Abstract
Vehicle automation is progressing rapidly and autonomous vehicles (AVs) are forecast to become a central feature of transportation systems globally. This development has the potential to result in profound changes in walking behaviors. The present study examined this issue from the perspective of relevant experts for the purpose of informing health policy. Interviews were conducted with 44 key stakeholders in Australia (n = 34), the European Union (n = 5), the UK (n = 4), and the US (n = 1). The stakeholders represented a wide range of sectors including government, AV manufacturing/servicing companies, transport policy consortiums, technology firms, insurers (public and private), trade unions, consumer representation organizations, and academia. Two potential scenarios were evident in interviewees’ discussions of the ways AVs are likely to be introduced and the implications for walking behaviors. The most beneficial scenario, but the least likely to eventuate, was considered to be the situation where people relinquish private vehicle ownership and rely on a combination of walking, public transport, and on-demand transport. The alternative scenario involved greater private AV ownership, traffic congestion, and urban sprawl, resulting in less walking activity. The convergence of the stakeholders’ views around the opposing identified scenarios highlights the need for proactive policy development to ensure the emerging transport transformation does not result in substantial increases in sedentarism.
Introduction
Autonomous vehicles (AVs) are in development and have been introduced in the form of trial applications in many countries around the world (1). AVs are forecast to bring a diverse range of health and well-being benefits across individual, social, and economic domains (2). These benefits include a reduction in injury and death resulting from crashes, greater mobility for the elderly and disabled, enhanced safety for pedestrians and cyclists, and increased leisure time (3–11). Once widely implemented, AVs are expected to be a highly cost-effective form of transport provision due to the large reduction in labor costs compared with forms of transport with paid drivers (12).
The aim of the present study was to investigate the ways in which AVs may influence walking as a form of physical activity to inform ongoing discussions about how to optimize public health via the introduction of vehicle automation. Levels of physical activity are already well below recommended levels in many countries (13), and it is therefore of importance to human health to identify whether the emergence of AVs is likely to exacerbate or alleviate this situation through impacts on walking and other forms of activity. Recent survey research suggests that once they have access to AVs, individuals will be substantially less likely to engage in various forms of active transport, including walking (14). Similarly, while modeling studies have produced varying results dependent on implementation scenario assumptions (e.g., whether autonomous public transport is made widely available), they have generally concluded that a reduction in physical activity is likely (15,16).
Much previous research examining the interaction between pedestrians and AVs has focused on the specific ways in which these entities are likely to interact with each other and the technical implications for how AVs should be programmed to anticipate pedestrians’ road-use actions (7,17–19). Little is known about how broader system changes resulting from the introduction of AVs could impact on people’s activity-related choices and behaviors (10,20). The present study represents an early investigation of this issue via consultations with a broad range of stakeholders with relevant expertise to explore the particular implications of AVs for walking-related behaviors.
Method
As part of a larger study examining the potential effects of AVs on various aspects of human life (21,22), 39 interviews were conducted with 44 stakeholders representing a wide range of sectors involved in the implementation of vehicle automation. The represented sectors included government (local, state, and federal departments responsible for health, transport, and/or infrastructure), AV manufacturing/servicing companies, transport policy consortiums, technology firms, insurers (public and private), trade unions, the law, consumer representation organizations, and academia. Ethical approval to conduct the study was obtained from the University Human Research Ethics Committee and informed consent was provided by all participants on the basis of assured anonymity in reporting.
Most of the interviews were conducted in Australia (n = 29), with international context incorporated via additional interviews undertaken in the European Union (n = 5), the UK (n = 4), and the US (n = 1). The need to access stakeholders with specialist knowledge resulted in the use of a recruitment process involving a combination of selective identification based on interviewees’ profiles and snowball sampling (23). This process involved Internet and literature searches to identify relevant individuals and asking interviewees to recommend other potential study participants. Almost all of the interviews were conducted in a one-on-one format, the exceptions being three of the Australian interviews that were conducted with multiple representatives at the request of the participating organizations.
The interviews were primarily unstructured, with each interviewee asked to discuss issues relating to AVs that were particularly relevant for their roles, organizations, and sectors. In this manner, topics relating to how human behavior is likely to change in response to the introduction of AVs emerged naturally throughout the interviews. The average interview length was around 70 min. All interviews were recorded with permission and the transcripts were provided to enable any corrections. The interview transcripts were imported into NVivo 11 qualitative data management software for progressive coding and thematic analysis (24). Coding was undertaken by a single coder (the author) due to the highly exploratory nature of the study and the resulting emergent coding approach rather than the use of a pre-established coding framework (25).
