In the coming decades, fully autonomous vehicles are set to take over the mainstream. A 2017 study released by Intel predicted that, by 2050, autonomous vehicles will represent a $7tn revenue stream worldwide. Allied Market Research projects the autonomous vehicle market to grow worldwide from $54.23bn in 2019 to $556.67bn in 2026 – suggesting that the developed world is speeding toward ubiquitous adoption of autonomous vehicles.
The signs are already there: Tesla’s driver-assistance software, Autopilot, is inching closer to true autonomy all the time; in 2019, the United Kingdom government announced its intention to have fully driverless cars on British roads by 2021; and Uber has been testing self-driving cars in the US since 2016.
Autonomous and automated vehicle (AV) technology offers compelling and innovative opportunities for individuals, business and wider society, but such opportunity is also coupled with considerable risk. Therefore, in order for AVs to develop pervasively, they must be regulated.
Given the technology’s potential to alter the fabric of society, the rate of its development has been surprisingly fast. But the scope for disruption is so total that the purported rewards of AV technology – in the best case – are set to be similarly all-encompassing, with both the public and private sectors having much to gain.
‘There’s been no precedent in history for a technology which is attractive to both sectors,’ says Richard Threlfall, partner and global head of infrastructure at KPMG.
‘Public authorities all over the world, at national and city level, can see the huge social and economic benefits of AVs. On the other hand, you also have private sector companies who can see that there’s an entire new market opening – indeed, The Economist said it was a new US$10 trillion market emerging and companies are therefore piling in at pace. It’s that twin push and pull of both public and private interests, which is why this technology is advancing so rapidly.’
For the public sphere, the case for AV technology almost makes itself. According to the World Health Organization, around 1.35 million people are killed in road accidents every year, with 90% of these accidents being caused by driver error alone. McKinsey’s 2015 report, ‘Ten ways autonomous driving could redefine the automotive world’, suggests that AVs could diminish such deaths by a mammoth 90%.
‘Eliminating all of the deaths and other accidents on roads from human error is a big prize,’ Threlfall comments.
‘AVs are expected to be much safer than non-AV vehicles,’ says Claire Bennett, general counsel of FiveAI, a leading European software company for self-driving vehicles, which has cars already being tested and demoed on London’s public streets.
‘There were about 2501 people killed or seriously injured on London streets in 2016. [Autonomous vehicles are] never going to get tired, drunk, emotional or distracted – unlike humans.’
A 2017 report by Intel, Accelerating the Future: The Economic Impact of the Emerging Passenger Economy, predicts that autonomous driving will reduce heavy traffic and alleviate congestion on world roads by decreasing more than 250 million hours of commuting time every year and turning this extra time into productivity.
This goes beyond potentially making individual road users more efficient. AVs may develop into a more efficient road transport system, for example through fleets of autonomous taxis.
‘This could lead to less private car use, particularly in cities, and therefore fewer vehicles and less congestion on the road,’ says Nicholas Paines QC, Commissioner of the Law Commission of England and Wales.
‘Congestion is a problem,’ says Bennett. ‘Some of the figures I’ve seen, UK drivers lose an average of 170 hours a year whilst driving – that’s a ridiculous amount of time. With AVs, you could use that extra time profitably. Personal car use in London takes up nearly half of the street space but only accounts for 13% of distance travelled. That’s a real waste of space. On top of that, personal car use just isn’t sustainable and that creates a series of interconnected problems that are harming different aspects of our lives. Personal driving is bad for the planet, bad for our health and bad for our pockets. If you buy a car, it will depreciate in costs. You then have insurance costs, the servicing, fuel, MOT and tax. That’s quite a significant burden.’
Sarah Owen-Vandersluis, partner and head of public sector mobility at KPMG, notes that the declining demand for personal car ownership may be informing this change: ‘This is especially true among urban millennials, and companies have picked up on this trend. OEMs, technology companies and aggregators are together conceptualising new business models which recognise that fewer people want to own cars. Instead, they simply want to get in the vehicle and go.’
The effects of new alternative methods of transportation can already be seen, with the ride-sharing apps like Uber reaching near-ubiquity in cities across the world. Driverless cars have the potential to be similarly transformative.
‘I think we’re envisioning a future where there is the ability to use your phone to say, “I want to go to this place,” and for your phone to automatically know all the different ways that you can do that, whilst taking into account your own preferences in terms of time and cost,’ says Threlfall.
‘Once you confirm, you wouldn’t need to look at a timetable, consult a map, or pay a fare, because it’ll all be on the back office of the app, in the way that we’re used to for so many other services today.’
The future envisioned by Threlfall breaks the distinction between public and private transport, making it less likely that individuals will own their own car in favour of car use on an ad-hoc basis, where the vehicle will more likely be owned by fleet companies.
‘I think that sort of model, or more use of ride-hailing-type services is going to become prevalent. Obviously, in an AV world, you don’t even need to drive the thing, you just need to get in it and it takes you to where you want to go.’
However, according to counter research outlined in the Journal of Transportation Policy, there is a risk that AVs may actually add to congestion, rather than reduce it. There is concern among policymakers around the world regarding road capacity, specifically that AVs on their streets might not be an optimal scenario.
