smarter, and more user-centric MaaS solutions [58], which may shift the urban mobility paradigm and, thus, affect the form of urban development towards more sustainable patterns. Newman and Kenworthy [59] analyze the relationship between the evolution of transport networks and the development of cities from the ancient and medieval “walking city” of mixed land use and compact development to the “transit city” of the industrialization era, which developed linearly along public transport routes with sub-centers around main terminals and, finally, to the post-World War II “automobile city” with suburbanization and land use zoning as its main features. If this approach is projected to the “MaaS city” of the digital era, it can be assumed that the city’s development will depend less on the obvious choice of the “dominant” mode and more on the individual travel choices of people with different backgrounds and behaviors, having access to a variety of mobility solutions and customized transport services dynamically addressing their mobility needs. Autonomous mobility may play a key role in this evolution.

3.3. Availability of Public Space

A result from the autonomous road vehicles’ uniform travel behavior and enhanced driving capabilities is the need for less roadway space compared to human-driven vehicles to navigate in traffic or to conduct parking maneuvers [60,61,62,63,64,65]. Furthermore, AVs can search for refueling/charging stations and parking spaces at a distance from the trip’s destination. The spatial reorganization of parking and refueling/charging infrastructure can have a significant impact on the availability and use of space in dense city centers, recreational and touristic areas, health, or educational facilities, etc. [66]. Studies suggest that the deployment of AVs can lead to the decrease in parking demand in city centers by 67–90%, depending on the share of AVs as privately owned and shared [15]. Thus, it can be assumed that the wide-scale deployment of AVs, especially in the form of collective transport, will free up public space, which can be used for the development of sustainable mobility infrastructure or other facilities, e.g., green infrastructure [65,67]. This opportunity can be linked to the current urban planning trends for redesigning the city at the “human” scale, such as the “15 min city” in Paris, the “Superblocks” in Barcelona, the “5 min to everything” in Copenhagen, and the “20 min neighborhood” in Portland as well as in Melbourne [68,69].

In a few recent studies, the use of dedicated lanes is examined as a design scenario for the introduction of AVs in the urban road network [70,71,72,73]. The effectiveness (in terms of the use of roadway capacity and the provided level of service) from introducing a dedicated lane for AVs along a street previously used by mixed traffic, i.e., both autonomous and conventional vehicles, depends on the total traffic flow and the share and type of AVs [74]. This will consequently affect the ability to free up public space, as described above.

The mass deployment of autonomous mobility will affect the form of urban space, as physical and digital infrastructure for traffic, parking, and refueling/charging should be appropriately adjusted to ensure safe and seamless autonomous operations [75]. Moreover, the implementation of CCAM, using electric and shared AVs in MaaS systems, as suggested by the aforementioned EU policy targets, will require the integration of the transport, energy, and telecommunication networks in the context of Smart City infrastructure [76]. The integration into the telecommunication network is essential to achieve the exchange of big data between vehicles, people, infrastructure, and management systems (V2X connectivity), while the integration of the electric AVs and charging infrastructure to the city’s electrical network will enable demand-responsive charging and the provision of ancillary services to the grid when the vehicle is idle [77]. However, it should be noted that the capacity for smart and integrated networks depends on the specific features of technology readiness as well as on the local needs and development objectives of each city. Thus, this aspect is not included in the analysis of Section 4 and Section 5.

The challenges from the possible effects of AV deployment regarding the availability of public space, the access to facilities, and the location choices, as described in the above sections, may be further discussed in the context of urban planning theories. The deployment of AVs in cities is expected to add to the existing complexity and interconnectedness of the urban system, while it may contribute to flexible and citizen-oriented planning approaches and to the availability of data for planning in the context of smart cities. For example, the compact development and urban intensification suggested by the smart growth theory to address urban sprawl and to improve local and global sustainability advocates pedestrian and bicycle friendly neighborhoods and transit-oriented development [78]. The potential use of AVs should be closely considered in the context of such planning theories. As already mentioned, private AVs may create the conditions for further urban sprawl, while shared AVs, especially in the context of MaaS, can support similar goals if appropriately integrated into the urban planning strategy.

3.4. Mobility Conditions: Congestion and Road Safety

A main benefit that is expected from connected and autonomous mobility is the normalization of traffic flows with positive impacts on the reduction of congestion, due to more stable driving and speeding profiles [79]. Moreover, as approximately 30% of traffic in city centers is due to drivers searching for parking spaces [80,81], the aforementioned ability of AVs to park and refuel/charge away from the trip’s destination may reduce parking-related congestion [82]. According to Meyer et al. [53], the implementation of fully autonomous road vehicles can increase the capacity of the road network by up to 40%. In addition, by removing human errors from driving, which correspond to 90% of road accidents in Europe [83], one of the primary goals of autonomous road vehicles is to radically reduce road fatalities [84] as well as accident-related congestion.

The potential of autonomous mobility to reduce congestion and improve road safety depends on the penetration rate of AVs and the duration of the transition period until the complete shift to road automation. During the period of transition, vehicles of different levels of automation will coexist, while they will share the road network with active transport and micromobility modes. Researchers argue that in this period the benefits for comfort, safety, and congestion may be limited or even negative because of the different requirements and capabilities of different modes and technologies [85]. For example, as already mentioned, autonomous cars need less lane width and headway distance than conventional cars. This may have an adverse effect on the mobility conditions of both human-driven cars and active transport and micromobility modes [86]. Thus, the share of public, shared, and private AVs and the duration of the transition period are two important factors to be considered for the estimation of impacts on congestion and road safety. In addition, the integration of AVs into the existing mechanized and active transport networks will require specific traffic regulations to ensure the safety of all road users and especially the more vulnerable ones [87].

