Pwyllgor yr Economi, Seilwaith a Sgiliau - Y Bumed Senedd
Economy, Infrastructure and Skills Committee - Fifth Senedd23/05/2018
Aelodau'r Pwyllgor a oedd yn bresennol
Committee Members in Attendance
|Adam Price AM|
|David J. Rowlands AM|
|Hefin David AM|
|Joyce Watson AM|
|Lee Waters AM|
|Mark Isherwood AM|
|Russell George AM||Cadeirydd y Pwyllgor|
|Vikki Howells AM|
Y rhai eraill a oedd yn bresennol
Others in Attendance
|Dr Paul Nieuwenhuis||Uwch-ddarlithydd mewn Logisteg a Gweithrediadau, Prifysgol Caerdydd|
|Senior Lecturer in Logistics and Operations, Cardiff University|
|Dr Wolfgang Schuster||Cyfarwyddwr technegol, Atkins, Aelod o Grŵp SNC-Lavalin|
|Technical Director, Atkins, Member of the SNC-Lavin Group|
|William Sachiti||Prif Swyddog Gweithredol, Academy of Robotics|
|Chief Executive Officer, Academy of Robotics|
Swyddogion y Senedd a oedd yn bresennol
Senedd Officials in Attendance
|Abigail Phillips||Ail Glerc|
|Robert Lloyd-Williams||Dirprwy Glerc|
Cofnodir y trafodion yn yr iaith y llefarwyd hwy ynddi yn y pwyllgor. Yn ogystal, cynhwysir trawsgrifiad o’r cyfieithu ar y pryd. Lle mae cyfranwyr wedi darparu cywiriadau i’w tystiolaeth, nodir y rheini yn y trawsgrifiad.
The proceedings are reported in the language in which they were spoken in the committee. In addition, a transcription of the simultaneous interpretation is included. Where contributors have supplied corrections to their evidence, these are noted in the transcript.
Dechreuodd y cyfarfod am 09:38.
The meeting began at 09:38.
Good morning, welcome to the Economy, Infrastructure and Skills Committee. I do apologise: we are running slightly late this morning because, as committee members, we took part in an exercise to promote active travel, so we've all been taking part in some walking, biking and coming in by different modes of transport this morning. So, we are running a little bit late as a result of that. I move to item 1. We don't have any apologies this morning. Are there any declarations of interest? No, there are none.
In that case, I move to item 2. This morning, we are one away from our last session in regard to automation and the Welsh economy. This morning, we're looking in particular at self-driving vehicles. We have a panel this morning of experts—I'm going to call you experts, I think, in your fields. I'd be very grateful if you could just introduce yourselves for the public record, if I could start from my left.
Hello. My name's William Sachiti. I am the founder of a company called the Academy of Robotics. We're a Welsh company that designs and builds autonomous vehicles. These are autonomous vehicles with a particular interest in delivering in rural areas such as rural Wales, and that is us.
Right. I'm Dr Paul Nieuwenhuis. I co-run the Centre for Automotive Industry Research in Cardiff Business School, which is—we take a social science perspective on the car industry. And I also co-run the electric vehicle centre of excellence in Cardiff University, which is a joint venture between us in the business school, the School of Psychology, because many of the barriers to the introduction of new technology are psychological, and the School of Engineering. And then I also help to co-ordinate the transport futures research network within the university. Because we've found that, as you can imagine in universities, a lot of people work away on exactly the same stuff, so we just thought we'd co-ordinate that, so we don't duplicate stuff. So, that's me.
Good morning. My name is Wolfgang Schuster, and I'm the intelligent mobility technical director at Atkins, part of the SNC-Lavalin group. As part of intelligent mobility, we focus on developing innovative solutions to make mobility easier, safer, and greener. As part of that, we're looking at new managed mobility concepts, such as mobility as a service, the use of smart technology, such as autonomous vehicles, and the smart infrastructure supporting those autonomous vehicles, and then underpinned by advanced data analytics, such as machine learning and artificial intelligence. Of course, we want to develop solutions that are safe, secure, robust and resilient, so a core focus for us is also on cyber security. In the autonomous vehicle domain, we are leading a number of projects, including two pioneering connected and autonomous vehicle projects, funded, or co-funded, by Innovate UK: VENTURER, of which I am the technical director, and FLOURISH, of which I am the chair.
Thank you. I'm very grateful for your time being with us this morning. If I could ask the first, very open, general question—not expecting you to go into any great detail, because other Members will probably ask you some more detailed questions—but to start with, what are the barriers that prevent people from using autonomous vehicles? I'm looking at Paul, perhaps.
At the moment, there are very few of them around, of course, fully autonomous. Are you familiar with the different levels of autonomy? So, fully autonomous is the so-called level 5, and that's when there is no steering wheel, as it were, so the vehicle does everything. There are very few of those available at the moment—they're experimental, aren't they? People are working with those. I think, at the moment, the industry is probably at about level 2, perhaps moving towards level 3. Some Tesla drivers believe they are at level 3, when they're not, for example. The barriers—I think perhaps you have a view on this.
Yes. I think there are obviously human factors barriers, potentially—it is about trust and acceptance. Because we need to make sure that, if we want to roll out these vehicles successfully, we need to ensure that people trust and accept these vehicles. So, there's an element of understanding the needs of the human being, and to design these vehicles around the human being. We've done a number of studies as part of VENTURER and FLOURISH towards that regard. So, for example, in VENTURER, we've looked at the trust and acceptability issue, both from the perspectives of individuals inside the vehicle, as well as from the perspective of other road users outside of the vehicle, looking at typical scenarios that you could encounter and seeing how the individuals react towards that in an autonomous vehicle. There are of course also barriers associated with insurance and legislation. So, who is responsible, who is liable, in the case of an accident—is it the vehicle manufacturer, is it the software provider, is it the individual inside the vehicle who didn't press the right button, or is it the telecoms provider because there was an interruption in the service of the download to the vehicle at the time?
I should also say—. Let me just briefly distinguish between different types of vehicles. I know we're talking about autonomous vehicles, but there are also connected vehicles, there are autonomous vehicles, and connected and autonomous vehicles, which are, essentially, three different types. Connected vehicles we already have today, and vehicles become increasingly connected. So, the simplest form of connection would be through a navigation system, through GPS, but then you have other connectivity, such as eCall, and all sorts of different services that can be provided to vehicles. Autonomous vehicles are vehicles that can drive just by themselves, without any reliance on external input except for their own sensors on board. Then you have the combination of the two, which is, essentially, where we are going, because that's where the real, the ultimate, benefit will lie. Now, in terms of the challenges associated with that at that level, it is really about having the right connected infrastructure in place first of all, because we need, obviously, digital connectivity for that. And then it is also about creating the right—to ensure that the technologies that we are rolling out are safe and secure.