‘It comes down to how many movements you can enable in a particular period of time in a particular space. In a mass transit mode, buses, trams and trains offer far greater capacity than private vehicles, so the worst-case outcome from a city point of view is that the adoption of AVs increases the amount of private vehicle use, making congestion in cities worse,’ Threlfall notes.
Then, and perhaps more prominently, there are the concerns surrounding the safety of such vehicles. Despite the aforementioned projections showing that the technology can be used to make the roads safer, there is some latent anxiety around getting into a car that isn’t being driven by a human being.
‘It’s the risk, or rather the lack of comfort that people have around knowing that a human is not in charge. But this technology will not be permitted to go on the road unless it’s proven to be as safe as a human or, indeed, safer. There is also a danger that we don’t harness this technology and the advantages it brings, so it’s really important that we get the appropriate validation and verification systems in place to make sure that this technology is safe, and to make sure the process that underpins approving AVs is right, so that we can utilise the full benefits of this technology,’ says Bennett.
Though fully self-driving vehicles have the capability to reduce vehicle-related deaths, this outcome is not certain. While advocates insist that enhanced safety is one of the biggest draws of the technology, incidents involving self-driving vehicles tend to be well-publicised.
In 2019, a fatal crash involving a Tesla Model X made the news for happening just 10 seconds after the driver had engaged the car’s Autopilot feature. In 2018, an human-controlled Uber test vehicle operating in self-drive mode collided with and killed a pedestrian that it had failed to detect.
‘[In Uber’s case] there was both a design flaw in the machine, but also a social flaw in the supervision,’ explains Paul Thagard PhD, cognitive scientist and professor emeritus of philosophy at the University of Waterloo, specialising in ethics and AI. ‘That is where I think Uber is probably irresponsible, because it did not only have a machine that really shouldn’t have been driving around, but it also had inadequate supervision of that machine.’
UK Law Commission
In June and October of 2019, the Law Commission of England and Wales, in collaboration with the Scottish Law Commission, published consultation papers 1 and 2, reviewing the legal framework surrounding the use of AVs and their potential use as public transport and passenger services.
‘This is the first time that we, or any other law reform agency, have been asked to design law for the future, rather than to solve problems with the law in the present,’ says Law Commissioner Nicholas Paines QC. ‘The Law Commission’s two consultation papers are part of a wider effort with the UK government, both domestically and on the international plane.’
‘We’re very much in listening mode rather than the conclusion stage at the moment in terms of our project generally – we’re putting out consultation papers and chewing over the responses. But, we’re learning a lot about the problems and framing provisional proposals for dealing with them. It’s a challenging extension of our normal work. What we are doing at the moment is analysing existing law, talking to stakeholders, getting a handle on what the problems are likely to be. And we are producing consultation papers, which are published to the world at large.’
The Law Commission’s ‘Consultation Paper 1’ focuses on three themes: safety, liability, and the need to adapt road rules for artificial intelligence.
‘Consultation Paper 2’ focuses on the regulation of what it calls ‘Highly Automated Road Passenger Services’ (HARPS), which refers to services using highly automated vehicles to deliver journeys to passengers without human intervention. The Law Commission recommends that there should be a new regulatory regime for HARPS, as opposed to a combined regime that currently applies to taxis, private hire and public service vehicles.
‘Consultation Paper 3’, which has yet to be published, aims to delve deeper into the issues identified over the course of the consultation process by formulating more detailed proposals on the way forward, leading to a final report with recommendations on all issues in 2021.
Notwithstanding other successful tests taking place in closed surroundings, companies will eventually require the deployment of AVs on public roads at scale in order to establish their level of safety, thus placing members of the general public at the centre of likely hazardous research environments.
‘You can see how quickly public opinion can get nervous about the technology, but we have to hold that in perspective against the number of people who get killed on roads by human error,’ adds Threlfall.
Rules of the road
Malini Bose, a mobility policy expert in KPMG UK’s infrastructure advisory group, points out that the potential benefits and risks of AV technology ‘Are contingent on how policy and regulation evolve. Outcomes are not predetermined. It’s only with the right policies and right regulation that the right incentives can be created.’
Indeed, as the technology is advanced almost entirely by the private sector, lawmakers around the world are having to adapt nearly in tandem, either by updating the pre-existing standards or creating a new legislative structure for when fully autonomous vehicles finally grace our roads.
Richard Threlfall oversees KPMG’s Autonomous Vehicles Readiness Index (AVRI), which ranks 25 countries on their readiness for the oncoming autonomous vehicle revolution, assessing each based on four pillars: policy and legislation; technology and innovation; infrastructure; and consumer experience.
The Netherlands ranks first, having announced a legal framework for autonomous driving in 2018 that allows AVs to be tested on public roads without drivers, and for being proactive in assessing how AVs can assist the country’s freight industry. The next best countries are Singapore, Norway, and the United States.