The innovative features of autonomous driving in comparison to conventional driving in the city, as described in the above sections, can be summarized as follows: i. Accessibility by a wider range of travelers; ii. Ability to engage in other activities, instead of driving; iii. Safer and more comfortable traveling with smoother acceleration/deceleration; and iv. Driverless parking and/or refueling/charging. The improved accessibility, safety, and comfort of AVs may lead to the increase in the number, frequency, and distance of daily trips [87]. Shared AVs providing demand-responsive services will probably move for longer periods and cover larger areas. AVs that will park and refuel/charge at a distance from the trip destination may alleviate parking-related congestion but lead to new trips with the parking and refueling/charging stations as their new destination. Such considerations should be reflected in the discussion of the impact of autonomous road vehicles on mobility conditions.

3.5. Environmental Impacts

The potential impacts from the deployment of AVs on mobility conditions are strongly connected with the environmental impacts in terms of emissions and energy consumption. In Europe, the “100 Climate-Neutral and Smart Cities” mission, i.e., one of the five ongoing EU missions, which combines the appropriate research, policy, and legislative actions to deliver 100 carbon-neutral cities by 2030, has allocated considerable funding for research and innovation in “co-designed smart systems and services for user-centered, shared, zero-emission mobility of people and freight in urban areas”, involving the support to the Connected, Cooperative and Automated Mobility (CCAM) and the 2ZERO (Towards zero emission road transport) partnerships [88]. The normalization of traffic flows and the decrease in congestion due to autonomous mobility is expected to decrease emissions and energy consumption [89]. Researchers also argue that Shared Autonomous Electric Vehicles (SAEVs) have the potential to boost the shift to electric mobility with significant environmental benefits [90]. However, the total energy efficiency and the environmental benefits of electric AVs, compared to vehicles with internal combustion engine (ICE), depend, on the one hand, on the penetration rate of AVs affecting their impact on traffic flows and, on the other hand, on the electric AV lifecycle emissions, the energy mix to produce electricity (Well-to-Wheel) and the availability and grid integration of charging infrastructure. For example, according to the simulation model of Patella et al. [91], which was set up and run for the city of Rome, the construction, maintenance, and end-of-life activities for battery electric autonomous road vehicles lead to GHG emissions, which are higher than for ICE vehicles. The same study estimates that a 100% electric AV penetration rate in the future would lead to a reduction of up to 60% in the total life cycle GHG emissions compared to the current scenario, where electric vehicles correspond to less than 1% of the city’s fleet composition. The above model and other studies, such as [92,93], use different methodologies and produce different results, confirming the complexity of assessing the environmental impact of AVs. However, they all depend on assumptions on the penetration rate of AVs, with higher penetration rates linked to more environmental benefits. Thus, it is important to consider the aforementioned transition period towards the higher penetration of AVs, when vehicles of various automation levels will coexist in the city’s road network. As already discussed, this coexistence may lead to local congestion, affecting energy consumption and emissions. In addition, if the wide-scale deployment of AVs leads to increased traffic volumes due to the above-described improved mobility conditions, i.e., accessibility, safety and comfort, the total energy consumption from urban mobility may be negatively affected. The increase in vehicle-kilometers is also linked to the shorter life span of each vehicle, especially in the case of shared AVs that will probably be used more intensively than private ones. Considering the effect of manufacturing, maintenance, and end-of-life processes on the life cycle impact of AVs on climate change, the life span of the vehicles may play a significant role in the overall environmental performance of the city’s vehicle fleet.

Table 1. Potential impacts per field from the deployment of AVs in cities.

Field of Impact	Short Description of Potential Impact	Type of Potential Impact
		Positive	Negative
Accessibility to activity locations	Accessibility by a wider range of social groups (depending on inclusiveness, affordability, and access to technology).	X
	On-demand access to underserved and remote areas.	X
	Ability to access areas of poor road geometry.	X
Location choices and spatial development	Enhanced MaaS solutions for planning against private car-oriented development.	X
	Urban sprawl due to the reduction in the value of time for AV users.		X
Availability of public space	Freeing up public space in congested or sensitive urban areas (depending on the increase in traffic due to induced travel demand).	X
Quality of urban environment	Improved mobility, road safety, and environmental conditions (depending on the share of AVs and the increase in traffic due to induced travel demand).	X
	Local increase in congestion, accidents, emissions, and energy consumption due to mixed (autonomous and non-autonomous) traffic in the transition period.		X
	Increased impact on climate change due to manufacturing, maintenance, and end-of-life processes (depending on technology, policy, and business model development in the related sectors).		X

summarizes the above findings from the literature review according to the field and type (positive or negative) of possible impact. The multicriteria analysis, as presented in the following sections, considers these findings, and turns to research and innovation experts to evaluate their significance for sustainable urban development and then to assess different policy pathways for the promotion of collective and/or private AV transport in European cities.

4. Materials and Methods

The scope of the current research is the analysis of different policy pathways for autonomous mobility from the perspective of sustainable urban development in Europe. According to the specific research approach, the evaluation of these pathways is linked to their potential impacts on different fields of sustainable urban development. The ability to assess these impacts by the current forecasting models face three main obstacles, i.e., the scarcity of data from the wide-scale implementation of AVs, the consequent uncertainty of impacts, and the complexity of impacts, which refers to the interactions between them and the dependence on whether the AVs are implemented as collective and/or private transport modes. Moreover, the comprehensive description of these impacts requires the analysis of both quantitative and qualitative parameters. In order to overcome these obstacles, the current forecasting models adopt various assumptions on potential changes in travel choices or on different AV penetration rates, which seldom derive from specific policies [42]. On the contrary, the present research is based on the principle that the use and penetration of autonomous road vehicles in cities will depend on wider policy objectives, expressed by strategic planning decisions, for promoting sustainable urban development. Given that the wide-scale deployment of AVs will take place by 2030, as outlined by the corresponding EU policy target, the proposed methodology aims to appropriately structure the decision problem of how to realize this target with the highest-possible contribution to the sustainable development of European cities.