The term 'automated vehicle' has been proposed as a sort of blanket term for the whole lot by people at the University of California, Davis. They recently launched a book, which I've recommended in my small submission, and they call it three revolutions, so, they say the—
I'd just better—. Sorry—
That's fine. You ask the questions.
I'm just conscious that we've got a lot of Members that want to dive in and I've got a lot of questions following what you just said, but I'm going to hold back and let some other Members come in here. So, Vikki Howells.
Thank you, Chair. I'm interested in how Wales's road network would have to change to allow autonomous vehicles to operate.
So, we—. I think the gentlemen both touched on types of autonomous vehicles. Interestingly, our ones don't have a steering wheel. We don't even have a driver compartment, because we're completely autonomous, and we make our technology to work with the existing infrastructure. So, the roads themselves don't have to change, but what we find is a challenge for us is the speed of connectivity between other devices we need to communicate with. For example, it's a really good advantage if an autonomous car knows where other cars are in real time, so it can predict futures better or it can navigate better. It's better if it knows when the traffic light's going to change before the traffic light changes so it knows whether to speed up or slow down. So, these things—. All these—an autonomous car, rather, needs many other devices to communicate with it for it to be able to do its job better. And, so, the big one for us is connectivity—more so that the cars can communicate with other connected devices so they can perform their job better.
Yes. And I would absolutely agree with that. That is one of the key points—that we need to have the right digital connectivity in place to enable that. There's another point that may have to change and that is that connected autonomous vehicles will also be low-carbon vehicles—so, for example, electric vehicles—and there's an implication here on the road infrastructure in terms of putting in place the right charging infrastructure.
And hold the thought on connectivity, because I think that's something that we want to come back to a bit later in the session as well. Vikki.
I know the Cabinet Secretary has intimated that perhaps we would need different lanes for autonomous vehicles compared to conventional vehicles on the road. What are your views around that?
So, we've conducted some studies for the Department for Transport looking at the impact of introducing autonomous vehicles with legacy vehicles—the safety impacts and the efficiency impacts. So, essentially, what we are saying is that we are introducing—let's refer to them as—strict rule-based vehicles with human-driven vehicles that are not necessarily always following all the rules. So, by mixing these two, what we've seen is that, initially, you could see potentially a decrease in the efficiency of the overall network-wide operations because of the level of cautiousness that will have to be associated, especially initially, with the autonomous vehicles. There are also, potentially, safety impacts because if an autonomous vehicle follows the rules and a human-driven vehicle does not follow the rules and the autonomous vehicle then decides 'I have right of way' that could cause safety incidents. So, it's a little bit like when we first introduced the fuel-based vehicles with the horse and the cart—initially it got worse. There were more accidents on the road until things stabilised, until you reached a certain threshold when things become better. So, that's when you start mixing the two. Now, of course, if you keep them separate, you don't have that issue.
I think they will always be mixed. I think autonomous vehicles, automated vehicles, are an additional mode. They are different from privately owned vehicles. Privately owned vehicles will obviously go down as people will use autonomous vehicles more, but they won't disappear. If you look at the introduction of new technologies historically, old technologies don't disappear; they persist. We still ride horses, we still have sailing boats and things like that, and the same will be the case here. So, the vision that's often presented that we will all be driving around in automated pods is very unlikely. There is no historic precedent for that sort of change.
To answer your question directly on whether separated lanes for autonomous cars and normal cars is a good idea, in my opinion, yes it is, though the way we make our cars is that we make them to always work with people and not have their own dedicated lanes. Autonomous cars have not been done before, so it's good to have an opportunity to be able to try this in the real world. Though I don't think it's the best way to go, it's better that we've had that opportunity to maybe try both ways so that, together, we can find a solution that works. Because I think we'd all agree that autonomous cars are the future, but any sort of barrier to stop that happening—or let's say that the path of least resistance is a better choice, and if that option is there then we'll take it, but I think that the future definitely will rely on cars where it makes no difference, because you'll always have people on the road.
Hefin, do you want to come in here?
Yes, just on something that Wolfgang said earlier about cyber security—and it's in your paper. You say:
'Data and information must be protected from external and internal attacks that will occur.'
So, you're fairly certain that they will occur. So, with someone with malign intent, how safe would it be?
Well, that's a very interesting question. Obviously, looking back at history, as we develop technologies and we are challenged to develop more advanced technologies, there have always been people on the other side trying to interfere with these technologies. That becomes all the more relevant where we are developing fully autonomous vehicles that have no human input and no capability to control the vehicle. If you're now talking about an integrated transport system, to make it more seamless, and that is the actual evolution to which we're going, then you're connecting a lot of different stakeholders, potentially, together who don't all have the same level of security associated with their devices, so potentially then providing the weakest link to enter the system. So, for example, if you connect a smart home device with your vehicle, with your bank, with some entertainment system, it provides the potential to infiltrate autonomous vehicles through these other systems, if that level of security isn't appropriately managed across the whole system.
And is cyber security technology keeping pace with that kind of technology?
Cyber security is making progress. Is it at a level where I would say that autonomous vehicles are fully secure today? Clearly not. The thing is that we are making progress towards that with every day because we are learning, and through projects such as FLOURISH, where we are looking specifically, as one of the components, at cyber security, we're learning how we should ensure that the vehicle is absolutely secure.
Okay. Can I ask one more? So, if we say we are at level 3, which is conditional automation, including platooning and highway pilot—. You said that some would say we're nearly there. Somebody said that.
Yes, I suggested we might be.
That's the point at which—. We're at the tipping point there where you could really cause problems if the technology goes wrong.
Well, some cars have already been hacked. It is still a problem that is underestimated and misunderstood in the car industry. The ICT industry has a much better understanding of cyber security than the car industry, so a lot of cars out there are probably hackable today; in fact, some of them have been hacked. So, what we need to avoid is a scenario whereby somebody with evil intent could suddenly hack tens of thousands of cars and use them to run over people in cities or something like that, which, theoretically, would be possible. So, this is why cyber security is crucially important, and some of my colleagues specifically in computer science look at that as a major issue. You guys look at it. It is going to be a potential barrier if that can't be brought under control.
