The UK wants driverless vehicles on public roads by 2025, thanks to a £100m government plan that prioritises safety through new laws. But how likely is this? Ursula O’Sullivan-Dale reports…
The successful implementation of autonomous vehicles (AVs), or self-driving cars, offers a wealth of social and economic benefits. In August this year, the UK government announced a sudden gear shift in its autonomous vehicle agenda with the expectation that self-driving cars would be ready to hit some public roads by 2025. The initiative, dubbed the ‘self-driving revolution’, covered the creation of potentially 38,000 new jobs and an investment of £100m to deliver an ambition to improve road safety, better connect communities across the country and put Britain at the forefront of the global implementation of this technology.
However, despite the ambitious nature of the government’s plans, even when accompanied with substantial financial backing, there are major ongoing considerations before even low-level autonomy becomes the norm in the UK. Specifically, there are critical issues around the compatibility of current self-driving systems, vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2X) technology with UK roads, the viability of road infrastructure to support widespread vehicle autonomy, the awareness, training and familiarity of human drivers, as well as issues surrounding liability and insurance in the event of collisions or fatalities.
Nations that rank as some of the most prepared for AVs include the USA, Japan, China, Germany and South Korea, all of which lead in the number of global patents issued around self-driving technology, according to a report by Select Car Leasing. What’s more, the USA and China have already piloted level 4/5 autonomous robotaxis (self-driving private hire vehicles) in public areas within geofenced zones. Germany is the only European country to sit in the top five on this list, with the nation making efforts to establish the bedrock for an AV legal framework on public roads as far back as 2015. Specifically, in July last year, the Act Amending the Road Traffic Act and the Compulsory Insurance Act (the Autonomous Driving Act) was implemented, allowing road vehicles with autonomous driving capabilities to operate in specified areas of the public road network.
So where is the UK on the global playing field? Many questions hang over the future of self-driving cars in this country, chief among them being whether the government’s ambition is even achievable. However, in January this year, the Law Commission, an independent body that provides ongoing legal reviews, published a report offering recommendations to support the “safe and responsible introduction of self-driving vehicles”. Then, in April, proposed revisions to the Highway Code were laid before Parliament to provide for self-driving cars. Additionally, in June, Parliament’s Transport Committee announced that it would begin investigations into how to best bring about the deployment of AVs, with the committee holding its first evidence session in October.
While these undoubtedly represent milestones in the UK’s journey toward real-world use of AVs, they also signify limited amounts of progress in a much broader body of work needed to prepare the UK for driverless cars legally, socially, technologically and economically.
Hitting the road
One of the key bumps in the road to a successful rollout is determining the scope to which vehicles will operate autonomously, and when human takeover of a vehicle’s operation will become necessary. Kevin Vincent, director of the UK’s Centre for Connected and Autonomous Automotive Research at Coventry University, believes terminology is a key consideration here.
“The preferred term for clarity is self-driving or automated as autonomous implies a level of self-determination on the part of the machine that may not be desirable,” says Vincent. “The timing for self-driving vehicles is very realistic as the first regulations allowing passenger cars to take over the driving task without monitoring by the user in charge – Automated Lane Keeping Systems or ALKS – has been certified by Mercedes for use in Germany and vehicles with this feature are expected on UK roads very likely some time in 2023.
“In addition, early-stage commercialisation trials of self-driving vehicles providing services such as passenger shuttles and urban last-mile logistics are expected in 2024.”
Vincent points to a near-future of partial automation, with higher levels of autonomy further down the road. While uncertainty remains as to whether the UK can match the speed of some of its European counterparts when it comes to autonomy, Holly Watson-Nall, head of special operations at UK-based autonomous technology firm StreetDrone, claims AVs are already operating successfully in the UK.
She cities the UK government-funded 5G Connected Automated Logistics project (5G CAL) at Nissan’s manufacturing plant in Sunderland as an example. There, StreetDrone successfully completed a proof-of-concept that culminated in the running of autonomous deliveries between a warehouse and the Nissan factory.
“We’ve integrated all the systems to deploy autonomous and remotely driven yard shunters on a 3km route, using a private 5G network for a safe deployment of our teleoperation functions,” explains Watson-Nall, before suggesting that a broader focus on low-speed applications is needed first. “Starting with smaller, slower delivery vehicles makes it easier to achieve these goals.”