Within this scope, the Multicriteria Analysis (MCA) is selected as the appropriate methodology to structure and solve the above decision problem by including all the necessary quantitative and qualitative parameters to address the complexity and uncertainty of the subject and by capitalizing on the knowledge and insight of a group of independent experts in the field of research and innovation. Another characteristic of MCA is the transparency in the appraisal of the examined alternatives, as the expert preferences are linked to the overall research objective [94].

Different MCA methods are usually combined in the literature according to the scope, approach, and resources of each case [20,95,96,97]. In the current research, the Analytic Hierarchy Process (AHP) [98] is applied to structure the hierarchy of the decision problem by analyzing its fundamental parts and reflecting the significance of each one of them. More specifically, AHP is used for extracting the weights of the criteria in relation to the goal, and the performance of the alternatives in relation to each criterion. A benefit of AHP is that it allows group decision making, with the consistency of the group’s answers being checked through the consistency ratio [72,99,100,101]. VIKOR (VIseKriterijumska Optimizacija I Kompromisno Resenje) is then applied for the overall ranking of the alternatives [102]. The main advantage of VIKOR over AHP in terms of ranking is that the former requires the meeting of the condition of “acceptable advantage”, referring to the highest group utility, as well as the condition of “acceptable stability”, referring to the lowest individual regret. Another advantage is that the difference between two successively ranked alternatives is large enough for one to be characterized as undoubtedly better than the other. The combination of AHP and VIKOR in consecutive stages of the process allows for the exploitation of the advantages of both methodologies.

The steps of the methodology are briefly presented in Figure 1.

The implementation of the methodology leads to the calculation of the $S_i$, $R_i$, and $Q_i$ values for each alternative (Step 12). These values are used to rank the alternatives in terms of preference. Namely, the most preferable alternative is characterized by the lowest $S_i$, $R_i$, and $Q_i$ values, while the least preferable one is characterized by the highest $S_i$, $R_i$, and $Q_i$ values. Moreover, the condition of acceptable advantage, i.e., the best alternative based on the above-mentioned values of maximum group utility and minimum individual regret, is checked along with the condition of acceptable stability, i.e., a large enough difference between two successively ranked alternatives in order for one to be characterized as undoubtedly better than the other.

The list of equations implementation of the AHP and VIKOR, respectively, are presented below.

4.1. Equations Used for the Application of AHP

$$\mathrm{A}=\left[\begin{array}{cccc}{a}_{11}& a_{12}& \dots & a_{1n}\\ a_{21}& a_{22}& \dots & a_{2n}\\ \dots & \dots & \dots & \dots \\ a_{n1}& a_{n2}& \dots & a_{nn}\end{array}\right],$$

(1)

where

• $a_{\mathrm{i}\mathrm{j}}={\mathrm{w}}_{\mathrm{i}}⁄{\mathrm{w}}_{\mathrm{j}}$ = element of matrix A, i.e., the relative importance of the criterion (i) over the criterion (j), or of the alternative (i) over the alternative (j), with i = 1, 2, …, n and j = 1, 2, …, n; (${\mathsf{\alpha }}_{\mathrm{i}\mathrm{j}}=1/{\mathsf{\alpha }}_{\mathrm{j}\mathrm{i}}$ and ${\mathsf{\alpha }}_{\mathrm{i}\mathrm{i}}={\mathrm{w}}_{\mathrm{i}}⁄{\mathrm{w}}_{\mathrm{i}}=1$).
• ${\mathrm{w}}_{\mathrm{i}},{\mathrm{w}}_{\mathrm{j}}=$ weight coefficients of the criteria or of the alternatives (i) and (j), respectively.

$${(\mathrm{A}-\mathsf{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}})·\mathrm{W}=0,$$

(2)

where

• $\mathrm{A}=$ pairwise comparison matrix.
• $\mathrm{W}={({\mathrm{w}}_1,{\mathrm{w}}_2,\dots ,{\mathrm{w}}_{\mathrm{n}})}^{\mathrm{T}}$ = priority vector for each hierarchy level.
• ${\mathsf{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ = principal eigenvalue of matrix $\mathrm{A}$.

  $$\mathrm{C}\mathrm{R}=\mathrm{C}\mathrm{I}⁄\mathrm{R}\mathrm{I}<0.10,$$

  (3)
• $\mathrm{C}\mathrm{I}=$ consistency index, calculated by the Equation:

  $$\mathrm{C}\mathrm{I}={\displaystyle \frac{{\mathsf{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{n}}{\mathrm{n}-1}},$$

  (4)
• $\mathrm{R}\mathrm{I}=$ random consistency index, as shown in

  Table 2. Random consistency index values for n elements.

  n	1	2	3	4	5	6	7	8
  RI	0.00	0.00	0.58	0.90	1.12	1.24	1.32	1.41

  .

In order to ensure the consistency of the pairwise comparison matrix, CR must be lower than 0.1. In case it is calculated higher than 0.1, the comparison must be repeated for the corresponding pairs that generate the inconsistency.

4.2. Equations Used for the Application of VIKOR

$$\begin{gathered} \begin{array}{cccc}{C}_1& C_2& \dots & C_n\end{array} \\ D=\begin{array}{c}{A}_1\\ A_2\\ \dots \\ A_m\end{array}\left[\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1n}\\ x_{21}& x_{22}& \dots & x_{2n}\\ \dots & \dots & \dots & \dots \\ x_{m1}& x_{m2}& \dots & x_{mn}\end{array}\right], \end{gathered}$$

(5)

where $x_{ij}=$ performance of the alternative $A_i$ regarding criterion $C_j$, with $i=1,2,\dots ,m$ and $j=1,2,\dots ,n$.