The transcripts were initially read in full, followed by an iterative coding process that involved line-by-line coding of transcript content and use of the constant comparative method to generate an interpretation of the data (26). The resulting coding hierarchy comprised deductive and inductive codes. The deductive codes were those derived from the AVs, transportation, and health literature, and included topics such as public transport, ridesharing, and aging. The inductive codes represented additional topics raised by interviewees that had not been incorporated into the initial coding hierarchy prior to commencing coding, and included topics such as trip length, working in transit, and grocery shopping.
The quality of the emergent interpretation was enhanced through the use of several strategies. These included source triangulation, which involved incorporating the perspectives of a diverse range of interviewees with differing areas of expertise (27), and providing verbatim quotes to illustrate the study findings (28). In addition, data collection and analysis continued until saturation was achieved and no new concepts emerged during coding (29). The results reported in this paper relate to the findings pertaining to the potential implications of AVs for walking as a specific form of physical activity.
Results
The interviewees described a range of potential implications resulting from the advent of AVs. It was generally recognized that the effects would be different depending on the form of roll-out that occurs. At one end of the spectrum would be the proliferation of privately owned AVs, resulting in increased congestion through the presence of empty vehicles traversing the roads on their way to or from their owners:
If we just have this idea that, ‘Great, I can send it home, it can drop me off in the city and send it home’, then it’s actually made it more congestion because it’s two trips now. It’s driving in, back, in, back, instead of going in, stop, park, and then back. We’d have all these ghost rides running around (insurance provider representative, interviewee #37).
At the other end of the spectrum is a scenario characterized by wide-scale investment in shared AVs in the form of ridesharing fleets and autonomous forms of public transport. This was described as facilitating highly cost-effective ‘mobility as a service’ (MaaS) arrangements that enable people to access affordable transport when and where they need it without owning their own vehicle. In this scenario, traffic congestion is reduced and space previously dedicated to parking is liberated for other purposes:
I’ve seen some modelling and some studies say that there’ll be probably 30 or 40 per cent less vehicles on the road once they become mainstream ... So, in terms of congestion, [if] you’ve still got the same capacity we’ve got now – if 30 or 40 per cent less vehicles, it will all flow a lot better (government, transport division, interviewee #31). As parking is less required, we’ll start shedding it and using it for other things. Turning it into wider playgrounds, putting in more street trees – actually claiming back our cities for livable human spaces (government, local council, interviewee #8).
Based on these two different implementation scenarios, the discussions indicated the possibility of two highly divergent situations, one associated with the maintenance or improvement of current walking levels and one associated with a substantial decrease. The characteristics of these two alternative realities are summarized in Table 1 and outlined below.
Potential outcomes of the introduction of AVs on walking behaviors.
AVs: autonomous vehicles.
Scenario 1: AVs promote physical activity
In this scenario, large numbers of people choose to eschew private vehicle ownership because advances in technology mean that alternative autonomous transportation options are highly affordable and convenient. This is expected to translate into increased daily activity due to people choosing to walk their shorter trips and pay for on-demand or public transport when taking longer trips. However, it was noted that highly responsive on-demand vehicle availability would be critical in determining whether this arrangement would be adequately attractive to encourage individuals to relinquish the convenience of owning their own vehicles given that active modes of transport are not always desirable or feasible:
It’s all very well if you’re walking to work and you’ve got to walk a mile or two miles and it’s a nice day. It’s not so much fun when it’s a howling gale as it is here quite often, and it’s raining, and you’re carrying a bag and wearing a suit (technology company, interviewee #25).
Increased walking activity was also seen to be a likely consequence of the anticipated large reduction in parking infrastructure required in city areas, resulting in higher density living as apartments utilize space previously allocated to parking. It was noted that inner-city living is associated with both reduced personal car ownership and greater engagement in more active forms of transport.
I don’t think personal owned cars will be something in the future. Looking at movement of the populations into city centers, then it’s already shown that there’s a movement away from car ownership. Because they don’t need it, because they can walk, bus, cycle, scoot... (AV development company, interviewee #2)
The advent of AVs was described as being conducive to safer road environments, potentially promoting greater levels of both incidental walking and walking for exercise and leisure purposes. This increased safety was attributed to a range of factors including reduced traffic levels, lower emissions, and non-susceptibility to human lapses in attention. In particular, interviewees discussed how AVs will ultimately rely on multiple forms of detection that involve both (i) on-vehicle sensors and (ii) communications with various forms of infrastructure that can convey information relating to conditions beyond the immediate detection range of the vehicle. Compared with traditional vehicles, the advanced sensing abilities of AVs will provide them with much greater ability to anticipate and avoid collisions with pedestrians and other road users:
[They will] communicate with infrastructure and utilize that data so you get a ‘look ahead’ capability. So it’s not just what the vehicle can see around it, it’s what the infrastructure is also telling it is there ... That helps a lot because you can see round corners effectively, you can tell if there’s been an accident up ahead, if there are pedestrians on the road (technology company, interviewee #25).
