HoC 85mm(Green).tif


Transport Committee

Oral evidence: Self-driving vehicles, HC 519

Wednesday 17 May 2023

Ordered by the House of Commons to be published on 17 May 2023.

Watch the meeting

Members present: Iain Stewart (Chair); Mike Amesbury; Mr Ben Bradshaw; Jack Brereton; Ruth Cadbury; Paul Howell; Karl McCartney; Grahame Morris; Gavin Newlands; Greg Smith.

Questions 261324


I: Rt Hon Jesse Norman MP, Minister of State, Department for Transport; and Claire Wren, Deputy Director for Future Transport Systems and Environment, Department for Transport.

Written evidence from witnesses:

Department for Transport

Examination of witnesses

Witnesses: Rt Hon Jesse Norman and Claire Wren.

Q261       Chair: We now move to the second part of our production today, with a little cast change at the top table. Moving to self-driving vehicles, a very high-level question: are they a good thing or not?

Jesse Norman: Self-driving vehicles?

Chair: Yes.

Jesse Norman: That is a very interesting question. The answer is yes, of course. It is like a lot of these things: it is not impossible to see dystopias as well as utopias. We talked about them and sketched them out in some DFT work we did when I was in the Department in 2019. In general, there is tremendous potential.

It is obviously not just about safety, although there would be the possibility, scouted in some of the literature, of reducing the number of accidents by an order of magnitude or even, conceivably, two orders of magnitude. At the moment, we kill about 1,700 peoplewell, we don’t, but 1,700 people a year die on our roads. It is an appalling statistic. It has fallen over time. It would be very good to find a big, structural way of getting that number way down.

There are questions of inclusivity. Self-driving vehicles potentially allow much greater inclusivity. They allow us to reach parts of the community that might really struggle to get out. They allow much greater potential connectivity. They could substitute for much more heavily carbon-emitting kinds of vehicles. The self-driving small bus is a very interesting and attractive area of mobility as a service. We are just beginning to understand the full range of applications, but, in general, yes, I think there is a tremendous case for them.

Q262       Chair: Is there a risk that self-driving vehicles will increase the volume of car use, to the detriment of public transport?

Jesse Norman: It is very hard to know in advance. You can look at two possible scenarios. The dystopia is pod world, in which you leave your house, get into your office, have a meeting in Manchester, get back into your office and work all the way back to your house or flat in London, or whatever it might be. If every single individual is doing that, the pattern of car ownership that we have at the moment, which is, roughly speaking, narrowing towards one person to one car at all times—a steady escalation—could be made even worse.

On the other hand, if what you have is an environment in which people think, “Actually, it doesn’t make any sense for me to own a self-driving car; I will just order one when I need it,” the usage of cars could go down. Instead of being unoccupied for 98% of the time, it becomes 50% or 60%. It is just not clear what the options are. Even if one had a model that was highly authoritative in its assumptions, you would wonder whether there would be effects that were unanticipated coming out in the way that would play out over time.

Q263       Chair: To follow up on that, the Government’s national road traffic projections found that the widespread use of self-driving vehicles would create “a world of fast growing traffic demand”. Won’t that just increase congestion?

Jesse Norman: As I said, I think it could go either way. Of course, the thing that will really make the difference is cost. If the effect is to lower the cost of lower-carbon alternatives, people would be interested and excited to do that. If the effect is to raise the cost of the alternatives, people will use cars and use self-driving cars. The fact that self-driving cars will become, in the way they operate, potentially a relatively low-cost vehicle will make a difference.

There are definitely negative potential air quality impacts, in particular, because of having lots of vehicles that generate lots of particulate matter from braking, road use and tyres, but there are other scenarios in which you can see much more hopeful and potentially attractive alternatives. That is before you get to issues of safety and so on.

Q264       Chair: Picking up the issue of cost, do you see self-driving vehicles being the preserve of the wealthy for the foreseeable future, or do you see a mass application for any user?

Jesse Norman: I see both. If you are talking about the private car market, which is the place people often start—they think it will be people with Teslas who will pay to buy a car at the super-premium end of the market—there is no doubt there will be some of that because of the bling factor, early adopters and all that stuff. It makes economic sense to launch at the top of the market for many of these prestige, as it may be, goods. There will be people who will want to continue to own their car, even if it does not make economic sense, because they just like having a car in front of the house, or wherever they might be.

This week in Scotland, the Scottish Government have launched CAVForth, which is a self-driving bus system across the Forth bridge. That is the other end of the market: very exciting, low-cost, highly accessible transport in communities that will really benefit from it. I think that is another very exciting alternative. We are going to see a very wide range of changes across the market.

Q265       Greg Smith: Good morning, Minister. Coming off the back of that, and accepting that there is not a precise answer to this, although the Government must be doing some modelling around it to prepare, realistically, how many self-driving vehicles will be on the roads of the United Kingdom in 10 years’ time?

Jesse Norman: I stand in awe of any modeller who could give you, in any sense, a closely accurate figure. The DFT, as you know, is the home of high-quality economic modelling in Government. Our boffins will try their best, but I would not be surprised if they were a long way off. In 10 years’ time, possibly, it is relatively easy to predict because there will not necessarily be that many. The really interesting question is in 15 years’ time, at the point when you have started to see take-up gear up very hard.

Of course, lets not forget that, when we talk about self-driving vehicles, we are talking about a very particular kind of beast. There are lots of intermediate driver assistance technologies that sometimes get confused with self-driving. One of the challenges for Government and for the industry is to be clear about what is one and what is not. For example, automatic lane keeping is a form of self-driving technology. As you know, in our CAM document, we look forward potentially to that being on our roads in a matter of a few years.