Performance values for benefit functions:

$$f_j^{*}={\mathrm{m}\mathrm{a}\mathrm{x}}_i\left(x_{ij}\right)\mathrm{and}{f}_j^{-}={\mathrm{m}\mathrm{i}\mathrm{n}}_i\left(x_{ij}\right),$$

(6)

Performance values for cost functions:

$$f_j^{*}={\mathrm{m}\mathrm{i}\mathrm{n}}_i\left(x_{ij}\right)\mathrm{and}{f}_j^{-}={\mathrm{m}\mathrm{a}\mathrm{x}}_i\left(x_{ij}\right),$$

(7)

$$S_i={\sum }_{j=1}^n\left[(f_j^{*}-x_{ij})/\left(f_j^{*}-f_j^{-}\right)\right],$$

(8)

$$R_i={\mathrm{m}\mathrm{a}\mathrm{x}}_j\left\{w_j·\left[\left(f_j^{*}-x_{ij}\right)/(f_j^{*}-f_j^{-})\right]\right\}$$

(9)

where $w_j=$ weight of each criterion, with $j=1,2,\dots ,n$.

$$Q_i=v·{\displaystyle \frac{S_i-S^{*}}{S^{-}-S^{*}}}+\left(1-v\right)·{\displaystyle \frac{R_i-R^{*}}{R^{-}-R^{*}}},$$

(10)

where $S^{\mathrm{*}}={min}_j{S}_i$, $S^{-}={max}_j{S}_i$, $R^{\mathrm{*}}={min}_j{R}_i$ and $R^{-}={max}_j{R}_i$.

5. Analysis and Results

5.1. Definition of the Decision Problem and the Overall Goal

Assuming the wide-scale availability of AVs by 2030, according to the above literature review, the reference period (2028–2034), i.e., the next EU programming period, is considered by the specific research as the period of wide-scale integration of AVs into the transport networks of European cities. Within this time horizon, the objectives of the analysis are:

• The evaluation of the significance of possible spatial effects regarding their contribution to the overall goal of sustainable urban development for the next EU programming period (2028–2034).
• The assessment of the positive contribution of specific alternative transport policy pathways regarding each possible spatial effect for the period 2028–2034.
• The assessment of the alternative transport policy pathways in relation to the overall contribution to sustainable urban development for the same period.

5.2. Definition of Criteria and Alternatives

The criteria of the analysis describe the possible spatial effects due to AV deployment in a city, which are summarized in Table 1. These effects can be either positive or negative. In this context, the selected criteria include the following:

• Effect on physical access to activities by vulnerable groups, e.g., people with impairments or reduced mobility of lower income households, etc.—PHYS AC
• Effect on the city’s spatial development, e.g., leading to urban sprawl or enhanced accessibility of remote and/or under-serviced urban areas, etc.—SP DEV
• Effect on public space availability, e.g., due to less requirements in roadway and parking capacity or additional infrastructure for autonomous transport operations, etc.—PUBL SP
• Effect on congestion and related emissions and energy consumption, e.g., due to smoother traffic flows or induced transport demand, etc.—EMIS/EN
• Effect on road safety, e.g., due to the decrease in human errors or due to the coexistence of vehicles of different autonomy levels, etc.—R SAF

The above effects from the large-scale deployment of AVs depend significantly on the way that AVs are used, i.e., as privately owned or as part of a collective (shared and/or public) transport system. Policy makers, supported by the scientific advice of research and innovation experts, can direct the deployment to ensure a positive contribution to sustainable urban development, i.e., to enhance the positive impacts and mitigate or decrease the negative ones. This can be achieved through appropriate transport policy pathways, which will then be implemented through the corresponding policy mix (infrastructure, technology, incentivization, regulatory framework, etc.). In this context, the following alternative policy pathways are examined:

• Implementation of autonomous collective transport to service specific routes and areas (e.g., congested and/or underserved) and/or specific periods (e.g., peak and/or low demand)—COL SPEC R/P
• Implementation of autonomous collective transport to service specific routes and areas and/or specific periods and incentivization for private autonomous mobility (e.g., economic incentives and/or exclusive services)—COL SPEC R/P & PR INC
• Implementation of autonomous collective transport to service the whole Functional Urban Area, FUA, (full network coverage) at all times—COL FUA ALL T
• Implementation of autonomous collective transport to service the whole Functional Urban Area, FUA, at all times and incentivization for private autonomous mobility—COL FUA ALL T & PR INC

It should be noted that the Functional Urban Area (FUA) consists of a city and its commuting zone. Based on the definitions by OECD [103], the FUA consists of a densely inhabited city and a less densely populated commuting zone whose labor market is highly integrated with the city.

5.3. Hierarchical Structure of the Decision Problem

The hierarchy of the decision problem is presented in Figure 2. The top level of the hierarchical structure refers to the overall goal, the middle to the related criteria, and the bottom to the alternatives.

5.4. Selection of the Group of Experts

The assessment of the pairwise comparisons of criteria and alternatives to extract the weights of criteria and the performance of the alternatives in terms of each criterion should be conducted by a group of 8–15 independent experts, as recommended in relevant literature [72,96,104]. The execution is conducted anonymously, so that the judgments are expressed independently and freely [72,104,105,106]. In the present research, 10 experts in the field of transport research and innovation with experience in autonomous road vehicles were selected. More specifically, the selected experts represent academic and research institutions from Germany, Greece, Hungary, Luxembourg, the Netherlands, Switzerland, and the UK, while the group also includes a member of the European Commission’s Directorate General (DG) for Research and Innovation. The survey complied with the General Data Protection Regulation (GDPR).

5.5. Weighting of Criteria

The above-mentioned experts participated in the survey through an appropriately designed questionnaire, supported by bilateral communication with the research team. In particular, the experts initially evaluated the significance of the given criteria, i.e., possible effects, regarding their contribution to the overall goal of sustainable urban development by executing a total of 10 criteria pairwise comparisons based on Saaty’s scale (

Table 3. Relative importance scale for the pairwise comparison of the criteria.