In some sense, if I could add to that, it could become a single point of failure, because, if you can infiltrate through a specific channel and you can suddenly affect thousands of vehicles, you could bring an entire city to a standstill.
I just want to add that I don't think it's going to be that much of a problem because what you find with technology is that people like us—we make autonomous cars and they work, but before we deploy these we then procure the services of the best security people to now secure them. So, it's a step-by-step process. So, I don't believe—
But there's an awful lot of in-points, aren't there—an awful lot of entry points to this?
Absolutely. I don't think it's a fair assessment to assume that because right now they're not the most secure thing in the world that they're always going to be. They are steps in the process. When we get to the point where deployment comes, believe me, a lot of companies and a lot of people will focus on this, and they will be secure, just like banks. But it's all a testing phase at the moment. This is why you'd find maybe our cars don't have the level 6 encryption as you'd expect—not yet, because we're not there yet. We'll seal it up once we're done, once it's a valid working product.
The only point I would add here is a remark made by David Friedman, who was the former head of the National Highway Traffic Safety Administration in the US. He said, 'Do remember, these products will be brought to you by the same people who brought you diesel-gate, the General Motors ignition switch disaster, which caused 112 deaths and a recall of 37 million cars, the Takata airbag disaster, which caused a recall of 36 million cars as well'. So, do remember that. It will eventually be delivered by companies whose primary priority is saving money and optimising profits.
Which is where, obviously, the likes of Government—there may be a requirement for intervention to ensure a certain level of security.
I'm detecting something of a divide between William and the two of you. [Laughter.]
When you've got a bunch of experts in—[Inaudible.] [Laughter.]
I'll bring Adam Price in.
I'm interested in what this means potentially for the car industry as traditionally conceived, because that's quite an important sector in Wales and I'm wondering whether we're seeing—. One potential scenario is, you know, a bit like mobile phones, where we used to have Ericsson and Nokia, but essentially that industry has moved to California because it's the technology firms, really, that are—. It's the technology that becomes important, particularly information and communications technology and digital. Could we see a similar shift in the car industry, basically, over to the likes of Google et cetera, based in California, which isn't a particularly positive prospect for the traditional car industry with jobs in Wales?
Yes, you're right to identify that as a potential issue. The bottom line is that the car industry ideally wouldn't want to do this. They're being driven into this direction by the ICT—by Silicon Valley specifically, and particular players in Silicon Valley. For example, if you look at companies like Uber, and Lyft, who are their competitor in the US, their current business model does not work until they get rid of drivers. So, as long as they still have human drivers, they will not be profitable. So, they're desperate to get rid of drivers. So, they're really driving a lot of this agenda. The car industry really doesn't see any need to get rid of drivers because they are their customers. So, the car industry has said, 'Well, we don't want these guys to run away with it', so they are beginning to respond—they're doing something about it. But they are lagging behind because they can see exactly what you're seeing: the car becoming commodified, which is something they've been fighting for decades to avoid, the car becoming no longer privately owned, which I suggest will not be the case. But I think demand will be reduced. Also, if we think in terms of the three revolutions that the Californians are working with, which is electrification, automation and shared use, that cuts out the privately owned car, largely. So, again, they will put up resistance against that development in various ways, although at the same time they will also be developing the technologies that allow them to keep up with it. But there is that culture clash between these two cultures: automotive and ICT.
I suppose the flip side of the question is: if that's the direction of travel, how could we, in Wales, actually become a test bed then and make sure that not all of the innovation and growth and jobs et cetera is concentrated in California?
Yes. You're the man, William.
Yes, sure. I was actually going to say, yes, you're very right. So, autonomous cars are more a computer on wheels than they are a car, and this is why you'd find the tech companies leading the charge. But on the flip side for, say, here, in Wales, we are a company that was born from Aberystwyth University in Wales. Our staff, most of them are Welsh, and we are doing the same in the UK. We're making autonomous cars. The world adapts, we adapt. So, the support that I think is needed is that companies like us, or similar to us, should be supported more so that we can continue to grow in scale and hire in the UK, because our vehicles are made from raw metal to a complete car here in the UK.
I was wondering, just in the context of it, just to dig a little deeper into that, the traditional or the current test beds that we know—Milton Keynes, for example—tend to—. You know, a grid-like new town is one kind of lab environment. But actually, in some ways, what you're doing is something very different, which is taking the topography of Wales and rural roads, et cetera, and, in some ways, a bit like Nokia did with the mobile phone—they used the fact that it was a very sparsely populated rural country to actually be a kind or innovator—it's the companies that can make this technology work not in a kind of Milton Keynes-like perfect environment, but make it work on the B roads of Ceredigion—.
That's why we don't test in Milton Keynes.
Indeed, yes. So, there's an opportunity there for Wales, because if you can do it in Wales, effectively, you can do it anywhere.
Can I answer that? I think there's an opportunity here to invest in some fundamental research associated with the core technologies, smart technologies that are associated with autonomous vehicles as well as the communication infrastructure, as well as also focusing on the wider operational environmental impacts that need to be considered as part of this development. Then, there's also something around focusing on the social and human factors questions that need to be addressed, which are specific to areas as well.
We're going to move on to skills in a moment and some questions from Joyce Watson, but I think Mark wants to come in on this section as well.
Yes, very briefly. To be successful, a motor manufacturer, like any other business, has to anticipate and meet customer demand. Many drivers enjoy driving not just for recreational purposes, but for practical purposes—going shopping, going to visit relatives, going out for the day or towing a caravan, whatever it might be. You mentioned sailing and horses, but that's primarily recreational use these days and it's not people's primary transport choice. How will we ensure that the right of people to choose to drive themselves for practical as well as recreational purposes will be protected, or is there a risk that legislators could, ultimately, deny people that freedom?
I think that's an important—. I'm a car guy myself. I rally classic cars and things, so I sympathise with the picture that you paint there. The problem is, I think, that legislators often see the car primarily as a transport mode and, as you have identified, it's a lot more than that; it's a cultural, social—. It's used for social differentiation, it's self-expression, and things like that. If we ignore those aspects, we are, from my perspective, in very dangerous water. This is why I believe that those rights will be protected at some level, and that, therefore, human-driven cars will continue to exist in some form. They will also, to some extent, be connected, they will be electric, et cetera, et cetera. But there will still be human-driven cars around.