Q266       Greg Smith: I accept that the modelling is incredibly difficult to get right. I was just trying to get a feel for whether the Government think that 1% of vehicles on the road will be self-driving, or 5%. Where will we get? You gave the example of the self-driving bus on the Forth bridge. Where do you see the significant first shift, accepting the Chair’s question about those who possibly have the money to just go out and buy whatever early technology comes? Do you actually think that public transport, such as buses or, as I saw at Heathrow the other week, the pod car park—where you risk your life in some sort of bubble thing to get to the airport—will be the ones that take off faster?

Jesse Norman: The irony, of course, is that when the technologies are fully proven, they are the ones where you won’t be risking your life going to the airport.

I do not think it is having it both ways to say that we are going to see a whole variety of different things happening. As I say, I think we will see automatic lane keeping relatively soon. I do not think there is any reason why we should not see a whole bunch of different, new shared-use automated vehicles coming through. Those could be pods or buses. They could be smaller, intermediate vehicles such as the seven to 10-seater used in defined suburban areas. I think we will see all of those things.

If you look at the numbers, you would do better to trust the numbers coming from some of the companies than from just one single Government source. You tend to get more when there is a bit more commercial skin in the game. When you have a plethora of different estimates, you will get a better sense of what their expectations are.

Q267       Greg Smith: We can flip it on the head, then. What proportion of vehicles do the Government want to be self-driving in 10, 15 or 20 years’ time? Is there a DFT view that it would be a good or a bad thing to have 20%, or 50%?

Jesse Norman: It is a fantastic philosophical question. Claire may want to come in, but my own view is that the safety benefits are so great that once you get to a critical mass in particular uses, that is highly desirable. There is some modelling that suggests you start to get that when 30% or 40% of the cars in a given context are self-driving; you start to get real benefits. If you have 80% or 90%, you get incredible benefits.

We are going to be very use-case constrained, but overall I think the Department is, in a way, more preoccupied with getting the transition right, getting the public messaging right and, frankly, maintaining the competitive advantage that we have historically had in this country from an industrial strategy standpoint than necessarily focusing on what the final outcome is.

Q268       Greg Smith: Within that transition, clearly the Government have spent a lot of taxpayers’ money on financially supporting people to transition to EVs over the last five, six or seven years. There have been different schemes over that time. Do you expect there to be similar public subsidy schemes for people to transition to self-driving?

Jesse Norman: I am not sure that I see any case for public subsidy on this at all. The reason why we did that in the case of EVs was that there was a very clear market failure that required a process of firing up their use in order to prime the charging infrastructure and the use. You will have seen that we have just launched the ZEV mandate, which is another big move to get private investment into charging.

In this case, the appetite is so palpable in different contexts, and the competitive energies to pioneer, trial and then roll out in other geographies are so great, that I would expect there to be tremendous drive. Claire is nodding, so she may want to come in.

As you know, you can pick up a self-driving taxi in Phoenix, Arizona, as we speak. We have some companies, in which the Government, to their enormous credit, have placed some pretty foundational early bets, doing phenomenal work in this country. We have to try to get behind that whole process and really push things as hard as we can now.

Claire Wren: I absolutely agree. I think this will be driven by the manufacturers. Going back to the point about whether we would have an ambition on the percentage, I think it would be foolhardy to have a short-term ambition on the percentage. The safety benefits, as that goes out, will be quite substantial.

In terms of uptake, obviously we always jump to the robo-taxi idea because it is often the day-to-day things, like freight and mass transit, that we are likely to see first.

Jesse Norman: If you talk to the Oxboticas of this world, or people like that, they are as excited about the in-company use, the in-field use, as they are about the public highway use.

Claire Wren: Of course, our roads wind a little more than those in Phoenix, Arizona.

Q269       Greg Smith: Indeed. I will ask my last question; other colleagues are going to come in on safety later. From the clear answers you have given that you believe there to be safety benefits, I get the sense that the Government are enthusiastic about self-driving vehicles. What budget do you have in the DFT for assessment of self-driving tech, preparing what regulations need to come in in the future to ensure that we have the right regulatory environment in the UK and examining other examples and learnings from overseas? What are the Government spending of taxpayers’ money on this sector, and how does that compare with other transport budgets around rail, road building, active travel or whatever it may be?

Jesse Norman: Three minutes with the DFT accounts will remind us all that the vast majority of DFT money goes on rail and road, and of course particularly on HS2, and that does have a distorting effect.

Greg Smith: Don’t start me, Minister.

Jesse Norman: No, no—you’re red flagging me.

Greg Smith: You were doing so well until then.

Jesse Norman: You may not have taken an interest in that, Mr Smith. However, the thing that is really interesting about this area is that there has been a consistent early process from the very beginning of getting behind this. If you talk to the companies involved—I know you have done; I have read the testimony—they will say that the reason why it is working here is that you have CAVPASS looking at the whole process of certification. You have done all the work on the Law Commission. You have the most carefully nuanced legal definitions involved. You have the trialling context in which they can do it. You have a phenomenal supply chain and academic support. It is that ecosystem that has created a moment of competitive advantage for us, and we need to get behind it now.

Q270       Greg Smith: I agree with that, but my question was that the buck will ultimately stop with the DFT to get the regulatory environment right in any other schemes that need to be put in place to enable this technology to be used on UK roads, be it for private use or public transport. I totally understand that the bulk of the money goes on road and rail, but what is the substance in the DFT—the budget and the number of civil servants working in this area—that is getting behind this?

Jesse Norman: I have no idea. We have CCAV. Claire will want to talk about the numbers.

Claire Wren: I can talk about the numbers. First of all, it is important to say that, at the moment, we do not feel constrained from budget in terms of the safety and security processes. It is obviously smaller than building a road because of what we are doing—

Jesse Norman: But it is still a lot of money.

