The past decade has seen significant hype and investment in autonomous vehicles. There is no doubt that technology has advanced considerably in this time. We are now at a point where there are very large bets being placed on the future of transport as being a driverless one. Now is also the time for the hard problems to be considered: what advances in technology, infrastructure, regulation and social policy will be required to pave the way for autonomous vehicles.
KWM recently hosted two events on Future Transport: Changing Gears (in Sydney on 28 August 2018, and in Melbourne on 30 August 2018), which debated how society might change following the introduction of driverless vehicles and how far away adoption is in Australia. Panellists included Paul Retter (Chief Executive, National Transport Commission), Professor Michael Milford (Leading Robotics Professor, Queensland University of Technology) and Jonathan Spear (Executive Director & General Counsel, Infrastructure Victoria), along with KWM partners Cheng Lim and Adrian Perkins.
Watch panellists from our recent Future Transport event discuss autonomous vehicles –regulation, AI, technology and when we might see these vehicles on our roads.
Some of the key takeaways from our panellists include the following:
Current regulations do not support automated vehicles and reform will be required in order to facilitate their adoption. The National Transport Commission has been tasked with the development of a regulatory framework to enable the safe operation of driverless cars.
- Existing transport laws assume that there is a human driver – this is a major obstacle for regulatory reform. In highly automated vehicles, humans will not technically be driving the vehicle for at least periods of their journey. Legislative change is therefore required to ensure that autonomous vehicles can legally operate in Australia, and it is an open question as to whether we should pass brand new laws or amend existing laws to account for autonomous vehicles (e.g. by changing the legal definition of driver, as has occurred in Tennessee, where "automated driving system" has been added to the legal definition of driver for liability purposes).
- At present, there are no regulations in place imposing safety requirements for autonomous cars, and there are no international standards or design rules that apply to autonomous cars, though the National Transport Commission is currently developing a safety assurance regime. Safety standards will be important in promoting public confidence in autonomous vehicles, particularly as driverless technology raises novel issues such as responsibility for deaths or property damage caused by the autonomous vehicles. Moreover, as Australia no longer manufactures cars, any safety standards or regulations will need to be consistent with international developments and imposed as an import requirement.
- We will need to think carefully about the liability regime that will apply to autonomous vehicles. As vehicles become more autonomous, there is likely to be a shift of responsibility from driver to vehicle – will the liable party be the driver, owner, manufacturer, or automation system provider? Is it appropriate to have a fault based liability regime or a no fault regime? Will there be a clear delineation of responsibilities of car owners (whether individuals or fleets) to ensure that their vehicles are always running the latest software updates? Should owners even have the option not to update their vehicle software? How will insurance regimes respond?
Rollout of Autonomous Vehicles
- There are two principal models for the large scale adoption of autonomous vehicles – fleet based or privately owned. A fleet based model will involve companies owning a fleet of vehicles that provide transport services to passengers (effectively a "mobility as a service" offering). There are a number of major new players clearly interested in this opportunity, including ride-sharing companies. In addition, traditional car manufacturers are also developing autonomous vehicles for personal ownership. It remains to be seen which model prevails, or whether a combination of the models will persist over a long transition period.
- A further uncertainty is where autonomous vehicles will be deployed. The weight of current investment in autonomous vehicle development indicates that the private sector will be focussed on large scale deployment of autonomous vehicles in the centre of affluent cities. This is more practical and commercial for a number of reasons. Confining the geographic rollout gives autonomous vehicle providers an opportunity to complement their autonomous technology with sensors placed in cities. This means that the technology itself would not need to be perfect in order to commence operation. Deployment in cities also provides access to a large population that is already accustomed to using taxi and ride sharing services.
- Absent regulatory intervention, or government policy direction, our panellists considered it unlikely, for technological and commercial reasons, that autonomous vehicles will ever become universal or widespread in country areas with lower population densities – yet, in terms of reducing deaths on our roads, those are the precise areas where autonomous vehicles could significantly reduce road accidents.