Despite ranking just seventh overall, the United Kingdom posted the second best score in AVRI’s ‘policy and legislation’ category, thanks to a number of initiatives taken by UK lawmakers to further the country’s AV agenda. For example, the Automated and Electric Vehicles (AEV) Act was passed into law in July 2018, establishing that liability for accident or damage rests with the insurer of the vehicle. If the AV is not insured, then the owner of the AV will be liable for any accident.
‘Westminster’s focus on Brexit last year has not been particularly helpful in passing any legislation, let alone for AVs. But, given the recent election results, we hope to have a bit more movement,’ says Bennett.
‘We’ve been trying to educate policymakers and influencers about this technology. This includes working with the Law Commission to try to make sure that the right environment is created to encourage the safe expansion of this technology. The Law Commission’s first consultation paper on AVs in 2018 refers to our “Certification of Highly Automated Vehicles for Use on UK Roads” paper and the Law Commission’s report on the outcome of this consultation paper makes multiple references to our views. As one of the leading AV tech companies in Europe working in this area, we believe we can help this process,’ says Bennett.
One specific legal question raised by AV technology is how the removal of the human element impacts the overall chain of liability when things go wrong.
Paines QC outlines how the Law Commission’s Consultation Paper 1 addresses who should have what criminal and civil liability, in a world where a car does not have a driver.
‘We analysed various civil and criminal liabilities that exist in connection with AVs and motoring. We looked at the liabilities that are imposed on the driver, because we had to rearrange those where there won’t be a driver, and made provisional proposals as to how those liabilities should be reallocated. We focused on the criminal side of things, because the AEV Act has already done a lot of work in reorganising the civil liability – basically by placing liability on the insurer who compensates the victim and can then make claims against people who can be proved to be legally responsible for an accident.’
‘On the criminal side, we looked at the range of offences that can be committed by a driver. These fall into two broad compartments. There are the ones about the way in which you drive: speed limits and shooting red lights. But there are other offences which don’t relate to a driver’s manner of driving: like the offence of driving without insurance, driving without MOT or road tax. Under the current system, the reason why the driver has those responsibilities is because they are the person that causes the car to venture onto the road. But, where you haven’t got a driver, somebody else has got to be responsible.’
‘Another new framework that we are contemplating is the ongoing safety monitoring once these vehicles are on the roads. This involves a range of measures, including data collection to better understand performance and the creation of an accident investigation branch to investigate accidents which are caused by malfunctioning AV systems, in order to learn lessons for the future.’
There is a large supposition that automated driving systems (ADSs) will proceed to advance until the user has little-to-no contribution to road safety. This transfers liability for most road accidents from the driver to the ADS vehicle manufacturing supply chain. Considering a situation where there is an accident on the road involving an AV with no user in charge, where would liability fall?
The UK’s introduction of the aforementioned world-first AV insurance legislation is distinct from that of most of the EU (and world) in that it requires that the driver be insured rather than the vehicle itself. The UK Act neatly deals with the issues surrounding AV liability by effectively placing a requirement for insurers to deal with all claims, whilst retaining both the right of recovery against manufacturers and the right to exclude liability where an individual fails to keep their AV and ADS software up to date. This allows the normal and established process of the courts to then allow the insurer to take action against any party they think can be reasonably held responsible.
‘That feels like quite an easy solution to me. Other countries have been slower to look at this issue. The UK, by looking quite carefully at this issue, will start to become the lead in terms of where other countries go on this,’ Threlfall remarks.
Where a user is only in full automation mode for part of the journey (under the classification of ‘user in charge’), they will only be responsible for the driving they do before and after the automation.
‘We think that cars which are not capable of self-driving in all circumstances should have a user in charge, explains Paines QC.
‘The plan is that they won’t be responsible for the driving whilst the car is self-driving – on a motorway, for example. They obviously will be responsible for the driving that they do before and after the motorway, and we reckon that they’re the right person to have wider responsibilities, such as ensuring that the car has its tax, insurance, and is roadworthy and so forth.’
‘With regards to a highly automated road passenger service, operators of these services would carry all the legal responsibilities.’
With the introduction of a new and revised chain of liability, the role of lawyers within the case handling is also set to change.
‘The established business that lies around driver insurance and the role of lawyers in disputes between insurance companies will almost certainly completely disappear. It will be replaced with more of a business-to-business-liability-type profession,’ comments Threlfall.
‘I think the insurance sector is quite worried about how AVs will affect the insurance industry,’ says Bose. ‘Given the sheer range of uncertainties involved, pre-determined laws by themselves are not going to be exhaustive. Lawyers are going to have to navigate through everyone from manufacturers, to insurers and to users in charge, in order to write contracts and provide the necessary legal support.’
The future that is delivered by AV technology will depend on the answers to these questions. The technology exists at a nexus between public and private concerns – safety, practicality, ethics – and both spheres have their role to play. But Thagard argues that this nexus shouldn’t be forgotten, and that a principled approach should be taken when deciding how to develop, regulate and deploy this technology.
‘The principle is “need not greed.” The danger in these cases is that you’ve obviously got commercial companies regulated by profit, which is legitimate, but there is also a grave danger that human needs will be neglected. The greed for profit will lead to companies rushing out products which are simply not ready to deal with all the physical and social conditions that driving involves.’