Claire Wren: We have £34 million specifically on safety and security processes. That is separate from the £66 million we have on commercialising CAM. That £34 million is just focused on the safety and security that we were talking about. At the Centre for Connected and Autonomous Vehicles, obviously the numbers move about a bit from day to day as the team comes and goes, but it is about 30. That is a joint team with the Department for Business and Trade and DFT—it is mostly DFT officials; it is partially Department for Business and Trade. That is exactly bringing together the point about talking to business, working very closely with the sector and the importance of commercialising as well as understanding safety and security. CAVPASS, the programme that the Minister mentioned, is a very specific programme that we have which looks at the very detailed processes and procedures that we need to put in place to be able to assess the vehicles when they come forward.

We are already at the point of thinking about the standards that we would think about and the procedures that we will need to go through, as well as what hoops the manufacturers will need to do. We are thinking really carefully about this in a multi-year programme. From the top level, you think, “We need to get these vehicles on the road,” but as you start to think through who is going to do it, how they are going to do it and how long the procedures take, those are the things we are already working on. To repeat the point, at the moment we do not feel constrained on budget.

Q271       Jesse Norman: It is worth mentioning one point. I will take an example that is not self-driving technology, which is obviously BlueCruise. We have just approved BlueCruise. That was a standard that came out of the UN, where we have a lot of involvement in this area; we are one of the leading players, as you might imagine. It is backed by a ton of data from effective roll-out on properly enabled roads in the USA. Why did Ford launch it here? Because the motoring environment and the fiscal and commercial environment are the most satisfactory from their point of view. That is the kind of thing we need to be able to do with automated vehicles and self-driving vehicles as we go forward.

Q272       Chair: Before I turn to some of the safety aspects, I want to follow up on a line of questioning. We have heard some concern from the private companies who are committing a lot of investment into developing this technology that for that to continue they need to have a full regulatory framework. It goes back to the point that Ben was making about e-scooters. The question mark as to whether there is going to be a transport Bill to introduce some of this is potentially going to divert that investment away from the UK. Is that a concern you share?

Jesse Norman: Absolutely. I have spoken to several of the companies involved. I am acutely aware of the concern. I have read all the testimony that they have given to you. It is absolutely a concern that I share. The Secretary of State has rightly said that we cannot make any anticipation about where this will come out, but you can rest assured that we are making the case as vigorously as we can on transport, on safety and on industrial strategy grounds for this to be a priority for the Government.

Q273       Ruth Cadbury: Minister, you just said that the safety benefits of autonomous self-driving vehicles are so great—

Jesse Norman: Potentially, yes.

Q274       Ruth Cadbury: Potentially. Okay. You are absolutely right that, on the one hand, a massive proportion of incidents and collisions are caused by driver error or risky behaviour, and many of the rest might be caused by error or risky behaviour by other people in the vicinity, or passengers. The driver error goes when you have autonomous vehicles, but, on the other hand, there are all the unintended and unforeseen incidents that could happen with the use of AI, which this is.

Some of us had the benefit of being passengers in level 3 vehicles around part of north London recently. We sat in a hands-off but driver-in-charge vehicle. Those of us who are drivers ourselves experienced different incidents. Often, the vehicle was slowing down when a rational and careful driver would have said that there was no need to slow down at that point. In our car there were a couple of incidents when the driver had to take over because the car had not anticipated a potential incident.

How confident are you that the research is going to be robust enough, particularly on the average urban road, to predict the behaviour of a wide range of people? A vehicle cannot predict in what direction a toddler will go, or children dribbling a football along a pavement. In both cases they are probably going to be entirely safe, but a driver can anticipate that and can make a split-second decision. The minimum time for the driver to take over is two seconds, yet I make decisions in those circumstances within a split second.

We are also talking about level 4 and level 5 vehicles, for which there will be no physical driver in the car. The car will be controlled somewhere else. What do we need to do to be fully confident, and how long is it going to take for that level of research and testing to be done? How do you test on toddlers without using real toddlers?

Jesse Norman: Thank you for the question. There is a whole series of questions. Let me pick them apart in due course. For what it is worth, I have been in a bunch of self-driving cars with different technologies from different companies. In one case there was a problem with a pod about five years ago, but not since then. That does not say anything of course; it is just an anecdote, but it is important to put that on the record. Everyone is always putting on their best show for the Minister, so it is very interesting that you say you have had that experience. You have had testimony from the chief executives of two of the major companies explaining how their vehicles tend to negotiate a fallen tree or the unexpected, and pointing out that in some cases, because they use technologies that are more sensorily enabled than human senses, they ought to be able to, and they expect them to be able to, detail and engage with the older person who walks out into the road, and so on.

The horrendous thing about this is that we can be certain that there are going to be some scare stories, particularly in the early stages, because of the diversity of human life. It is a really interesting question of consent and democratic engagement as to whether or not we, as a polity, as a country, are able to weather that moral panic in the pursuit of a future that might lead to an enormously larger number of people not dying. Without being too philosophical about this, that is what I think the question is.

If you look at the technologies, one of the reasons why the UK is so attractive that I didn’t mention earlier to Mr Smith is that it offers an incredible variety of use cases. People and companies get a testing environment here that is second to none and much superior to open boulevards in other countries because of the nature of our road system. The only thing we can do is to work very closely and carefully with them on graduated processes by which automation is introduced on to our roads under constrained and regulated circumstances. If we do that, and it is done carefully, I think it ought to be possible to do it safely, effectively and with public consent.

Without dwelling too long on BlueCruise, one of the most interesting things about that is that it is easy to see it as an individual technology, but it actually came out of 30 years of previous work on cruise control, adaptive cruise control, automatic braking and all the different elements that make that up. As people understand the difference between these forms of assistance and self-driving, we ought to be able to create a public environment in which there is acceptance and understanding of that situation, but we cannot bank on it.

Q275       Ruth Cadbury: How are you going to judge whether self-driving vehicles are actually safer once they are introduced on to UK roads?