Importance Value	Explanation
1	Equal importance
3	Moderate importance of one over another
5	Essential or strong importance of one over another
7	Very strong importance of one over another
9	Extreme importance of one over another
2, 4, 6, 8	Intermediate values between the two adjacent judgments

). An indicative part of the pairwise comparisons is shown in

Table 4. Indicative part of criteria pairwise comparison regarding the overall goal, i.e., sustainable urban development.

The criterion on the left is more important than the one on the right (select the degree of relative importance)									The two criteria are of equal importance	The criterion on the right is more important than the one on the left (select the degree of relative importance)
Physical access to activities	9	8	7	6	5	4	3	2	1	2	3	4	5	6	7	8	9	City’s spatial development

.

The “aggregation of individual judgments” method was applied for the aggregation of the experts’ answers, using the geometric mean (GM) of the values attributed to each criterion, while the consistency check of the experts’ answers was executed using the AHP consistency ratio (CR) from Equation (3). The experts’ answers, along with the corresponding geometric mean values, are shown in Table 4. It should be noted that, if an expert selects the criterion on the left side of

Table 5. Expert judgment and geometric mean (GM) values for the criteria compared in pairs.

CRITERIA/EXPERTS	EX1	EX2	EX3	EX4	EX5	EX6	EX7	EX8	EX9	EX10	GM
PHYS AC vs. SP DEV	3	1/8	1/3	7	1/5	7	5	3	1/8	4	1.2483
PHYS AC vs. PUBL SP	6	1/7	1/3	5	3	5	3	5	1/4	2	1.6619
PHYS AC vs. EMIS/EN	7	1/7	1/5	9	5	1	7	3	1/3	1/6	1.2651
PHYS AC vs. R SAF	1	1/6	1/6	1/5	5	4	1	1/4	1/8	1/6	0.4745
SP DEV vs. PUBL SP	4	4	3	3	5	1/4	1/3	3	8	1/2	1.9308
SP DEV vs. EMIS/EN	5	6	1/5	5	3	1/7	1/2	1/3	8	1/7	1.0937
SP DEV vs. R SAF	1/3	8	1/7	1/7	7	1/5	1/5	1/5	1/8	1/7	0.3746
PUBL SP vs. EMIS/EN	4	3	1/6	5	1/3	1/4	1	1/4	7	1/7	0.8548
PUBL SP vs. R SAF	1/5	5	1/6	1/5	1/5	1	1/3	1/6	1/8	1/8	0.2994
EMIS/EN vs R SAF	1/6	6	1/3	1/7	1/5	5	1/5	1/3	1/8	1/3	0.4094
PHYS AC vs. SP DEV	3	1/8	1/3	7	1/5	7	5	3	1/8	4	1.2483
PHYS AC vs. PUBL SP	6	1/7	1/3	5	3	5	3	5	1/4	2	1.6619
PHYS AC vs. EMIS/EN	7	1/7	1/5	9	5	1	7	3	1/3	1/6	1.2651
PHYS AC vs. R SAF	1	1/6	1/6	1/5	5	4	1	1/4	1/8	1/6	0.4745
SP DEV vs. PUBL SP	4	4	3	3	5	1/4	1/3	3	8	1/2	1.9308

, the selected value is used in the analysis, while if an expert selects the criterion on the right side of the Table, the reverse of the selected value is used.

The AHP pairwise comparison matrix for the criteria (

Table 6. Pairwise comparison matrix for the five criteria.

	PHYS AC	SP DEV	PUBL SP	EMIS/EN	R SAF
PHYS AC	1	1.2483	1.6619	1.2651	0.4745
SP DEV	0.8011	1	1.9308	1.0937	0.3746
PUBL SP	0.6017	0.5179	1	0.8548	0.2994
EMIS/EN	0.7905	0.9143	1.1698	1	0.4094
R SAF	2.1074	2.6695	3.3401	2.4426	1

) is formulated by inserting the above presented geometric mean values in Equation (1).

Table 7. Normalized pairwise comparison matrix, priority vector and consistency control for the five criteria.

	PHYS AC	SP DEV	PUBL SP	EMIS/EN	R SAF	PRIORITY VECTOR (W)
PHYS AC	0.1887	0.1966	0.1826	0.1901	0.1855	0.1887
SP DEV	0.1511	0.1575	0.2121	0.1643	0.1465	0.1663
PUBL SP	0.1135	0.0816	0.1099	0.1284	0.1170	0.1101
EMIS/EN	0.1491	0.1440	0.1285	0.1502	0.1601	0.1464
R SAF	0.3976	0.4204	0.3669	0.3670	0.3909	0.3886
λmax = 5.0264    CI = 0.0066    CR = 0.0059 < 0.10 ✓

✓ refers to the last parameter (CR).

presents the normalized pairwise comparison matrix for the criteria and their corresponding weights, represented by the priority vectors from Equation (2). The results from the respective consistency control are also presented.

5.6. Evaluation of Alternatives Regarding Each Criterion

After the evaluation of criteria, the experts evaluated the given alternatives, i.e., transport policy pathways, by taking into consideration the extent of their positive contribution regarding each criterion, through the respective pairwise comparisons leading to six pairwise comparisons for each one of the five criteria. The comparisons are executed based on the scale of

Table 8. Relative preference scale for the alternatives regarding each criterion.

Importance Value	Explanation
1	No preference
3	Moderate preference relation
5	Essential or strong preference relation
7	Very strong preference relation
9	Absolute preference relation
2, 4, 6, 8	Intermediate values between the two adjacent judgments

. An indicative part of these comparisons is illustrated in

Table 9. Indicative part of pairwise comparisons between the alternatives regarding each criterion.