There are other advantages that give a degree of resilience. If you think of a major natural disaster, for example, like the hurricanes in the Caribbean, which completely knocked out large chunks of infrastructure, the human-driven car can still negotiate a lot of the post-disaster environment, whereas a totally connected and automated vehicle would really struggle with that sort of completely changed and anarchic environment. So, even for the resilience of a transport system, it would still be desirable to have human-driven cars. This is not a widely shared view, incidentally.
Can I just add to that? I would agree with that. I think that connected and autonomous vehicles will be one element of that wider transport system. It is much more about the integration and optimally using each of these different modes, rather than totally replacing one of the levels. And you're right: manually driven vehicles, in some sense, could be a lot more resilient in the case of failures. So, for example, if I wanted to create gridlock in London and let's say we have only autonomous vehicles, you just need to jam GPS and the vehicles won't know where to go. So, there's an issue there that needs to be addressed. If you have manually driven vehicles, they could still manoeuvre that.
I'll just add to that by stating that I can see a future, for sure, where it's likely going to be illegal for humans to drive because cars do a much better job, even in the unexpected environment where something horrific has happened. They'll overcome that, I'm sure, in the future. I think we probably will also see a future where you'd have, similar to us having bus lanes today, lanes for human drivers only.
Wow. Thank you, Mark. I'll come to Joyce Watson then.
We're all going to move to the future, so that's going to necessitate different skills within this market, whether it's production or wherever. So, do you want to tell us what they might be—those skills—and how we might acquire them to keep up with this, or drive ahead?
Going back to my point on the development of smart technologies, which really underpin all of this new mobility ecosystem, which incidentally doesn't only include autonomous vehicles on the road, as it will include autonomous vehicles in rail, it will include autonomous aircraft, commercial and private drones and what not, so, there's an underpinning capability here that needs to be created and there's an opportunity here for Wales to create that for future generations.
So, what are the skills?
So, they are basically technology skills—core technology skills associated with automation and connectivity.
Okay, so those are the skills, but we need the people to teach the skills.
That's correct. I recently graduated from university as a mature student. I studied at Aberystwyth University artificial intelligence and robotics. As someone who'd come from the corporate world as an entrepreneur, it was quite interesting to learn how so many people have degrees, or are studying degrees rather, which won't have as much value for them in the future and they have no idea, and then here we are looking for graduates in computer vision and computer science and there are too few. We're having to look abroad for the talent because there are not enough. It's there in the UK, but it's just not enough. There are people studying things that are not going to help them at all in the future. So, perhaps some sort of push to help people to understand that this is what you need to study now because that degree in—I don't know; I don't want to name any—
Be careful. [Laughter.]
I'll be careful not to name any—might not be as advantageous in a world where everybody is looking for some sort of—. And it doesn't have to be something really technical. There are fewer technical roles within this industry that still would be very useful, as compared to other degrees that aren't very useful.
So, what about upskilling, because not everybody's going to go back and start university? Most people who we're going to employ are already there, so we have to think in those terms. So, what about tackling that?
We could do. So, if we go back to the autonomous car industry, it's going to take about 20 years for us to see autonomous cars everywhere, because this is the life cycle of a car. So, I believe we should probably focus more on just before university, because these are the people who are going to be hired. The penetration now on autonomous cars is less than 1 per cent. Imagine it's 10, 20 or 30 per cent. That's a five or six-year period from now. These are people who are still in school, who haven't done their GCSEs yet. These are the ones, we think at least, who we should focus on. Yes, I see that value for today, but we have a functional system today; people have jobs today. It's the ones in the future I'm worried about, where everything's automated.
I absolutely agree. It's really building that capability for future generations, and, obviously, beyond automation activities, there are some core skills in the sciences, technology and engineering and mathematics that will always be relevant in these fields.
I suppose, in representing a university, I should really contribute here. Clearly, the areas include, on the car side, transformation from internal combustion engine vehicles to electric vehicles. That's happening anyway, so there's a predictable skill set. The cyber security area, we've already mentioned, and that's a key area. Human factors, again, where some of my colleagues work, and how humans interface with these new technologies, those are all predictable areas. So, I think we can to some extent guide people in those directions already, and you probably—robotics is another area that is highly relevant. So, yes, it's sort of inspiring young people, kids, that these are exciting areas and that they will be inspired to follow a career in that, and don't listen to their parents who grew up in a completely different world where different things were relevant, basically. Put some money into Techniquest to put some exciting stuff in there.
Okay. I hear you, but it doesn't answer the fact that the majority of people will be looking for work who might be displaced or already there. The facts are this: you have more people of advanced years than you have of young people. So, I would like some thoughts about the existing population that you're going to find are in the workplace, because even in five years' time, the 15-year-olds now are going to be 20, the 20 are going to be 25 and so on.
I'll give you a more relevant answer to that. So, if you look at—. We make autonomous vehicles for delivery and one of the arguments is we've replaced delivery drivers. One of the jobs we're recruiting for most at the moment is people skilled in understanding an area, because these will become fleet operators—someone who knows where the cars are and they can oversee where all the cars are. As a small company, it's expensive to take someone through that training. We don't expect the Government to do it, because it's very specific to us. We don't mind training them. So, I think maybe some sort of support structure to help tech companies be able to train people, to upskill them or adapt their skills or the knowledge they already have to be able to do the same job or a very similar job but tech-related.
If there are no more questions on skills—no—I'll go to Mark Isherwood.
Thank you. We know that autonomous vehicles rely on some level of internet or mobile connectivity. We understand the UK average for data coverage on A and B roads is 45 per cent, but only 22 per cent in Wales. How do you believe this might affect our ability to develop and use autonomous vehicle technology in Wales?
If I can start, autonomous vehicles don't require connectivity. If you want a purely autonomous vehicle that can drive by itself, of course you need some navigation interface, which you have through, for example, GPS—it's a global system—but you don't require, necessarily, an interface to operate the autonomous vehicle. Now, if you want to optimise a network of autonomous vehicles, that's when you start requiring connectivity. That will be most relevant in areas such as larger cities, to evolve these vehicles as part of smart cities, which are also much more likely to have full coverage. Of course, if you want a whole area-wide, and also the more remote areas, covered to conduct better network management, then of course you will require connectivity and it will require the development of advanced digital connectivity.