Jesse Norman: Obviously, it is something that will be intensely publicly scrutinised by the Department, by independent agencies and by the media and the general public. There will be a gigantic amount of review. That carries risks as well as benefits.

Q276       Ruth Cadbury: Apart from using the power of social media, which often does not follow the actual facts, what metrics will you be using to judge whether it is safer or not?

Jesse Norman: That is an interesting question. As you know, the Department does not always have as much access as it would like to the specific circumstances of particular injuries or deaths. That is a matter for the police. There will need to be some pretty close liaison on both sides to make sure that there is an understanding of what those are in the early stages, and therefore an understanding of what the causative factors may be. Unfortunately, because of the nature of the hyper-energised media environment we are in, people jump to a conclusion about what causation was without actually knowing. It is quite important to wait for all of the facts to come through. That itself creates a risk because, inevitably, that takes some months, and the train of public opinion may already have left by that point.

Q277       Ruth Cadbury: You said that your safety ambition for a self-driving vehicle is that of a competent and careful human driver. What does that mean in practice? Do you plan to set that standard in law?

Jesse Norman: My ambition would be to put that standard into primary legislation so that people understand that it is the goal; they can see on the face of the Bill that it is the goal. Inevitably, how that cashes out and whether we can continue to drive standards more quickly upwards is something that we could reasonably place in regulation. We can say, “We consider this to be the standard, but we are going to upgrade that standard.” The debate for us is, do you put that standard on the face of the Bill, which then, if you can go further and do better than that, requires further legislation, or do you place it in an ancillary piece of public communication—perhaps a speech by a Minister—which then allows the Department to upgrade it?

Inevitably, over time, you are going to want to put numbers on that. You are going to say, “So far it looks like were getting the following reduction in killed and seriously injured”—I hope and, goodness knows, we all pray—“but we think we can do better. As with our decarbonisation targets, we are going to turn the screw a bit to try to get better performance overall. That would be how I would think of it.

Claire Wren: Can I give some examples of what that means in practice? A careful and competent driver would not break the speed limit. They would not jump a red light. If you go out on the street, you see both of those behaviours all the time. They are two very specific ways in which you would see a self-driving vehicle—

Jesse Norman: Also, to go to Ruth’s point, a two-second reaction time is slower than the split-second reaction time of an individual, but then cars do not get distracted by mobile phones. There is a whole series of other things.

Claire Wren: They have metaphorical eyes in the back of their head because there will be cameras all the way round.

Jesse Norman: And they will not necessarily have had a bevvy at the local before getting on the road. There is a whole series of other risk factors that it is worth taking into account.

Claire Wren: You mentioned the number of accidents that have a contributory human factor.

Jesse Norman: It was 80% or 90%. It is incredibly large.

Claire Wren: Yes. It is 88% on the latest figures, so almost 90%. It is a huge number of collisions. Thank you for that correction. It is absolutely a challenge as to how we will assess in practice what safety means. The examples are abiding by the speed limit, stopping at red lights and giving cyclists 1.5 metres to overtake. They are all really good examples.

Jesse Norman: I have just been handed a note by my parliamentary secretary telling me that I was wrong in saying that BlueCruise was approved under UN regs. It was actually approved under article 39 of GB type approval regulations. Let me put that on the record, with apologies to the Committee and thanks to my team.

Q278       Ruth Cadbury: I want to go back to my comment about toddlers. AI, as we know, and vehicle testing in the more traditional sense, tends to use—tends to focus on, lets saythe average white man. We know the history of crash test dummies, which, because they were based on the average male body, have caused avoidable injuries to women. Are you sure that the testing is looking at the whole of the population in a way that, too often, standard research does not?

Jesse Norman: A big shout-out to Caroline Criado-Perez. I totally agree. If you scale infrastructure to the expectation of male physical presence or behaviour, it is catastrophic for more than half the population potentially. That is absolutely right.

One would hope that the safety gains to be made are sufficiently great as to overwhelm differences in gender. Of course, there will be differences in gender use, as you might imagine, but I take the point you make. Claire might want to come in, but I can certainly put that point to officials. We will obviously want to be sureI think I am right in saying that we have already built in the question of scrutinythat, to a humanly determinable extent, there should be no embedded discrimination on any of the bases we are talking about.

Claire Wren: It is an incredibly important point. Earlier I talked about CAVPASS, and that is the way that we are developing these standards. We have not developed them all the way to the end, as you might imagine, because we have only just started, and it is a really complicated thing. We have started to think about what needs to be incorporated. I cannot give you the full answer because we are not there yet in the policy work. Conversations at places like this Committee are a really important part of that policy process. It is important that we are having this conversation here today.

Different companies have different technologies. Some of them are using machine learning and building on the day-to-day experience of drivers. They basically put sensors into existing vehicles and collect the data. Obviously, that picks up what drivers do around toddlers and in all the sorts of different environments, as well as the pre-programmed average white man, who is very much not the standard person. The different ways the different technologies think about, or rather consider that issue—“think” is probably the wrong word—are among the things that we will need to unpick and that will need to be presented as evidence.

Of course, there is also an ex ante and ex post issue. First of all, there is the question about

Jesse Norman: Latinate, Claire?

Claire Wren: There is the question about what evidence we need in advance of approving a specific vehicle. This is all going to be done on a case-by-case basis. It is not that we will put a big tick in the box and everyone will be able to go out, but that individual models will need to come forward. After those vehicles are out on the road and have been listed as self-driving and approved to be on the road as self-driving, there will be the collection of data and continued learning from that. Both of those things are important.

Jesse Norman: It is worth saying one other thing. You were focusing rightly, Ruth, on the behaviour of pedestrians and others who are road users and whether that might change. Of course, there is also the effect on passengers in self-driving vehicles, taxis and the like. One is going to be looking at both of those things, I would think.