The alternative on the left is preferable to the one on the right (select the degree of relative preference)									No preference between the two alternatives	The alternative on the right is preferable to the one on the left (select the degree of relative preference)
Autonomous collective transport to service specific routes and areas and/or specific periods	9	8	7	6	5	4	3	2	1	2	3	4	5	6	7	8	9	Autonomous collective transport to service specific routes and areas and/or specific periods and incentivization for private autonomous mobility

.

Following the same methodological steps as for the criteria, the input data from the experts’ pairwise comparisons of alternatives per criterion, as well as the geometric mean value for each one of the five criteria were calculated. The normalized pairwise comparison matrices for the alternatives, the priority vectors, and the consistency control for each criterion are shown in

Table 10. Normalized pairwise comparison matrices, priority vectors, and consistency control for the alternatives in terms of each of the five criteria.

Criterion: PHYS AC	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC	PRIORITY VECTOR (W)
COL SPEC R/P	0.2293	0.1739	0.2259	0.2758	0.2262
COL SPEC R/P & PR INC	0.1517	0.1150	0.1234	0.0802	0.1176
COL FUA ALL T	0.4426	0.4067	0.4362	0.4317	0.4293
COL FUA ALL T & PR INC	0.1764	0.3044	0.2145	0.2123	0.2269
λmax = 4.0619   CI = 0.0206   CR = 0.0229 < 0.10 ✓
Criterion: SP DEV	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC	PRIORITY VECTOR (W)
COL SPEC R/P	0.2451	0.2642	0.2188	0.3491	0.2693
COL SPEC R/P & PR INC	0.0936	0.1009	0.1163	0.0719	0.0957
COL FUA ALL T	0.5612	0.4348	0.5011	0.4364	0.4834
COL FUA ALL T & PR INC	0.1001	0.2002	0.1638	0.1426	0.1517
λmax = 4.0751   CI = 0.0250   CR = 0.0278 < 0.10 ✓
Criterion: PUBL SP	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC	PRIORITY VECTOR (W)
COL SPEC R/P	0.1534	0.2838	0.1288	0.2339	0.2000
COL SPEC R/P & PR INC	0.0462	0.0854	0.1250	0.0557	0.0780
COL FUA ALL T	0.7031	0.4033	0.5903	0.5620	0.5646
COL FUA ALL T & PR INC	0.0974	0.2276	0.1559	0.1484	0.1573
λmax = 4.2341   CI = 0.0780   CR = 0.0867 < 0.10 ✓
Criterion: EMIS/EN	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC	PRIORITY VECTOR (W)
COL SPEC R/P	0.1710	0.2714	0.1445	0.3118	0.2247
COL SPEC R/P & PR INC	0.0539	0.0856	0.1145	0.0534	0.0768
COL FUA ALL T	0.7081	0.4473	0.5984	0.5126	0.5666
COL FUA ALL T & PR INC	0.0670	0.1957	0.1426	0.1222	0.1319
λmax = 4.2383   CI = 0.0794   CR = 0.0883 < 0.10 ✓
Criterion: R SAF	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC	PRIORITY VECTOR (W)
COL SPEC R/P	0.1567	0.1989	0.1227	0.2387	0.1793
COL SPEC R/P & PR INC	0.1025	0.1302	0.1618	0.1021	0.1242
COL FUA ALL T	0.5853	0.3687	0.4583	0.4223	0.4587
COL FUA ALL T & PR INC	0.1555	0.3022	0.2571	0.2369	0.2379
λmax = 4.1028   CI = 0.0343   CR = 0.0381 < 0.10 ✓

.

5.7. Application of VIKOR for the Overall Ranking of the Alternatives

VIKOR is applied to assess the final ranking of the alternatives. For this purpose, the priority vectors of the alternatives (Table 10), which correspond to the ranking of the alternatives for each criterion, are used as input to develop the decision matrix of

Table 11. Decision matrix for the application of VIKOR.

	PHYS AC	SP DEV	PUBL SP	EMIS/EN	R SAF
COL SPEC R/P	0.2262	0.2693	0.2000	0.2247	0.1793
COL SPEC R/P & PR INC	0.1176	0.0957	0.0780	0.0768	0.1242
COL FUA ALL T	0.4293	0.4834	0.5646	0.5666	0.4587
CPF	0.2269	0.1517	0.1573	0.1319	0.2379

, based on Equation (5). It should be noted that all the criteria in the current research are considered as benefit criteria (benefit functions), due to the way that the questionnaire was formulated, asking experts to evaluate which alternative is preferable to the other regarding each criterion.

Based on Table 11, the $f_j^{*}$ and $f_j^{-}$ values are calculated using Equations (6) and (7), respectively:

• $f_j^{*}={max}_i\left(x_{ij}\right)$  = {0.4293 0.4834 0.5646 0.5666 0.4587}
• $f_j^{-}={min}_i\left(x_{ij}\right)$ = {0.1176 0.0957 0.0780 0.0768 0.1242}

Table 12. S_i, R_i, and Q_i values and alternatives ranking.

	COL SPEC R/P	COL SPEC R/P & PR INC	COL FUA ALL T	COL FUA ALL T & PR INC
Si	0.7240	1.0000	0.0000	0.7433
Ri	0.3246	0.3886	0.0000	0.2564
Qi	0.7796	1.0000	0.0000	0.7016
RankSi	3	4	1	2
RankRi	3	4	1	2
RankQi	3	4	1	2

presents the values of $S_i$, $R_i$, and $Q_i$ for v = 0.5 (according to Equations (8)–(10), respectively), which are calculated using the criteria weights (priority vector W) of Table 10. Table 10 also presents the final ranking of alternatives, where lower values of $S_i$, $R_i$, and $Q_i$ correspond to a higher ranking.