Any other comments? No. Can I ask a supplement, which relates to this: were Government to sell the 5G radio spectrum, what protections, if any, need to be put in place into the tender to incentivise companies to ensure access to areas that are not commercially viable?
Can you repeat the question?
The Minister here, in our Chamber, highlighted her concern, given the precedent for selling spectrum at 4G and so on, that were UK Government to sell spectrum at 5G, there's a risk that the successful bidders, commercial companies, would not have a commercial incentive to ensure access in non-commercially viable geographical areas. How can we protect against that?
I think there will have to be in place some regulatory framework to address that issue, because there are obviously a number of commercial challenges that we will be facing in the roll-out of these kind of technologies. You're absolutely right to point that out that, whilst rolling out a digital network in cities will be commercially viable, putting in a few masts around in very remote areas, which serve only a few people, will not be commercially viable, so there will have to be some sort of regulatory framework to address that.
My response to the study—that's the thing with this industry is that none of this has been done before. People look to us for answers all the time and sometimes it's important we say, 'We're not sure.' But if we ask around or if we work together, we can figure it out. This is why studies and consultations are there, isn't it?
Okay, thank you.
So, this is something you think Ofcom, who regulates the area, perhaps needs to consider.
I don't know. That's the thing. That's why I say we should put it to a consultation and have many companies chip in and add their input and then we can figure it out. I'm very cautious of over-regulation because often the regulators, when it comes to tech, don't understand as well that tech evolves so quickly or changes so quickly, so if you regulate this and it kills the entire future of a specific route to market—. So, maybe we should take a look at it first and have a consultation of some description where many people add input.
Maybe it's regulation in consultation. [Laughter.]
Okay. I'm going to come to—
That's absolutely right.
I'll come to Lee Waters who's got some wider questions.
Thank you. What do you think this all does to our idea of what public transport is?
Sorry, say that again.
What are the implications of this for how we view public transport or what we class as public transport?
Well, public transport will be very vulnerable to this because it has the potential—. The most vulnerable areas are buses and taxis, really, for automated vehicles. So, yes, it will have to change. That's my perspective, anyway.
I think autonomous vehicles will operate alongside public transport. One of the key applications of autonomous vehicles will be for first and last mile. There's also a question here around developing the right approach and the desired outcome by engaging with the right stakeholders with the appropriate incentivisation and with the appropriate commercial models to ensure that these vehicles will be used in a way that doesn't necessarily jeopardise public transport or other means of transport.
In rural areas, of course, they could enhance the public transport function. Where it's no longer viable to run buses on a regular basis with a driver, to run a smaller autonomous vehicle without a driver may suddenly become cost-effective. So, it may well enhance a public transport function in rural areas.
Couldn't you do that in urban areas too? You mentioned earlier the sharing—couldn't it just change our conception of what buses look like?
Yes, of course. This is where the sharing is so important. Once you start the sharing element, then of course it becomes more like a public transport mode, but also at the same time more flexible than what we currently think of as public transport.
I think, to add to that, there's an opportunity here to exploit how we can maximise the efficiency of public transport. So, if we consider these smaller vehicles as part of public transport, it is well known that, obviously, public transport is not always the most efficient. So, for example, the occupancy of buses or rail at specific times of the day may be very low, so it may be much less efficient to operate a large bus rather than replacing it with an autonomous vehicle that is much smaller.
At the moment, it's not cost-effective to operate a small bus because the driver costs the same as in a large bus.
Indeed, and that's why I'm interested to hear you think it's a threat to public transport because it could be an opportunity for public transport.
I think it could be for sure, in my opinion, but maybe not right now. In the next decade or so, it might begin to be a threat. The reasons are, very simply—imagine you've got a vehicle able to take you pretty much anywhere within about an hour, and it costs nearly nothing—probably cheaper than it would for, I don't know, your local bus ride—to take you from door to door. Now, everybody's doing this at the same time, so you've got many cars on the road that are constantly on the move. Would you still get up—it's raining—go to the bus stop, get a bus, get to the train station, get a train and then get to work? I don't think so. But then, it also sees the end of car ownership as well. Why would you own a car if a car can just pick you up, take you where you need to go, and then it's gone?
Indeed, but that's my question. Does it not change our conception of what buses and public transport are? Because in that scenario you paint, buses as we know them now wouldn't remain in aspic, they'd adapt as well. Just as you have on Uber a sharing option, why wouldn't you have this for these cars as well?
You mentioned Uber—so Uber-like apps, let's call them taxi apps—most people in a big city today would rather use a taxi app than they would public transport. The reason they don't is because it's more expensive, isn't it? If it was cheaper, which is would be with an autonomous vehicle, then it's logical to assume all people would want to use them because the cost isn't an issue.
Indeed. You're making my point for me. [Laughter.]
Can I just get back to your point? You made the point that autonomous vehicles could be, potentially, a lot cheaper as an option and therefore people would be transitioning towards autonomous vehicles rather than buses, but if you put in place the right incentivisation for people and the right commercial models to ensure that that doesn't happen then you can actually ensure the right mix of vehicles is being used. Otherwise, the danger is that we go down the route of suddenly creating in some sense additional congestion, and potentially additional pollution if these vehicles weren't electric or low-carbon vehicles.
I suppose the question is: as a consumer, would you rather have a seven-minute walk to the bus stop, because the bus is not going to drive down your street, or would you rather that a small car picks you up right in front of your door and drops you off at your door at work?
Yes, I think we may be talking at cross-purposes. My argument is that these things are going to be really expensive to buy—more expensive than conventional cars—
No, not really—
Well, electric cars are more expensive than conventional cars.
Only because of the scale of economics.
Well, that's my point. So, in the initial stage, these are going to be more expensive, so sharing would be an attractive thing. So, in that scenario, you could well see what you've just described—. If several people are wanting to do a similar journey, you effectively are creating a minibus.
Yes and no. Just going back to the cost thing, one of our autonomous cars that does deliveries is cheaper than a van, and we are a small company in the UK. Scale that up and the cost is going to be significantly—. Because this is just a milk float with a computer inside. It's a computer on wheels. A car is much more complex, by the time you got the engine, the transmission, all those things. Autonomous cars are not as expensive—. The only thing that's expensive is that there are so few of them, but let's get that number up by 10 per cent and the price drops dramatically—dramatically.