Q279       Paul Howell: You have talked about different people and different effects, but obviously one of the potential effects is the driver in a self-driving vehicle when the self-driving vehicle needs their intervention at a point in time. How do you prevent them from putting themselves in the space where they overestimate the capabilities of the vehicle?

This is part of the same question. To take the metaphor of airlines, a lot of pilots take off themselves, put the plane on autopilot to get to the other end, and then take control again for the landing. You can see exactly the same thing happening with big freight vehicles. They will do the complicated bit, get down the motorway on self-driving, and then do the complicated bit at the end. How do you ensure that the driver is up to speed—pardon the pun—with the vehicle at the point in time when they take engagement again?

Jesse Norman: Of course, it will depend on the context we are talking about. Because it’s accessible and we can think about it, in the case of BlueCruise, what allows the vehicle to be comfortable when the driver is taking their hands off the wheel? I have driven a BlueCruise vehicle, to the extent that you canI have seen it done, I should say; I have been in the vehicle. Your eyes are tracked by eye monitors and when you are not paying attention to the road for a period of time, it will flag it to you. If you do not take back control or show that you are in control and ready, it will bring the car to a safe stop.

It is a very interesting experience to do that. I have tested it. I looked away for a while to see what would happen and it started to slow down. It is very interesting. In some of those contexts you will want to have monitoring of a user in charge, just to make sure that that person is available for a transition even if the car is under self-driving mode. It will depend on context. It is potentially likely to be an offence not to be available for a transition request from a vehicle in those circumstances. There will be severe sanctions for people. There will obviously be the severe sanction of killing yourself or hurting yourself, but there will also be legal sanctions against you, in the way you would expect now against drunk driving or something like that.

Q280       Paul Howell: Do you think there is also the potential that the actual skillset of a driver will diminish because they are no longer doing as much of the driving, so therefore, when the intervention comes, their inherent skillset has—

Jesse Norman: I would have thought you will have two kinds of things. It is very interesting that if you talk to members of the younger generation—my children’s generation—many of them have no appetite to own a car or to drive. They are just not interested. They do not think it is a particularly interesting thing. My generation—

Q281       Paul Howell: That probably depends on where you live. If you live in some of the rural parts of the world, there is a very big appetite to drive. It’s the only way you can get there.

Jesse Norman: Of course that is true. People go round the houses, depending on where they are on this issue. You are absolutely right. I am not generalising on that. I am just saying that there are some people who have different views on these things.

I expect you will find that there are people who love driving, who want to do it and who will take steps to be able to drive even in a self-driving world, because of the pleasure, excitement and the sense of getting away that that gives them. For other people, there is clearly a risk that their driving skills will start to degrade in the intermediate stage because they relatively rarely do it. If you are doing, for example, automatic lane keeping, you still have to be a highly effective driver to get through streets, to get to your lane before you turn the system on. All that local stuff will still have to be embedded, which I think is probably the majority of someone’s driving skills. That is something we are going to have to watch. If you look at how people’s skills at mental arithmetic have changed in the face of calculators, you can see that automation is a constant potential threat to some skills.

Q282       Paul Howell: Moving to a slightly different section, you have talked about the driver who wants to drive and the driver who does not want to drive. Do you think that the public want self-driving vehicles on the roads?

Jesse Norman: Interesting. Part of the issue with new technology is its unpredictability. The joke is the two cave dwellers in the troglodyte era. One of them says, “Harry, do you know what? I think someones going to invent a wheel.” Harry says, “What do you mean?” and he says, “Well, it’s a round thing with a hole in the middle. Oh, okay, Ive invented it.” You can never tell what the future is going to be and what the reaction will be.

What is quite interesting with new technology is that sometimes people do not like new technologies, and they react against them. Very often, you discover a market as you go. We probably all use our mobile phones from time to time. That is an example. Nobody contemplated that 25 years ago. It is astounding that it has made such a difference.

Q283       Paul Howell: Do you think it is the Government’s role to convince people that these vehicles should be on the roads?

Jesse Norman: That goes back to a question raised earlier. I think the Government’s role is to enable the benefits to be realised and, if people pick that up, to support them. There clearly are social benefits that are worth moving towards if we can do it in a way that does not grossly oppress human freedom or pollute the atmosphere. There are plenty of environmental benefits if it is done right. There is clearly a role for Government policy both in enabling and in then regulating. In many ways one wishes that were not the case, but it is the nature of the industry and the nature of the technology.

Q284       Paul Howell: It is one of those areas that could be quite emotive in certain spaces very easily. We have seen with other innovations that there is a perception that it is the right thing to do for safety. There is a perception through data that something is safe. The easy example would probably be smart motorways, where the perception was that they were safe, but public opinion kicked back so hard that they did not perceive them as being safe and as a result—

Jesse Norman: It is interesting that you raise that because it is the other way round.

Q285       Paul Howell: How do you see the impact of that potential social pressure into the space—

Jesse Norman: Obviously, that is a very current example. There are plainly cases where there is a very clear, well-attested, documented and numerically supported public case for action and the public won’t wear it. That is just one of those things. In a democracy it is hard to object to that. We have to work with consent, in the way that we have described.

Q286       Chair: I would like to turn to another risk that self-driving vehicles potentially have. That is the vulnerability to cyber-attack from a malevolent terrorist group or a hostile state. How worried should we be about someone hacking in and using self-driving vehicles as a weapon?

Jesse Norman: It is a very good question. I will start by making a couple of distinctions that might help us to think about it. When we think about self-driving vehicles we are thinking about autonomy. Connectivity is a slightly different thing. It is the level of connectivity that creates the potential for, as it were, the widescale cyber-attack. Any cyber-attack is potentially highly problematic, as you might imagine. We are talking about technology. As I say, we all carry mobile phones. There are occasionally emergency patches that have to be put through on mobile phones because of the unpredictability of hacking attacks on technology. There clearly is an ascertainable risk.

