Speculation versus evidence: Consumers and automated vehicles
Sean Innis and Dr Ryan Young discuss the future of automated vehicles and the policy issues government will need to tackle to prepare our society for automation on our roads. This article flows from a workshop ANU hosted in collaboration with the Office of Future Transport Technology in the Department of Infrastructure, Transport, Cities and Regional Development. The views expressed are those of the authors.
In 1962, the American public was introduced to the Jetson family. The Jetsons lived in the future, 100 years in the future to be precise. 2062 is an exciting time where the technology dreams of the 1960s have become commonplace. In the opening scene, the Jetsons crowd together in their space vehicle for the journey to school and work. Ignoring some obvious stereotyping, what do we see? First, we see George in the driver’s seat, eyes front, controlling the vehicle. He looks happy and relaxed. Second, we see his children, Judy and Elroy, launch off in single-person automated space pods to continue their journey to school. And third, we see Jane “his wife” launch in her own automated pod to do some shopping.
The Jetsons provides one vision of the future for automated vehicles. In doing so, it reminds us of the difficulty in accurately predicting the path of technology and social change over time. So, with this reminder, what can we expect from automated vehicles, what are the policy issues and how should we prepare?
Automated vehicles today
In 2019, we remain 43 years away from the world of the Jetsons. Automated vehicles are not a key part of our lives but there are signs that they are coming. Vehicles entering the market today often have sophisticated driver-assist technologies, including automated braking, lane-centring and parking. Tesla Model 3 advertises itself as having full self-driving capability in the future — only the software is needed.
Public reactions to these new technologies are still playing out. Automated line centring and braking change the driving experience, and take a bit of getting used to. Some drivers like the new technology, while others do not.
Current approaches to advertising are instructive. One advert depicts a man in a ute using an automated system to parallel park in an impossible looking space to impress a mate. Another depicts a grey future where everyone drives automated vehicles, and encourages people to have fun driving while they can. Still others seek to sell the safety features of new technologies, or simply the excitement of having the latest technology. This diversity suggests no settled view exists on the value of autonomy.
Five levels of autonomy are generally recognised by industry. These start with driver-assist technologies (levels 1 and 2) and end with full automation (levels 4 and 5). Of these, level three, which requires drivers to take control of the vehicle in emergency situations, looks problematic. Reaction times and concentration dynamics make level three automation potentially less safe than existing vehicles. As a consequence, at least some car makers are seeking to move directly from driver assist to full automation, where driver control only occurs in planned circumstances.
Why automation?
Considerable excitement exists around the potential benefits of full automation. This common feature of new technology is captured by the Gartner curve — where hype about the potential of technology peaks early before falling sharply and then rising slowly again as the true value of change is revealed. Automated driving technologies look to be at a relatively early stage of this process, with past predictions about the potential and take-up of technologies underestimating the challenges involved.
While the true potential of full automation remains unproved, current thinking is that the potential benefits could be large for individuals and society in general. These benefits lie in five broad (and somewhat overlapping) areas:
they are safer: a key promise of autonomy is that it will make driving safer. Fewer accidents means fewer deaths and injuries, fewer insurance claims, and less money spent on repairs. Fewer accidents also means fewer traffic jams. Greater compliance with the road rules could also allow policing resources to be devoted to other priorities.
they are cheaper: drivers represent a major cost for freight and people-transport industries. Automated systems could improve productivity by reducing transport costs overall, and unlocking resources for other uses (of course, the impact of this process on existing workers and businesses also needs to be considered).
they provide convenience: automated vehicles may relieve us of tasks we prefer not to have or find difficult. Parallel parking, for example. Automated vehicles could also turn the drive to and from work into productive time, rather than dead time, for commuters.
they improve the efficiency of infrastructure: automated vehicles, if widely taken up, could reduce congestion and travel times by improving traffic flow. Automation, combined with other changes, could also reduce the need for parking — improving the utilisation of land.
they provide access: automated vehicles could provide mobility to people (such as the elderly or disabled) who might otherwise find it difficult to get around, improving agency and social connectedness.
Three features of the potential benefits stand out. The first is that private benefits and public ones are intertwined but are not identical. People may, for example, be willing to pay to improve their own safety, but this will not capture the full societal benefits of, say, fewer hospital visits. Second, some benefits rely on widespread take-up of automated vehicles. Delivering reduced congestion, for example, would likely require a high proportion of vehicles to be automated. And third, some benefits may not require automation. Existing ride-share and taxi services, for example, might be less preferred by elderly consumers to owning (or accessing) an automated vehicle, but achieve a similar purpose.
Take-up of automated vehicles is clearly key. Assuming the technology and potential benefits of automation are there (this shouldn’t be taken for granted), how consumers respond will largely determine whether the potential benefits are realised. This is a big unknown and will be shaped by many factors.
Moreover, the benefits of automated vehicles do not appear as clear, uncontested, and immediate as the benefits of past technologies that achieved a rapid take-up, such as the mobile phone, computers, or cars originally. This suggests a more uncertain and harder to predict future for automated vehicles.
Viewed today, a split in the community seems to exist. For some driving is a necessary evil — a chore rather than a pleasure. Access to fully-automated vehicles for these people could be very attractive. For others, driving is a pleasure that is valued highly, and may cause them to resist automation. And some might fall between these camps, valuing automation in some circumstances (driving to work) but not others (family holidays).