Can we move on to the point you were making about the first and the final mile? We have a paper from the Open University that talks about the work that they're doing in Milton Keynes. They talk about the use of passenger pods in the last mile in pedestrianised areas of the city centre, not on public roads. So, here we have a scenario where our current conception of what is an area where people walk where cars aren't allowed is suddenly becoming turned on its head. So, these pods, potentially, could be allowed anywhere. What are the implications of that, do you think, for public space and for disabled groups and for active lifestyles?
On the subject of the pods, I think maybe this is more an experiment than a reality. So, on pavements in the UK, we have a law called invalid carriage law, which applies to anything that is a specific size from knee height up to a mobility scooter. They can't travel at more than 4 mph and must give right of way to pedestrians. These are the current laws, which aren't looking to change any time soon. I think what's happening at the moment is that, in Milton Keynes, they've classified a lot of their pavements as roads so that they could test autonomous vehicles, but nearly every autonomous car company now is only testing on these but wants to have the cars on the road. So, invalid carriage law protects us against this, hence the 4 mph and the right of way. So, the only vehicles now in the UK on pavements are the little knee-high robots that do deliveries. And they're very small. You could just walk around them or they get out of the way anyway. They're existing in several cities—I think 100 cities in the world—but it's not been a problem yet, and it won't be a problem for autonomous cars because we'll not see the cars on the pavement.
But do you see a likely consequence of all this an increase in sedentary lifestyles?
Yes, but also a decrease in rural areas. So, for example, if you live in rural Wales or somewhere where it's very hard to get out or go anywhere, if an autonomous cheap car can come and pick you up and take you to a city—a one and a half-hour journey on a road with no markings; it does it itself—. So, yes, it works both ways.
I think autonomous vehicles have the potential to enhance inclusivity, essentially, because people, such as disabled people and the elderly who can't drive, would be able to have access to vehicles. That would bring them closer to society.
They can take a taxi, of course.
They can do. They certainly can do, but, for example, if you were very independent, let's say, during your whole life, and you wanted to, let's say, own an autonomous vehicle that stands there right in front of your door, you could still do that if you were 90 when you potentially couldn't drive any more.
Or it could decrease levels of physical activity too.
It depends on the right level of incentivisation—I get back to that point. In the end, we as humans evolve as part of this mobility system that we're transitioning towards. So, yes, if not managed appropriately, there is a danger that you end up with a system where people will only and exclusively use wheeled mobility as opposed to using their feet—or to cycle.
Because currently 60 per cent of us are overweight and a quarter of adults obese.
And the car probably has a role to play in that. It's probably one of the culprits for that.
It does, absolutely, and I'm just wondering whether this development we're discussing this morning could accelerate that further.
I think we need a lot more human factors and social studies to really address that question in detail.
Okay. Thank you.
I want to start to draw our meeting to think about what Government can do, and what you think Government should do. So, I'll come to David Rowlands.
I think we're coming now, really, to the raison d'être of us making these inquiries. What follows on from this is that we make recommendations to the Government. So, given that the Welsh Government says it's committed to advanced technologies, where would be the best areas for the Government to give support to help to facilitate the development of autonomous vehicles? And following on from that, how can we create an enabling environment for autonomous vehicles?
My view is that Government, first and foremost, should play the role of an enabler, facilitator, co-ordinator, to develop a strategy and a framework for collaboration and information sharing, and to also enable the development of a capability base in STEM skills, as well as, specifically, smart technologies in focusing on issues related to multimodal network management, which is going to be key, because autonomous vehicles will be one element of this wider integrated transportation system.
There's also a role to invest in fundamental research, addressing some specific social and human factor questions that need to be carefully looked at, to ensure the success of transitioning towards a new mobility system. I think there's also something around building upon the industrial strategy to bring together the public and private sector around artificial intelligence and intelligent mobility, aligned with the Government's growth strategy.
Then, one more point: it's related to trust. Government has a role to ensure that we can build trust in the machines, the technology, the data, the new services that are being developed; for example, by developing a framework for the testing of these vehicles in real environments, creating something like a CAVLA—a connected and autonomous vehicle licensing agency—on the sidelines of the DVLA.
As we have established, there's a lot of potential here for things to go wrong. I see the role of Government more hands-on, really. If this process isn't very carefully managed, then it will go wrong and it will never happen. So, Government, again, working with various stakeholders, really needs to play quite an active role to make sure this whole process goes in the right direction and that all the safeguards are in place to control that process. So, I see quite a strong role for Government in this whole process. There's so much scope—
And William, from the point of view of the innovators and the people who are producing these products.
Actually, I'll start by giving some stats. So, I think, in Wales last year, we saw 4,932 car accidents that were reported to the police. Of these, 1,005 were quite serious. With autonomous cars, the stats show that we'll be able to reduce these figures dramatically. The only small thing I'd love to see—I don't know if it's even possible—is companies like Uber or companies elsewhere who have, let's say, an autonomous car crash—. They're not very common, but they do have them. What worries me is a crash happens and then the data as to how that happened stays there—nobody ever knows. If it was within the law for us to be able to get 10 seconds before—the same data 10 seconds before, across the board, so any autonomous car company can share 10 seconds of data before any crash, we all learn, rather than we all have to repeat the same mistake over and over. There are these minor crashes and you get, 'Oh, it hit a post'. It would be really interesting if we had some sort of way, or have it written into law, so that this was shared, and then all the cars learn from the same knowledge and then it becomes a better future for all of us.
A collective black box.
Yes. I think that's quite vital, actually, this co-operation, because currently, a lot of products in the private sector operate still in silos, protective, especially of their data, because they don't know what the value of that data is. Therefore, just in case they have some value at some point in time—. But, obviously, anything that is for the public's benefit needs to be shared across the board and that includes safety-critical types of data.
Safety-critical data just 10 seconds before is all we need. Because we'll share ours as well, but it means that we can train all the cars to better be able to manage a specific scenario that could only be recreated under certain conditions that we never would've known until it happens to us.
Thank you. Mark Isherwood.
William Sachiti, you mentioned that you are an Aberystwyth graduate.