When people are being driven around, that risk potentially carries a very high cost so it is absolutely right to be concerned about it. When you have a vehicle that is more autonomous and more reliant on different technology, sensory inputs, in principle you have a greater degree of potential resilience in the face of a deviant input from a central controller, which is the worry that people have. As you know, many of these technologies have lots of different inputs. They have LiDAR; they have cameras; they have maps and other technologies that make them aware of where they are in the road.

What are we doing to protect against that? Obviously, it is a source of extreme concern within companies. It is something that we track closely in DFT. It is something on which we work very closely with the National Cyber Security Centre to make sure that there is a heads-up and awareness of the potential threat. We tend to be quite good at that in general in this country, but none of that is any reason for complacency. We need to continue to focus on it. The more autonomous the vehicle is, the less you can get the mass destructive effects that are really galvanising some aspects of the public imagination.

Q287       Chair: I would like to put a scenario to you that was given to me by the Motor Insurers’ Bureau. They have a real concern that many cars now have the latent technology in them for a software upgrade at some point in the future. There are daily cyber-attacks on BMW, Mercedes or whoever. The risk is that there is a mass targeting and that all cars with this technology are given an instruction to accelerate for so many seconds and then to turn right, causing carnage on the highway. How do we stop that?

Jesse Norman: I recall being in your chair and taking testimony from the chief executive of TalkTalk, when they had to admit, as a leading UK technology provider, that they had been cyber-attacked successfully by the simplest possible form of cyber-attack of an allied distributed service. There is every reason to guard against complacency on that.

Of course, in these cases we are looking at specific vehicles in specific use cases, licensed to do specific things according to pre-ascertained security protocols. We are also looking at companies that are hyper-vigilant across a lot of different vehicles and a lot of different users, potentially in different countries, about potential attacks on their device and system. It will be a requirement that all of the software be kept up to speed by the relevant organisation as part of the regulatory process.

We cannot rule it out, and indeed in some respects it looks like it is inevitable, that there will be attacks and that some of them will be successful, but we can take every step we can possibly imagine to try to head them off.

Q288       Chair: On the point about security patches, if that is the right term, should those be automatic? My understanding is that some of the software updates will have to be actioned by the user. Should the security level updates come in anyway?

Jesse Norman: That will be looked at very specifically in particular use cases. The default principle is that the institution or individual liable will be responsible for the update. If you are, for example, in an automated taxi or bus, it will be the bus operator who is liable, and they will be responsible for putting in the patch. Of course, as you go through the detail there will be some cases now where we think about whether we want to install the latest update of whatever it is. We have a choice. In practice, in the world we are talking about, there will be very constrained choices because it will have to track the liability.

Claire Wren: If you think about the computer example, which I am sure is in your head, you would not want to restart your computer when you were in the middle of a Select Committee hearing, but you would get increasing warnings of increasing urgency. We would anticipate a similar situation. If you get asked to do it as you are driving down the motorway, is that the time to update the car? No, it isn’t. I don’t know about the parliamentary system, but the DFT system insists at some point that you then do it before you can do anything else. We would expect a similar situation.

Jesse Norman: That is in a context where the liability is potentially likely to shift, and therefore you have to ask the user whether they wish to sign it off. In many cases in automated control, there will be no user in charge and the organisation will make that patch themselves.

Q289       Chair: Related to the scenario I gave about the mass attack, the Motor Insurers’ Bureau raised it with me because they have a real concern that such an event could potentially bankrupt insurance companies.

Jesse Norman: Yes, of course. It would be a black swan event. It would be like having a tsunami or a hurricane. It would affect so many people in ways that were not anticipated. It might come outside the structure of any normal insurance arrangements, and therefore, potentially, liability could rest with them. One can absolutely understand their concern.

Q290       Chair: I appreciate this is straying outwith the boundaries of DFT, but wearing your former ministerial hat in the Treasury, are there discussions under way to have a structure similar to the Pool Re system for property insurance in the event of a mass terrorist attack, so that there is almost an insurer of the insurers?

Jesse Norman: Yes. Of course, this is by no means the only case in which the potential mass effect of a terrorist attack can occur. There are other modes of transport and other modes of human behaviour and infrastructure where you could see a terrorist attack having a very significant and catastrophic effect. As you say, it falls outside DFT. I would like to think—I am sure officials will make sure in the light of this conversation—that the relevant Treasury officials and Ministers are fully alerted to this Committee’s concern. I am sure that they are in conversations with the MIB already, but if they are not, this will be a little additional nudge.

Claire Wren: We absolutely are.

Jesse Norman: I have a former Treasury official to my left, so that is a helpful intervention.

Q291       Chair: There are a couple of more general questions about insurance. How do you see the insurance market having to change for different levels of autonomy in vehicles?

Jesse Norman: Gosh, that is an interesting question. I have not really reflected on it in any depth. Claire may have some thoughts that she wants to add. I suppose the very obvious point would be that the whole system has been developed by the Law Commission, taking all the inputs and expert advice it has had, in order to be sensitive to the question of liability that drives the question of insurance. Given the genius, depth and fertility of mind of our insurance industry, one would expect there to be products coming forward that address the very wide range of potential liabilities that could arise. Of course, we will have criminal sanctions against some, but you will still have users in charge who may drink or who may not be, as it were, fully capable.

There are environments in which there will be existing kinds of offence for which insurance may be triggered, and there will be new ones that come out of the self-driving mode. Those will have to be covered by insurance. One thing I think we can say with absolute certainty is that a product is not going to be launched if people cannot be insured to use it. That is going to be a very live topic of conversation over the next two or three years as we move into those deployments.

Q292       Chair: What access will different groupsaccident investigators, insurance companies and the like—have to vehicle data when collisions and accidents happen?