While it appears the UK is ready for limited applications of self-driving technology in controlled environments, the scope to which this can be expanded within the near future could be key to unlocking this debate. Mohammad Mousavi, professor of software engineering at King’s College London and principal investigator of the verifiability node in the UKRI Trustworthy Autonomous Systems programme, expands on the importance of not perceiving autonomous systems as binary.
He says: “Autonomy has different levels, it’s a spectrum, it’s not a case of you’re either autonomous or you’re not, you can be autonomous at different levels. In fact, there is a widely-used definition of autonomy ranging from level zero to level five.” His key suggestion on developing a roadmap for realistic deployment is to “look at the different cases for different levels of autonomy and thoroughly consider whether they make social, ethical and economic sense. For example, level four is very much deployable in confined areas and places such as harbours and warehouses, where there is not much other traffic”.
However, in places where there is interaction with other vehicles, Mousavi believes the case for full autonomy may be stronger for heavy vehicles, trucks or lorries. “Looking to 2025, motorways and major roads present a better case than very crowded urban areas. I do think there are cases where we can already deploy level four autonomy by 2025, but it doesn’t necessarily make sense in all cases.”
Solihull trials
Solihull Council trialled a self-driving shuttle in March this year on the roads around Birmingham Airport, becoming the first UK local authority to purchase its own fully electric autonomous shuttle. The initiative came as part of the council’s wider ambition to see how AVs might fit into its transport network.
Of the results of the pilot, Colin Maltby, project manager for low carbon future mobility at Solihull Council, says: “We’ve found public attitudes to using and sharing the roads with AVs as being almost entirely positive. However, our trials have had on-board safety operators, as per current government legislative requirements.
“The 2025 target is ambitious, but our experience has shown us that the technology is developing at great speed. It is our view that, by 2025, it will be ready for application within certain specific use cases – primarily ‘low speed, fixed path’ – i.e. shuttles running at around 20mph along pre-mapped routes.”
Country roads (take me home?)
The radical difference in dynamic urban settings and less predictable rural environments, which often carry the burden of underfunded and inconsistent levels of road infrastructure, will likely complicate any mainstream rollout of AVs. Lack of critical physical features, including kerbs, white line markings, necessary signage and evenly paved roads, as well as a lack of uniformity in road shape and sizing, creates a host of challenges for the sensory systems on which autonomous driving systems rely.
This adds to the already difficult task of developing a competent partial or fully autonomous system, which can struggle with the simple predictions that come easily to a human driver and requires a sophisticated onboard computer with high processing power. Currently, even the most developed versions of this technology that exist can be confused by light rain or driving in reduced visibility.
On the potential use of AVs to provide improved connectivity with rural areas, Pryce Evans, senior associate director of technology solutions at engineering company Jacobs, believes the countryside is the missing link. “Which rural networks support autonomy and/or are barriers is yet to be fully assessed and may [even] be being ignored as trophy projects on motorways or in more controlled environments suggest that we can say the UK network is ready,” says Evans. “As with any complex technical system you may have parts that support and others that don’t”.
What’s more, Miles Elsden, former chief scientist at the UK Department for Transport, and professor and director of the Institute for Safe Autonomy at the University of York, adds that, for autonomous driving to work, “you need the right level of bandwidth and very low latency. That’s not available everywhere. So, 5G promises the earth, but I’m not sure whether it will be able to deliver it and it certainly won’t be able to deliver it everywhere.”
Thus, a lack of trialling in rural areas could create a critical knowledge gap in how we understand AVs operating in such environments. However, Sarah Gates, head of public policy at autonomous driving technology start-up Wayve, says: “As AV technology increases and its deployment becomes more widespread, we should expect transport in rural communities to become more accessible. The benefits of this are potentially huge in areas where there is less access to public transport and conventional ride-hailing is not cost effective for providers.”
Stay in your lane
One of the key questions around the proposed AV rollout is whether such vehicles should occupy their own space, or if overall road safety will be improved by their integration with other motor transport. One proposed system for the management of these vehicle-to-vehicle relationships is an ALKS, which Vincent mentioned earlier. With ALKS, he says, “AVs will share space with non-AVs and only work when the conditions – road type, weather, etc – allow it according to the regulation”. According to Vincent, should the conditions fall outside of what is permitted there is a protocol for returning control to the operator or in the worst case, bringing the vehicle to a safe stop.
However, Matthew Avery, director of research at vehicle technology research centre Thatcham, believes it’s very doubtful there will be any automated driving systems on UK roads by next year. “When it does come into the market, maybe in 2024 or 2025, those systems are going to be limited in use but capable in regular traffic,” says Avery. “AVs won’t need specific lanes, just a button on the vehicle, the same as normal cars [which enables the driver] to take eyes off the road and hands off the wheel. There will be nothing different about AVs and 99% of the time you will drive as you do today, with a steering wheel and pedals.”