And I think, Dr Nieuwenhuis, you referred to the links you have with the psychology department, because of the broader social, moral and human aspects of this. Last year, I attended a talk at Aberystwyth University between two academics, talking about those very subjects. And the two things I wanted to ask you about in the context of the question: they showed a film where somebody left work, ordered a taxi, which was a driverless taxi, said where he wished to go, which was four miles away—it was something like 'Mumby Road'—and because of a break in connectivity, the car registered that as 'Mumbai' and they found themselves in a container on a ship on the way to the Indian subcontinent. So, there's the technical aspect. And the moral one, which was quite interesting, was that they hypothesised a crash situation, potentially, and different scenarios. So, if a dog and a cat ran out at you from either side at the same time, the driver may be a dog lover, so he or she would've chosen to hit the cat, given that they have a choice of one or the other, or if they were a cat lover, they would've chosen to hit the dog. How do you programme driverless cars to reflect our own moral choices?
This is something we come across a lot and the answer I always give is, 'What would you do as a person?' It's a very rare scenario, where most of us don't know what to do, so how do we expect a human programmer to be able to do that? So, I take the question back: what would you do?
You can't hit the dog, Mark. [Laughter.]
I'm not going to tell you which animal I would choose to pick. I'd get hate mail. But the question is how, if at all, you programme the driverless vehicle to reflect the passenger's moral views when faced with choices like that.
I think that's quite an interesting question, because I think there are different components to that question. First of all, it is about the likelihood of actually being in that scenario. With an autonomous vehicle, you're less likely to be in such a scenario than you would be in a human-driven vehicle, because it has its eyes on the road all the time and further ahead than a human being would do. So, you're already reducing the likelihood of having such a scenario and encountering that. Then, of course, there is, obviously, that question surrounding, if you come to this very unlikely scenario, what do you do? How do you represent, essentially, the breadth of the human behaviour that you will encounter? So, one person, in your example, may be going for killing the cat and the other person may go for killing the dog. That's a question that needs a lot more research, especially when you're talking about new artificial intelligence that will manage these kinds of scenarios to ensure that we don't end up just sealing the fate of specific animals or human beings or segments of the population.
I think the fundamental question, though, is: will we accept machines going around killing people at all? Is that the big issue? At the moment, we have somehow accepted people driving cars and killing other people, at some level, which, we might argue, is also really unacceptable. But, is this going a step further, saying, 'All right, we allow these machines to make the same mistakes that we do'? It's fewer, and less common, but still—.
Of course, you could programme into it, 'If it's got two legs—'. [Laughter.]
Just to add to what Dr Wolfgang said, to give you an idea of the extent of how well these cars can predict futures or see ahead, we have, let's say, four different types of scanners, which each take 40,000 scans a second, which is—what is it—1.6 million scans a second, and then all this data is then used to be able to predict what's going to happen. This is per second. It can see 100m in any direction at all times, 1.2 million times per second. So, it's predicting ability is so much better than a human. Often you'll see these cars slow down—'Why is it slowing down? Why is it slowing down?'—and three seconds later, 'Oh—where did they come from?' The car's able to see it way before we're able to. So, as he says, the chances of getting to a situation like that are quite low because they outperform us already in most other—
They're not on their phone texting, are they?
They're not texting.
I just want to go back to some of the territory that Lee Waters was interrogating. Jan Gehl, who is the great apostle of redesigning cities—you know, Copenhagen, turning it into a liveable city, and the bike revolution, et cetera—he's a real critic of one potential nightmare vision of the future. Basically, the way in which the design of everything—the way that our cities are planned, our economy—was very much, for the latter half of the twentieth century, all about design around cars. We could simply have a version of that, except that they are autonomous vehicles. Shouldn't we be starting from the starting point of saying, 'Let's take the opportunity to actually make sure that people spend less time in vehicles', autonomous or not, and then see autonomous vehicles as a kind of adjunct? Because I fear what's happening already is that people are using the subway less. There's a marked—. You're right; people are choosing to get in the Uber or Lyft in New York, and that is a real problem in terms of the long-term viability of public transport. So, the car cities of the twentieth century were a problem, but might we be creating a different version of that problem with this technology?
We don't have steam engines anymore on trains. We just adapt.
But isn't the challenge, then, for Government—? Could we say that, essentially—? We build our communities around roads. We build the roads first and then think where we fit the people. Could we reverse that and say that, actually, with mini pods et cetera, you could build communities not just without cars, but without roads, actually, in the future? Because the default—. All of us start every day and end every day, don't we, as pedestrians of some kind. So couldn't there be an even more fundamental rethink of how you plan space? Then you fit the autonomous vehicles around the people and not the people around the roads and everything else.
Absolutely. I think Saudi Arabia's building a city at the moment—$500 billion into a city designed specifically like that, with that in mind. In Singapore as well, there's a smaller one. So, there are cities around the world where this is being experimented on. It has not been done before, but it is certainly the right way to think, because it's beginning to happen. The only thing is that we don't know what the future holds because these things are always hard to predict, because there's this thing we call putting the technology before the user experience; whenever you put the technology first, it might go wrong. This is what we think might happen with drones, for example, because they are very loud and—
And even the idea of commuting, as well, which really only exists in English-speaking societies—and it's to do with property and the way that people have had to move out of city centres. But there's an opportunity to have a fundamental rethink here at a deep level about how people work, how people live, and then fit the technology around that, rather than have the technology basically design the template.
I think the bottom line is that there's actually no need for this technology, so you have to decide, 'Is this what we want?' and start with the humans first. What would the humans get out of it? So, start from a human perspective to design your cities.
Of course, there have been cities like that before—medieval cities were like that. People walked around them. So, yes, I think that would be the right perspective. The only problem is that we are where we are, and certain things cannot be uninvented. That is, again, where I think the process needs to be managed. You need to decide, and the Government needs to decide, what we actually want to get out of this. Do we want it at all? Who benefits from it? I think that's the right perspective.
If I can add to that, I think you're absolutely right: at the core should be the social benefits and the individual human benefits that we can get from this. So, we need to take a much more holistic approach and to link it up with other kinds of issues that we're facing, not only social inclusion, but accessibility, congestion, air quality, and trying to look at it from a holistic perspective and see where we can address these problems and how we can address those problems. For example, autonomous vehicles. Road accidents cost the UK economy about £60 billion a year. Congestion costs the UK economy about £40 billion a year with, of course, a negative impact upon health and a negative impact upon the general well-being. So, we need to look at the wider picture rather than looking at autonomous vehicles in isolation. It's about how these would operate as part of wider society.