According to Table 12, the alternative COL FUA ALL T (implementation of autonomous collective transport to service the whole Functional Urban Area at all times) has the minimum $Q_i$ and, thus, it is selected as the best alternative. As regards the condition of acceptable advantage, where $0.7796-0.000=0.7796>1/(m-1)=1/(4-1)=1/3=0.3333$, and of acceptable stability, the specific alternative is also first in rank by $S_i$ and $R_i$. Consequently, the alternative COL FUA ALL T is considered the optimum alternative.

6. Discussion

Regarding the results from the implementation of the combined AHP–VIKOR methodology on the weighting of criteria and the corresponding priority vector (Table 7), the criterion of road safety (R SAF) was undoubtedly first with a weight of 38.86%. This result is aligned with the current literature, which suggests that the elimination of human error due to autonomous driving will decrease road accidents. The physical access of vulnerable groups to activity locations (PHYS AC) scored a weight of 18.87% and was ranked second, in line with another widely reported competitive advantage of autonomous road vehicles referring to their ability to provide mobility for travelers who are unable to access or drive conventional cars. An interesting finding was that the effect on the city’s spatial development (SP DEV), ranked third with the significant weight of 16.63%, even though this aspect is not widely covered by the existing literature. The weight related to congestion, emissions, and energy consumption (EMIS/EN) was found to be lower, i.e., 14.64%. This result possibly reflects the uncertainties regarding the effects of autonomous road vehicles on mobility conditions, as discussed in Section 3. The effect on public space availability (PUBL SP) ranked last with a weight of 11.01%, indicating, however, the potential of autonomous road vehicles to contribute as one of the planning tools for shaping urban space.

Regarding the evaluation of alternatives, the group of independent experts indicated that the implementation of autonomous collective transport always servicing the whole of the Functional Urban Area (COL FUA ALL T) is the optimum transport policy pathway in terms of promoting sustainable urban development in Europe for the next programming period (2028–2034) (Table 12). The implementation of the above alternative combined with the incentivization of private autonomous mobility (e.g., through economic incentives and/or exclusive services) (COL FUA ALL T & PR INC) was ranked second, closely followed by the implementation of autonomous collective transport to service specific routes and areas (e.g., congested and/or underserved) and/or specific periods (e.g., peak and/or low demand) (COL SPEC R/P). The alternative of implementing autonomous collective transport for specific routes and areas and/or specific periods while incentivizing private autonomous mobility (COL SPEC R/P & PR INC) was ranked last.

According to the priority vectors of the alternatives (Table 10), the overall optimum alternative, i.e., COL FUA ALL T, is also characterized by the best performance in terms of each criterion with considerable difference from the second-best alternative. On the other hand, the alternative COL SPEC R/P & PR I is ranked last in terms of all criteria. The alternatives COL FUA ALL T & PR INC and COL SPEC R/P are interchangeably ranked second and third depending on the examined criterion. Specifically, the alternative COL SPEC R/P is the second-best alternative in terms of the positive contribution to the city’s spatial development, public space availability, and congestion. The alternative COL FUA ALL T & PR INC is the second-best alternative in terms of the accessibility of vulnerable groups to activity locations, closely followed by COL SPEC R/P, and in terms of road safety. However, the considerably higher weight attributed by the experts to the criterion of road safety leads to the ranking of COL FUA ALL T & PR INC as second in the overall ranking.

The present assessment provides a preliminary evaluation of pathways for the deployment of autonomous road vehicles in cities, prioritizing collective (public and shared) transport services based on AVs. From a strictly cost-based approach, there are cost factors related to the current business models that may affect the viability of this pathway, which involve the cost-efficient supervision of passengers, vehicle repairing processes, and cleaning protocols [75]. The present research also confirms that the uncertainties regarding the impacts of AVs on mobility conditions and urban development patterns have not yet been fully addressed by the scientific community. Similar conclusions are drawn by a recent participatory visioning and multicriteria analysis conducted for Manchester (UK) and Melbourne (Australia) [107]. Furthermore, the research results highlight that the deployment of AVs in the context of either collective or private transportation will significantly affect urban development patterns (towards higher availability of public space or more urban sprawl, respectively), as also illustrated by the work of Legênea et al. [108]. From a reverse perspective, Wadud and Mattioli [109] assess the total cost of ownership and use of AVs in different scenarios incorporating private or on-demand vehicles to conclude that the rural location of households may increase the possibility that automated vehicle ownership will cost less compared to shared automated mobility services.

Regarding the limitations of the present research, the main limitation is related to the constraints of the applied methodology in the maximum number of the elements (criteria and alternatives), which can be compared in pairs. This is due to the fact that the human brain is allegedly capable of efficiently comparing in pairs up to 7 ± 2 elements at a time [110]. Thus, the present research provides a preliminary assessment of pathways, without accounting for all possible combinations of autonomous transport services and city contexts. Moreover, in order to address the low maturity and disruptive potential of AV deployment, the research team engaged a sufficient number (as discussed in Section 5.4) of experts in research and innovation with significant experience in the field of autonomous mobility, representing different organizations across the EU. In the near future, as cities experience the wide-scale implementation of AVs, this preliminary analysis could be used as the basis for a survey covering the whole spectrum of stakeholders (e.g., research and innovation experts, local authorities, industry, service providers, businesses, civic society organizations, etc.) in higher numbers to analyze their different perspectives. Furthermore, it can be complemented by methods and models to assess the cost-effectiveness of the alternative pathways in the context of a comprehensive transition study [111].