On the matter of keeping to one lane, Evans doesn’t believe self-driving cars should necessarily be “prevented from using all lanes” and that “the restriction of the nearside lane to being AV-only may support the ability for this lane to operate with an increased level of safety.” He continues that the freedom of movement between AVs and non-AVs “may operate assuming any potential collision might be mitigated by technology sensing potential stopped vehicles [and that] there has to be an optimum mix of human and automated”. Evans likewise wonders whether this will lead to a new ‘automated road rage’, as wholly compliant and rule-based automated driving could potentially frustrate human drivers.
What are the levels of autonomous driving in driverless cars?
Some cars are more self-sufficient than others. To help clarify the different types of self-driving vehicle, American standards organisation SAE International has published a classification system identifying six distinct levels of autonomy:Level 0: No driving automation
All driving tasks are performed by the driver.
Level 1: Driver assist
Most driving tasks are still controlled by the driver, but one aspect of driving – such as cruise control or lane assistance – can be performed automatically.
Level 2: Partial automation
The car can take over some aspects of driving, but the driver still needs to be engaged with their surroundings. Partial self-driving features include automatic emergency braking and lane assistance.
Level 3: Conditional automation
The car can cover nearly all aspects of driving itself, but only in certain environments, for example, on motorways and for very limited periods. The driver must take back control if needed.
Level 4: High automation
The car drives automatically without needing input from the driver, although there must still be a human on board. It can operate this way in most environments, but not all.
Level 5: Full automation
The car doesn’t need any involvement from the driver. It’s able to complete all driving tasks, under any conditions, and understand all scenarios, including traffic jams. The car wouldn’t need a steering wheel or pedals.
Not ready to level up?
A root of many of the anticipated problems with a widespread AV rollout comes back to the perception and understanding of the general public. While, of course, the competency and awareness of human drivers is key, this can only be reached with more extensive and open debate and education. Elsden calls for better engagement with the public, adding “with a lot of these things, it’s chicken and egg. There’s only a few [AVs] out there – people haven’t really seen them or been exposed to them. So, they don’t really know what the questions to ask are. And until there’s more penetration into the market, and people have more exposure, they won’t be able to think from an informed position – there’s a lot of hype out there.”
Part of this exposure and awareness will likewise benefit the ability of drivers to conduct safe and knowledgeable operation of self-driving cars. On the potential issue of human drivers needing to be re-trained, Evans says: “I believe a re-education would be required that focuses on the driver’s responsibility in driving a vehicle that may be under human or automated control. Evidence from the updates to Highway Code recently illustrates that – without a mandated training – drivers after passing the test will not be compelled to relearn and revert to applying experience and adapting.
“Can we allow this process to be done without control on the open road network and be based on interpretation? There needs to be some sort of shift, at least in part, whether this be a licence or addendum as a new class of vehicle to an existing licence.”
The blame game
A key consideration of the government’s proposed agenda is also the issue of accountability, with the Department for Transport planning to introduce a regulatory framework that will make manufacturers responsible for collisions that occur when cars are self-driving.
Elsden agrees with this. “As with all these things, black and white is relatively easy. If the person is in control and driving, they’re responsible. If the fully autonomous systems are in control and driving, you can see that the people who developed and designed and delivered that have a duty to make sure its safe.
“It’s that bit in the middle, the transition [from manual to autonomous drive], where things become more difficult. Who’s supposed to be in control and who has legal liability at certain points in the handover process?”
On the critical importance of data transparency in helping attribute blame and accountability, Avery adds that it is essential to make the data of crash incidents freely available, and that “the vehicle or the manufacturer driving must be held accountable and should compensate the insurer [in the event of an accident]. If that does not happen, then there will be a lot of reticence on behalf of insurance companies to cover these vehicles, meaning adoption will be much slower and the cost of insuring an AV will be much higher than expected”, which could present a range of major barriers to government plans.
So, while the UK’s strong history of technological innovation offers a great starting point for the development of an AV revolution in Britain, the journey to a self-driving road network isn’t as straightforward as it seems. Multiple legislative, technical, social and bureaucratic obstacles need to be considered before we fully understand the implications of putting self-driving cars on the road with human drivers.
This article was originally published in the November 2022 issue of CiTTi Magazine. Read the article in print.