Thank you, Chair. The advent of driverless vehicles is a key moment in history, really, with the public needing to embrace this huge societal shift. In my mind, it takes me back to the late 1700s when Edward Jenner discovered vaccination and the public fears around that, and how long it took us to overcome that. If you ask the average person in the street their views on driverless vehicles now, most people would just be aware of the very few incidents where this has gone wrong and led to collisions, and one of those being fatal. That doesn't even take into account the issues that we've talked about here today regarding cyber security, too. How do we overcome these sorts of barriers to make sure that the public are ready for this huge change, and will embrace it?
If I may answer that, so we're working on two, in a way, co-funded projects—VENTURER and FLOURISH. VENTURER is addressing the issue of trust and acceptability, so it is basically focusing on understanding what are those human needs at the core, what makes people react in a specific way to these autonomous vehicles, to understand that, and then to design the system around the human as opposed to designing a system and then, essentially, plonking the human into the vehicle or in front of the vehicle. So, it is about this co-evolutionary process that will generate the trust and that will generate the acceptability.
As part of FLOURISH, we are looking at a specific segment of the population—the elderly, who are likely to be the early adopters of such vehicles, given the obvious limitations. And we've held a number of interviews across a spectrum, and you get very different views from different people at each stage. We've conducted interviews at the very beginning without giving them any information, and then afterwards, after they knew more about the vehicles, and their views had already shifted.
So, I think it is about informing the public about developing the technology together with the human, because we as humans are part of this mobility system that we're transitioning towards. It is not 'transport system and humans'; we are part of it and we evolve with it, similar to when we went from the horse and cart to vehicles. We also evolved with that in terms of how we adapted to it.
I'll go back by saying that in 1876 in the UK, the first cars came from a city of horses and carts, and at the time people were scared—'What are these giant steam engine things?'—and so Government legislated that from this day forth, there shall be a man with a red flag waving down, walking in front of each car, saying that there's a car or vehicle coming. The result was that for 30 years, we were stuck behind a red flag and 4 mph, whilst Germany, Italy and France had booming car industries, and some say this is why we're still behind. So, it's important that we don't over-panic and legislate to be stuck behind a red flag. [Laughter.]
That's a great analogy, that is. Adam.
Just one final question from me. Sidewalk Labs, which is Google Alphabet's urban innovation subsidiary, has launched this idea of creating a smart city or an actual new settlement in Toronto where it's going to build out, basically—it's going to build the future. It's going to build an open air laboratory as a test bed of automation technologies right across everything, from manufacturing through to transportation, which is an interesting idea, and some other companies have ideas along similar lines. Is there an opportunity for us to do something similar? And we may be thinking about what that would look like in a rural environment. We've got these eco-towns that are, at the moment, just about environmental technology, but couldn't we actually create a kind of future model test bed where we can, in a real environment, including with human beings testing the technology as well, actually learn in that real context?
That would be fascinating, wouldn't it? I think that's a great idea.
Let's do it then. Let's do it in Wales.
I would entirely agree because the UK has the core skills and bases and expertise to actually develop that, and with the right framework in place—legislative framework, insurance framework—that can be achieved. And I would also add to that that Wales and the UK could lead on standardisation in the space, given our expertise.
And we've got relatively short communication lines in Wales because we are a relatively small country, so it's easier to do that sort of flexible programme.
Yes. Great. Well, there's an opportunity. It may be something for the committee to consider. Thanks.
I really like that 1876 fact. That was really good. That's going to make it into our report. [Laughter.] It's our job to listen to what you've said this morning, and other witnesses, and to bring forward recommendations to the Government. But is there something, perhaps one key area, that you think that we should be making a recommendation to Government on? If I can ask you—we're just over time now—to succinctly tell us what you think is your top ask of Government. What would that be?
So, autonomous cars, it's more a tech industry, and tech companies tend to be small start-ups most of the time, more than they are the large, sort of, straight to millions of funding. I'm in touch with the many tech entrepreneurs working in the autonomous car space. They may not be building the car entirely like us, but maybe bits to do with the car. The Government's done a great job of creating grants and funding opportunities, but the problem is the way it's structured is, you would need to have a team of grant writers to be able to ever win a grant as a small company. So, nearly every start-up you see doing amazing things were heavily reliant on venture capital funding or angel funding. We have to be so much better, and it's a big problem that we never get the Government funding. We've never had Government funding and we've built an entire factory. It would be great if we had more support. The support is there, but it's done in a way where it's only the big universities, university departments, that get it because they know exactly how to write the grants, because they have teams of people behind it. And, so, this is a skill that smaller companies, which are the ones innovating in this space, can never truly attain.
I've not come across a grant so far where it doesn't require filling out lots of paperwork, but I take your point, yes.
And it requires a lot of resource in a university as well that's often wasted.
I think the idea of a local experiment, making Wales, or a small location in Wales, an experimental space, that would really allow Wales to go ahead and take control of at least part of this agenda by integrating all these systems, not just the autonomous systems, but also the human systems in a small town or something, or in a rural area. I think that would be of a lot of value, and you would attract people in from all over the world to come and have a look at that.
Yes, I would agree with that. I think it is about facilitating the integration of stakeholders that would not normally come together. And we've seen that as part of these collaborative research and development projects, such as VENTURER and FLOURISH, where we have a very, very diverse set of stakeholders coming together to work together and achieve a common goal. So, for example, from academics via small tech companies, large industry, charities, insurance and legal companies, governmental entities, all working together as one to achieve that goal. Ultimately, these are specific projects. Then, we need to think at a higher level to really bring together the learnings from these different projects to ensure that we have a consistent roll-out of these solutions onto our roads. So, as part of that, I think what is absolutely vital is that Government facilitates collaboration and information sharing between all these different stakeholders.
I'm grateful. Thank you. We really appreciate your time this morning. We know you're busy people, so we appreciate your time this morning. We're very grateful, and that does bring this particular item to a close.
bod y pwyllgor yn penderfynu gwahardd y cyhoedd o weddill y cyfarfod yn unol â Rheol Sefydlog 17.42(vi).
that the committee resolves to exclude the public from the remainder of the meeting in accordance with Standing Order 17.42(vi).
Cynigiwyd y cynnig.
I move to item 3, under Standing Order 17.42 to resolve to exclude members of the public from the remainder of the meeting. Are Members content? Yes. Lovely. That brings our meeting to a close today.
Derbyniwyd y cynnig.
Daeth rhan gyhoeddus y cyfarfod i ben am 10:50
The public part of the meeting ended at 10:50.