7. Conclusions

The specific research focuses on the role of autonomous mobility for the sustainable development of European cities with its main innovation being the assessment of transport policy pathways for the implementation of autonomous road vehicles in view of the next EU programming period (2028–2034). By assuming the wide-scale deployment of autonomous mobility by 2030, as suggested by the literature, and by considering that the sustainable development of cities is a global goal, as indicated by the SDGs, the results of the specific analysis provide useful insights for planners and policy makers not only in Europe but also in other cities of the world. Moreover, the current research contributes towards bridging a gap in the literature by delivering a policy-oriented framework for the evaluation of alternative pathways for the implementation of autonomous road vehicles based on their potential spatial impacts on sustainable urban development.

The EU policy framework in the current programming period refers to the deployment of autonomous transport as one of the transformative policies for the green and digital transition of urban mobility. The framework also draws attention to the need to assess specific challenges for the effective implementation of autonomous mobility in terms of socio-economic and environmental sustainability. The review of international literature regarding these challenges leads to different, and sometimes controversial, results. An overall conclusion from the literature review is the uncertainty, due to the scarcity of empirical data and real-life experience, and the complexity, due to the wide range of impacts and their interactions, in the assessment of anticipated effects. A common finding from the synthetic analysis of recent research and policy documents is that the impacts of autonomous mobility on cities depend on the type of implementation, i.e., as privately owned vehicles or as collective transport modes. The present research explores different pathways that depend on the type of implementation.

The above-described uncertainty and complexity leads to the need of analyzing in an integrated way both quantitative and qualitative parameters related to both positive and negative impacts. The MCA approach, with the participation of a group of independent experts in research and innovation across Europe, was adopted by the present research to address this methodological challenge. Specifically, the combination of AHP and VIKOR is used, building on the advantages of both methodologies. The methodology provides an effective structure of the decision problem and a scientifically sound method for the evaluation of alternative transport policy pathways in relation to specific criteria for sustainable development. The main constraint of the AHP methodology refers to the number of the examined criteria and alternatives, which should be adequate for the comprehensive analysis of the problem but low enough for the experts to be able to efficiently execute the corresponding pairwise comparisons. Consequently, the present research focuses on specific types of criteria (mainly on potential spatial impacts) and the analysis of general categories of alternatives, i.e., the use of AVs as private and/or collective transport with partial or full spatial and time coverage.

The application of the methodology concludes that road safety plays a crucial role in the effective implementation of autonomous mobility in terms of sustainable urban development for the next EU programming period (2028–2034). Social inclusion, expressed in terms of the accessibility of vulnerable users, as well as congestion are also regarded as important. It should be noted that the above issues are often mentioned by literature as the main advantages of autonomous mobility. The survey, however, highlighted additional aspects, which are not yet thoroughly examined by literature, such as the effects on the spatial patterns of urban development and, less so, the availability of public space. These aspects should be further explored by the scientific community.

In terms of the transport policy pathways deriving from the analysis, the implementation of autonomous mobility as a collective transport system can contribute to the cities’ spatial development, the alleviation of congestion, and the enhancement of public space if implemented in specific areas and routes and at specific time periods. To unlock the full potential of autonomous mobility by including to the above the improvement in road safety and the accessibility of vulnerable users, it is important to develop an autonomous collective transport system with 24 h service of the whole Functional Urban Area. At the same time, the benefits for road safety and the accessibility of vulnerable users are expected to be significant through the incentivization of private autonomous mobility.

The present research accepts, as a given, the EU policy target for large-scale deployment of automated mobility by 2030. However, according to the research findings, one may question whether this target can be reached in time by all European cities. More specifically, the development and implementation of a collective transport system that combines autonomous shared and public transport solutions with the ability to cover the wider urban area at all times is a great challenge by today’s standards. It requires significant investments in digital and physical infrastructure, capacity building, and changes in governance and management, cooperation, and information exchange between private and public stakeholders, innovative business models, policies, and regulations, as well as other measures aiming to change current travel patterns and behaviors. It also requires the reconsideration of other ongoing policies referring to sustainable urban mobility and development, which should be adjusted accordingly. According to the authors’ view, the way forward for the EU policy to address this challenge involves four interconnected dimensions. Firstly, it is needed to capitalize on the long tradition and pioneering role of European cities in the promotion of sustainable urban development. Good practices and lessons learnt regarding sustainable mobility can be used as a guide to ensure the effective deployment of autonomous road vehicles. Secondly, technological innovation should be led by policy goals, such as the vision for zero road fatalities, the global goal for decarbonization, and the right of all citizens to have access to affordable and effective mobility services. Thirdly, policy measures should consider the different contexts, needs, and goals of each region, city, and social group. Lastly, flexible planning strategies and monitoring mechanisms for evidence-based policy making should be designed to account for the uncertainty of the long-term impacts of autonomous mobility on cities. Overall, the EU transport policy for the next programming period should regard the implementation of autonomous road vehicles as a tool for transformative change in sustainable urban mobility.

In conclusion, this specific research provides a coherent evaluation of transport policy pathways for the anticipated wide-scale deployment of autonomous mobility for passenger transport in European cities. In this way, it enhances the preparedness of planners, researchers, and policy makers to exploit the potential of the specific technology towards sustainable urban development. The main findings can be communicated to the public in terms of an open debate to increase awareness and to incorporate the users’ opinion in a bottom-up future research approach. To fully cover the subject, follow-up research is planned by the authors, which will address the autonomous mobility solutions for urban logistics. Moreover, there is a potential to expand the current methodology to cover the gap from the evaluation of transport policy pathways at the EU level to their adaptation at the local level, taking into consideration each city’s different degree of technological readiness, as well as its specific spatial characteristics and urban development goals. Another important aspect for follow-up research is the analysis and evaluation of implementation scenarios for the proposed transport policy pathways, taking into account the official instruments, policy tools, and jurisdictions in EU transport policy, as well as the regional and local differences. Finally, the present research defines specific criteria for sustainable urban development, which can be used as categories for the design of a set of indicators that could be used by cities in parallel to the implementation of autonomous road vehicles with the purpose of monitoring and assessing their impact.