Jesse Norman: Again, that is a matter of an evolving discussion, as I understand it. Take a worst-case scenario where someone is killed by some unexpected use-case development in which a vehicle manufacturer is at fault, or an entity that holds data acting as a relevant responsible liability-holding entity for this, such as a bus company. You would imagine that there would very quickly be a need to scrutinise the data record, which, by presumption, would be pretty complete about what had happened, in the same way that the police have powers now to check someone’s mobile phone if they want to work out what their movements were, whether they have been in a particular place or behaving in a certain way. I do not think that is hard to imagine, but it will need to be put into law and it will need to be developed. That is an ancillary further aspect of what we are talking about.

Claire Wren: The Law Commission recommended data retention for 39 months, and the Government have endorsed that. Some insurance companies would like a longer period, but that 39 months reflects the three years’ personal liability plus an extra three months to be on the safe side.

Jesse Norman: To get your claim in, yes.

Claire Wren: It is a balance. Data retention is obviously not cost-free, and you can imagine the amount of data. It is easy for the insurance companies to push the costs on to others. I think that 39 months is probably the right balance on that.

Jesse Norman: That is absolutely something we can continue to look at. You do not want to deny people justice. These things tend to be pretty front of mind when they happen, for obvious reasons. Obviously, if there is some ancestral ancillary thing, it may be that the cost of preserving data does not make it impossible.

Q293       Grahame Morris: On a practical issue, to ensure that autonomous and self-driving vehicles are roadworthy, presumably there will still be a requirement for an MOT certificate.

Jesse Norman: The entity that is liable for the self-driving mode will be required to make sure that the vehicle is capable of satisfying the legal requirements at the time, subject to the vehicle itself, which, as you know, coming out of the manufacturer will have certain requirements laid on it. I would expect there to be a roadworthiness requirement for that, for the same obvious reasons that we have talked about. You have to make sure not just that the vehicle has its tyres inflated but that the brakes are in good working order and all the rest of it.

Q294       Grahame Morris: The vehicles will be much more complex and have a much greater degree of electronics content that would be vital to their safe operation. Would you anticipate, or will you be specifying, that there will be particular centres that carry out the MOTs, or would they still be carried out at garages that would have to ensure that they complied with the requirements of being able to safely maintain an autonomous vehicle?

Jesse Norman: It is a very good question. I anticipate that the liability holder would ensure that the technology works, using garages or other checking mechanisms that have been properly licensed for the process, in the way that they would with an electric vehicle today, for example.

Q295       Grahame Morris: Is the liability holder the manufacturer or the software company?

Jesse Norman: It depends. If we are talking about the entity that governs when a vehicle is driving itself, it will be the so-called ASDE—the authorised self-driving entity—and if it is the entity responsible for running the service, like a bus operator, when it is running with no user in charge, it will be that person; otherwise, it will be the user in charge.

Q296       Grahame Morris: You have bamboozled me there. Is it the manufacturer or the software company?

Jesse Norman: It depends on what the problem is.

Grahame Morris: It depends on the circumstances. Thank you.

Q297       Greg Smith: On the MOT question, at the moment, when a petrol or diesel car, or even an EV, goes through its MOT, the things that it is checked against are, as you alluded to, brake pad depth, discs, tyre tread depth, windscreen wipers working, seatbelts working and all of those things. Surely, there will have to be additional legislation to change any future MOT on a self-driving vehicle to bring in, for example, whether the lenses on the cameras and the LiDAR sensors are cracked or scratched. Those are the very things on the vehicle that detect other things on the road or other moving objects. Surely, there will have to be a fundamental change to what an MOT is testing against, in addition to brake pads, windscreen wipers, and so on.

Jesse Norman: I have not reflected on it particularly, but that may well be true, yes. It may well be that we cannot rely on the liability holder alone to be the driver of the change. We will need to have some checking system; that is right.

Claire Wren: We will absolutely need to think about the roadworthiness of vehicles. If you think about something like automated lane keepingthe self-driving technology that we expect may be on the roads this year or next yearthose cars are very similar to the cars that we have on the roads at the moment. They will be MOT-ed in the same way. What you are all talking about is the pods, no-driving-wheel situation, where you will have another organisation responsible for the roadworthiness. The exact processes and procedures for assessing that roadworthiness are really important. As you say, they would need to consider things like, “Have we got all the latest patches?”, as well as—

Jesse Norman: Take the classic case of CAVForth at the momentthe self-driving bus in Scotland running over the Forth bridge. The bus partners are coming together to run that service. They are jointly responsible for its effective delivery, and they will be individually responsible. Alexander Dennis will be responsible for the running of the bus.

Q298       Greg Smith: I do not mean to cut across you, Minister, but my point is a very practical one. If you drop your iPhone—other brands are available—and smash the sensor that makes face ID operate, face ID will not operate, and you will have to go back to putting in a code. Likewise, if there is an incident and an e-scooter drives into a self-driving car and cracks the sensors that are there to detect the toddler on the road or another vehicle or whatever it might be, surely as part of any future testing for the roadworthiness of a self-driving vehicle, in addition to brake pads, windscreen wipers and all the rest of it, it must include a test of all the sensors and whether they are working or cracked. That will require legislation.

Jesse Norman: You can imagine that sensors will become like tyres. Software will not be something that a garage can necessarily scrutinise. For commercial reasons it would probably be highly undesirable for them even to try to, but that software liability will rest with the software provider. This is all about the edge cases on how the different parts of the component relate to liability.

Claire Wren: It is a really important point. In that particular scenario you would want the sensor fixed well before the vehicle went to its equivalent of the MOT. That is where the organisation responsible for the vehicle is responsible for getting the sensor fixed well in advance of, potentially, a year—you do not want it driving around without the sensor for an entire year. The MOT is not the only mechanism. The regular check is not the only mechanism.

Q299       Jack Brereton: I want to talk a bit about infrastructure and long-term infrastructure investment. Is that something that is being considered in adapting our road network for self-driving vehicles?