- There is an open question as to whether government should play a role in "picking a winner" in terms of the rollout models to incentivise the adoption of one model over the other, particularly given the broader benefits to society in lowering traffic density, reducing road tolls and the opportunity to rethink city planning and infrastructure usage.
- There will be a continuing need to accommodate the mixed usage of conventional and driverless vehicles on our roads and that raises several issues for infrastructure planners and regulators alike.
Autonomous vehicles also promise to provide more efficient travel, but this will require building adequate infrastructure for these benefits to be realised.
- Network infrastructure is a key concern – if autonomous vehicles require a constant connection to a wide network (rather than doing all processing independently and locally), then 90% of the Australian landmass will be unsuitable for them due to lack of network coverage. However, all the major players involved in autonomous vehicles are focussed on developing technologies which enable the vehicles to operate safely without a network connection.
- Autonomous vehicles are better suited to roads that have strong and consistent lines as lane keeping may be disengaged when the line markings are changed. In terms of signage, the current technology often has difficulty reading electronic speed signs. However, it is possible that technology will advance to the point where substantial infrastructural change is not required in these areas.
- Another infrastructure problem is managing congestion during peak hours. Autonomous vehicles may encourage a greater number of people to utilise vehicles to travel to and from work each day, increasing the number of cars on the roads in centralised locations. Strategies or pricing models will be required to moderate the number of vehicles at particular times, or to ensure traffic flow by other means.
- Autonomous vehicles are likely to be low emission vehicles – principally electric, hybrid or even hydrogen powered. The increased number of electric cars will have a material impact upon the electricity grid. We need to start planning for this and developing incentives to maximise efficient use of our existing infrastructure (e.g. discouraging charging during peak load times through appropriate price signals.)
- How we fund infrastructure in a future where a significant proportion of road traffic does not operate on petrol or diesel fuels will also require consideration, given fuel excises and levies will not be a means of charging users of electric and alternative fuel vehicles for their use of road infrastructure. It may be appropriate to move to a model of road usage charges and congestion charges.
- All of this demonstrates that providing the appropriate infrastructure for autonomous vehicles will require looking beyond the road and considering their holistic impact across the entire economy.
While appropriate technology, regulations and infrastructure are fundamental to the adoption of autonomous vehicles, there are also social impacts that must be considered.
- As noted above, the commercial drivers for the rollout of autonomous vehicles may not necessarily be aligned with social need. We will need to ensure that the benefits of autonomous vehicles also flow through to benefit those with the most to gain, such as those with limited access to public transport.
- Autonomous vehicles also raise important ethical considerations. Germany has been the first country to publish ethical guidelines for driverless cars. There needs to be a level of transparency around how the safety systems in autonomous vehicles are designed, and the priorities which are programmed into them. For example, do they prioritise passenger comfort and safety – and will that have an impact on how they behave in a critical situation (possibly to the detriment of those outside the vehicle)? Do they discriminate between individuals (the classic trolley problem)?
- Autonomous vehicles have the potential to dramatically reduce the number of collisions on our roads. Human error is at least a partial cause in the vast majority of motor vehicle crashes. But at the same time, while computers may outperform humans in many driving scenarios, in some unusual cases they may in fact perform far worse than a human. Compare roads to air travel – cars need constant control and kill small numbers of people frequently, whereas aircraft don't need constant control and kill large numbers of people infrequently. This makes cars harder to automate safely than an aircraft. But the benefits of being able to do so could be very significant.
How far off are we?
- Opinions from attendees varied upon when the deployment of autonomous vehicles would be a reality. Some attendees felt that the technology will be safe enough in the next 5 years to justify commercial deployment, while others felt that it would be longer.
- The interactive display outside the session demonstrated that technology is already better than human performance in some instances. Attendees were encouraged to "beat the machine" and identify whether two images of road locations were exactly the same. It is clear that the technology is very close to achieving commercial deployment. However, technological advancement will need to be aligned with shifts in regulation, infrastructure and social attitude in order for the large scale adoption of autonomous vehicles to be realised.
For further insights relating to the future economy, access the content and join the conversation at kwm.com/thefutureproject.