As the 1960s stereotypes in the Jetsons reminds us, viewing a future world through the preferences and social norms of today can be fraught. Generational changes are occurring in vehicle use — especially in cities — that may well increase the pace and level of demand for automated vehicles. But it is interesting that, in 2062, George Jetson remains firmly in control of his vehicle, hand on the wheel (or joystick in his case).
What, if anything, should governments do?
Driver-assist technologies do not fundamentally challenge existing policy or regulatory approaches. Yes, they are new. And yes, government has an ongoing responsibility to ensure they are introduced safely. But current regulatory systems seem generally well placed to cope.
Full automation, on the other hand, is a different story. Full automation — particularly if it involves networked, algorithmic decision-making supported by machine learning — represents a fundamental shift for our society. While the underlying task may not have changed, transporting people and goods over roads, the replacement of human control, and decision-making raises some very different questions for government.
At a practical level, Australian governments are already defining some elements of the pathway to full automation. Consumer choice is determining take-up, at least for driver-assist technologies. Work is taking place to make automated system producers responsible for operation of their technology. And localised trials are testing emerging technologies to understand better what changes are needed on the ground.
In considering the future of automated vehicles, government faces a chicken-and-egg problem. Widespread adoption of automated vehicles is likely to require inter-related investments in governance, regulation and (possibly) infrastructure. Yet justifying these investments requires confidence that widespread adoption will occur. Government also needs to understand and design for a complex mix of private and societal benefits.
Pre-empting technology development at this early stage would be a mistake for government, but so to would waiting with a view to muddling through as things change. The alternative is for government to engage and plan with a view to “being ready” as new technology emerges. This planning would be most effective if it focussed on the system as a whole.
Looking forward, trust will be key. Trust by consumers in the technology and the systems and regulatory frameworks supporting it will be critical to any successful transition to automated future. More broadly, citizens need to trust that their roads will become safer when automated vehicles are present and that someone will be responsible when things go wrong. In building this trust, six areas warrant at least some consideration by government as part of a planning process:
1. the need for an appropriate regulatory environment: the replacement of a human control over a vehicle will likely require significant changes to a range of regulatory frameworks.
At one end of the continuum, current road traffic rules are based on a person being in charge of a vehicle. If a traffic law is broken, the human driver is responsible. This approach holds for the driver-assist technologies but may not in a fully automated environment. At the other end of the continuum, governments will need to consider how to govern the complex ethical and practical considerations that emerge from machine-based decision-making.
Approaches to standards setting and licensing of vehicles may also need to be considered. Engagement with international processes will be important here. Automated vehicle technologies are being developed in a global context, with key design standards likely to reflect global considerations rather than national ones.
Other decisions may also be needed. One example might be whether to provide police with a ‘kill switch’ that automatically causes a vehicle to pull safely over to the side of the road. Another might be what rules should govern the use of data generated by automated systems.
2. the need for appropriate infrastructure: it is not yet clear how (or whether) road infrastructure will need to change to facilitate full automation. Technology development is currently focussed on working effectively in existing driving conditions. But capturing the full potential benefits of automation may require changes in the design and delivery of road infrastructure. Should widespread upgrades of road infrastructure be desirable or necessary, how upgrades are prioritised, sequenced and paid-for across the nation will become important.
3. the need for access to communications technology: large-scale automation requires access to high-speed mobile communication technologies. The impact of this demand on the operation and capacity of communications networks is not clear. Yet managing the interaction between communication technology and automation well could be key to the safe, effective roll-out of fully automated vehicles. Thinking now about how best to manage potential problems such as a widespread communications failure or cyber-attack may also help position Australia well for the future.
4. the need to manage social costs and opportunities from transition: the potential benefits of automation come, in part, from displacing existing (less efficient) employment and business models. This displacement has costs as well as benefits. Past industry transitions suggest that older drivers in the trucking industry, for example, may find it difficult to get alternative employment. Panel beaters and car repairers, insurances companies, and car park operators may also see existing business model disrupted.
Transition will also create new opportunities for business and individuals. New car-sharing models may emerge, and business may spring up supporting mobility for older people. A range of technology related opportunities will also likely emerge. Thinking early about the role government should play in both sides of this transition may be desirable.
5. the need to consider regional and remote access: uptake of automated technologies is unlikely to occur evenly across the country. This is particularly true if take-up requires specialised road infrastructure and access to high-speed mobile communications technology. While many paths are possible, from the vantage point of today, early take up of automated vehicles looks most likely for: (1) busy freight corridors; (2) urban locations where significant efficiency, safety, and access benefits exist. As always in Australia, access for regional and remote areas will need to be considered.
6. the need to manage changes in government revenue: automation should mean greater compliance with road and parking rules. Greater compliance is a good thing and should be welcomed. But it will also have revenue implications for some jurisdictions. More broadly, governments will need to consider how best to fund any costs needed to support the introduction of automated vehicles.
A final word
It is too early to judge the likely path and possible benefits of an automated future on our roads. But the scale and complexity of potential change warrants early investment by government in building the understanding and planning needed to be ready.
Establishment last year of the Office of Future Transport Technology in the federal transport department, together with cooperative arrangements with the states and territories, represent an important step. These bodies will need to take a broad view of their responsibilities and work well with many other agencies if Australia is to be well positioned for the future.
Sean Innis is inaugural Director of the Public Policy and Societal Impact Hub at The Australian National University.
Ryan Young is Director of the Futures Hub at the National Security College at The Australian National University.
This article originally appeared in The Mandarin on 11 November 2019.