The ability of AVs to be constantly vigilant with 360-degree vision was seen to be especially important given the increasing distraction of pedestrians due to their use of mobile phones while walking or exercising, resulting in impaired vision and hearing:
They are like zombies ... they are just focusing on their mobile. They have their headset, so they are not seeing or hearing anything (AV manufacturer, interviewee #21).
Older people were predicted to receive substantial benefits in terms of being able to stay active due to the availability of on-demand autonomous transport. It was recognized that once people are unable to hold a driver’s license, they tend to become more house-bound with fewer opportunities for physical activity. Access to inexpensive and convenient transport options that provide door-to-door service for short distances can assist in keeping those with limited mobility up and moving:
Transport is a really big issue for [older people]. It could be simply getting people to a dining room for breakfast, as opposed to them being socially isolated in their unit because it’s too difficult to get out, to walk to wherever it is (government – health division, interviewee #32).
The final identified potential positive impact on activity levels related to freeing up time used for driving for more active pursuits. Interviewees discussed how the work commute could be integrated into one’s working hours because of the ability to be productive in transit rather than undertaking the driving task. The saved time could then be allocated to other activities that may involve walking and other forms of exercise:
I can go walking in the park with my children, and I can go and play sport, and I can go and do other things. We win the time, and with this time you can do many things (AV manufacturer, interviewee #30).
Scenario 2: AVs discourage physical activity
This scenario is characterized by high levels of personal AV ownership and/or usage of on-demand services that result in the large-scale uptake of door-to-door transport. While this situation was described above as a favorable outcome for the elderly, applied to the population in general this would constitute a disbenefit where it results in substantial decreases in incidental activity. A further dampening effect would be observed where households can send the AV out to do the shopping and other tasks, without any need to leave the home:
Just think about the daily shopping that people do for food ... You can do the booking online and this optimized, environmentally friendly vehicle will deliver this to your door (transport academic, interviewee #3).
Even among those who do wish to venture out, distrust in the ability of AVs to detect and avoid pedestrians could reduce willingness to engage in forms of incidental or purposeful activity that involve road use. Interviewees provided anecdotes of people being concerned that AV owners would program their vehicles to prioritize the driver’s safety at the expense of other road users. It was acknowledged that high levels of distrust in the community require a very cautious approach to AV introduction:
Right now, we are running at quite low speeds, under 25 kilometers an hour. It’s a choice to not scare people, to go at kind of pedestrian speed, to help them overcome their fears or their worries (AV manufacturer, interviewee #21).
Finally, the counterpoint to the possibility of increased urban density in the previous scenario is the potential for private AV ownership to result in exacerbated urban sprawl. Some interviewees described a situation where large numbers of people take advantage of the benefits of AVs to enable them to live on larger properties further away from central business districts.
What will that do for urban sprawl is a really interesting question. Because the danger is that it may re-encourage urban sprawl because things like congestion, travel times to work, and all those kinds of associated things can be less important in an autonomous vehicle (government, local council, interviewee #8).
Implications
There was agreement among the interviewees that Scenario 1 is the most optimal in terms of human welfare in general and individuals’ activity levels in particular. However, it was also acknowledged that market forces are likely to work in favor of Scenario 2:
The only way to gain anything, to have less vehicles on the road per capita, is to make it attractive and safe for people to actually use [AVs]. There is a danger that we will get it wrong. If we just leave it to the manufacturers, they’re all about selling however many units they can, and how much money they can make. They’re not worried about whether this is great for the community – the bottom line for them is about profit (insurance provider representative, interviewee #37).
This creates the situation in which strategic public policy is required to ensure potential adverse consequences are anticipated and addressed:
We need to make sure that this comes with some very clear indications of strategies of how to make sure that people maintain and increase healthy living, healthy lifestyles, cycling, walking – that we’re not basically introducing a barrier to that, unintended as it might be (transport consortium, interviewee #28).
Discussion
The views of the interviewed stakeholders in regard to the potential effects of AVs on incidental and planned walking are consistent with more general recommendations in the literature for governments to take an active and early role in AV planning to optimize health and well-being outcomes at the societal level (11,14,30). The two future scenarios discussed in the present study have divergent possible outcomes for the amount of walking that will be undertaken by individuals in their daily lives, highlighting the need for governments to engage in strategic efforts to increase the likelihood that Scenario 1 eventuates. The specific elements of these scenarios have been noted across previous work forecasting the social and health effects of AVs (e.g., 2,3,9–11,31). The contribution of the present study is the explicit consideration of the effects on walking behaviors of the combination of these elements into overall scenarios.
Working toward Scenario 1 would involve the benefits of AVs being harnessed by governments with the specific intention of encouraging and facilitating healthy lifestyle behaviors, including walking, rather than allowing commercial business models to determine the ways in which AVs are introduced (2). Specific aspects of a strategic AV implementation process are likely to include (i) road usage or congestion charges to prevent ‘ghost rides’ and encourage individuals to prioritize active and public forms of transport where feasible (14,32), (ii) clear communications disseminated to the general public to allay safety concerns that may prevent them from being willing to walk in close proximity to AVs (33), and (iii) a focus on city livability and controls on urban sprawl in city planning processes (9,30,34).
A further finding that has received less attention in previous research relates to the potential for AVs to create more time for physical activity beyond that incurred in the form of active transport. Being able to use commute time for productive work and using AVs to perform tasks such as collecting food shopping were described by interviewees as releasing time for more enjoyable forms of physical activity. While occasionally hypothesized in previous work (10), empirical research on this potentiality appears to be lacking and could represent a useful topic of future studies. Similarly, the proposition that exercise could be done in transit by fitting AVs out with appropriate equipment is worthy of greater consideration in both future research and the development of proactive policy actions (20).
The identified themes provide insights of relevance to the ripple effects model of automated driving (6). This model differentiates between more immediate and longer-term outcomes of vehicle autonomy. For example, transport choices, including those relating to walking as a form of active transport, are contained within the first ripple, while the provision of active transport infrastructure is in the second ripple and ultimate public health outcomes in the third. The themes within each scenario represent a mixture of proximal (e.g., increased home delivery services) and distal outcomes (e.g., urban sprawl vs. urban density). Overlaying the ripple model on the emergent themes provides an indication of which trends are likely to eventuate first and the resultant policy implications. Extending the active transport example, the likely delay shown in the model between the commencement of changes in active transport resulting from the introduction of AVs and the provision of related pedestrian infrastructure suggests that active transport uptake may not be optimized in the early years of the transport transformation. Proactive efforts to establish such infrastructure prior to the wide-scale availability of AVs could thus serve to enhance active transport outcomes by avoiding this time lag (14).
The primary limitations of this study are the convenience sampling approach to recruitment and the overrepresentation of Australian stakeholders. However, previous research conducted in Australia identified key AV-related issues that were noted as being consistent with those identified in the international literature (31,35), providing some assurance that the national context is aligned with global trends. This was supported by the lack of notable differences in responses according to stakeholder nationality in the present study. Study strengths include the steps taken to achieve methodological rigor: (i) achieving deliberate sample diversity through participation of representatives from a wide range of stakeholder groups from public and private sectors across multiple countries to enable triangulation across sources and saturation of concepts (i.e., sample sufficiency) (27,29); (ii) the use of extensive coding processes and the constant comparative method to glean meaning from the dataset (36); and (iii) the use of thick transcription (i.e., providing verbatim quotes to enable the interviewees’ words to be heard) (28). Finally, few studies have made physical activity the primary focus of analyses of the likely health and social outcomes associated with the introduction of AVs. By clustering the issues identified by a highly diverse sample of interviewees around the two polarized future scenarios for walking behaviors, this analysis highlights the specific policy areas that will need to be addressed to minimize the likelihood of people becoming substantially more sedentary in the AV era. Future research could focus on the potential impacts of AVs on other forms of physical activity to extend this work.
In conclusion, this study presents early insights into the future world of AVs and the implications for physical activity behaviors. The results are purely speculative given that actual outcomes cannot be known until AVs are readily available (2,6). However, the convergence of the expert stakeholders’ views around the opposing characteristics of the two identified scenarios highlights the need for careful policy development to ensure this emerging transport transformation reaches its potential for promoting walking and other forms of physical activity, as well as the more commonly acknowledged outcomes of preventing injury and death through reductions in road-related incidents.
Footnotes
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was part-funded by the Bankwest Curtin Economics Centre.