Jesse Norman: It is a very interesting question. Different technologies place different demands on infrastructure. The technology nirvana is to have a suite of sensors and software that has enough redundancy to be safe but is sufficiently clever not to require very high demands on infrastructure. Of course, that then imposes public costs that the infrastructure provider, the Government or the relevant authority may not be able to deliver, which means the car cannot be run.

At the moment, obviously, they need to be connected and operated on 4G. In some cases, they will have other connectivity, but I would expect that to go down over time. They will become closer and closer to being run on regular roads in the regular way. If you think about automatic lane keeping, as a very simple example, it is important to have white lines in the road that help the sensors to determine where they are in the road. If, peradventure, some failure to keep those lines occurred—

Q300       Jack Brereton: Well, that is a regular occurrence on most of our roads.

Jesse Norman: Probably less so on the strategic road network, but, yes, that will be a demand on infrastructure that will need to be met in order for that technology to work.

Q301       Jack Brereton: We have a very varied road network, all the way through from the high-speed motorway system to a little backwater lane. How is the road network going to be able to cope? Do you envisage that self-driving vehicles will be able to be used on all of the different varieties of road infrastructure that we have in this country?

Jesse Norman: I think you have had testimony saying that the moment when an autonomous vehicle can pick you up from a remote farmhouse in Scotland and deliver you to some place in London, driving entirely autonomously, is a very long way away. One of the issues of equity is the question of whether we will be able to get adequate self-driving services and capability into rural areas so that they feel part of it. I come from a very rural constituency and have plenty of constituents who are already reliant on community bus services, dial-a-ride and that kind of stuff.

This technology potentially offers a really good option. It is quite interesting talking to one of the big companies that I was visiting the other day. They were saying, “Look, we are absolutely up to test some of our technology in those areas because it doesn’t place a lot of reliance on infrastructure. We would like to test it more under specific circumstances and build more experience for us in rural areas.” That is potentially a very exciting development.

Q302       Jack Brereton: Could you envisage a situation where we say that motorways are for self-driving vehicles, but the rest of the roads remain for ordinary vehicles?

Jesse Norman: Again, we are in the realms of the further future. You could certainly imagine that one lane would be reserved for automatic lane-keeping vehicles. If there were so many of them that it could use the lane effectively, in the short run what is more likely is that you will have ALKS vehicles cohabiting alongside regular vehicles. I don’t think there is any reason why that should not work pretty well.

Q303       Jack Brereton: Are there any specific steps being taken at the moment to prepare our road network for self-driving vehicles?

Jesse Norman: There are several key components. It is fair to say that the telecoms companies are aware of the need to make sure that there are proper upgrades to the mobile telephony system. We all know that there are parts of the country that do not get a 4G signal now. If we had autonomous vehicles today, those parts would be inaccessible, which is not acceptable.

I think that National Highways are extremely aware. They are a very early potential use case. They are extremely aware that the national strategic road network needs to be fit for purpose for the introduction of these vehicles.

Q304       Jack Brereton: Is there a specific budget that the Department or National Highways have dedicated to some of these—

Jesse Norman: I don’t know about a specific budget. I am not sure that it needs a specific budget, but it certainly needs them to be able to maintain the highway in a format that does not—

Q305       Jack Brereton: Or, as we have just discussed, maybe to a higher standard than—

Jesse Norman: It may not be necessary to have a higher standard. They may just be required to have the standards we have. It is a question one could put to the heads of the different self-driving companies involved and the different service operators. I do not think they regard themselves as needing anything other than the current quality of infrastructure, appropriately maintained.

Claire Wren: That is exactly the case, Minister. The self-driving vehicles will need to be able to use existing infrastructure. We all want that infrastructure to be maintained, but the self-driving vehicles will be able to use it as it is now. It does not need a separate budget. In the future you might choose to do some different things, such as hazard warnings around corners that the cars can detect. That is a further-away thing, but for self-driving vehicles on the motorway or on the strategic road network, the aim is that they use it as it currently is.

Q306       Jack Brereton: You mentioned National Highways and their awareness of these issues. Are there any specific requests that you have made to them or to local authorities on their network?

Jesse Norman: Within DFT, that is more a matter for the Roads Minister than for me. As I say, and as Claire said, we do not see that there are specific infrastructure demands being made at the moment. I think we slightly underestimate, on the national strategic road network, how much work is being done on different enabling technologies. They are really quite advanced. An interesting session you might want to have could be about the different road-use technologies that they are developing, in the way of sensors.

The other thing to think about is that, obviously, capability differs very widely between some local authorities and others. Some local authorities are fantastic in urban areas and some in rural areas. You cannot really predict. The smaller ones tend to be more stretched than the bigger ones in this kind of area. Certainly, the city mayoralties are very seized of the interest. If you talk to Transport for West Midlands or the authorities in Greater Manchester or Liverpool, they are thinking hard about this, to say nothing of TfL.

Q307       Jack Brereton: As we go through that transition, do you envisage there being some challenges? We are going to have conventional vehicles operating alongside self-driving vehicles.

Jesse Norman: I do not think there is any doubt but that there will be challenges of different kinds. There are all kinds of reasons for thinking that this is going to require very careful attention and management. The maintenance of public consent and adequate communication is really important. We have some quite stringent rules in preparation for people who mislead about the capabilities they have, such as advertising driving assistance technology as a self-driving technology, precisely because we do not want people to blur the gap there and—

Q308       Jack Brereton: Which already exists.

Jesse Norman: Quite. We are already working with different regulators to make sure that that is inhibited and, if necessary, prosecuted.

Q309       Ruth Cadbury: Given that most local authorities can’t even afford to fix the potholes, are we likely to have sufficient digital infrastructure for self-driving vehicles across our road network?

Jesse Norman: That is a great question. You can see a situation in which there might be local road hazards that are conceivably less detectable by self-driving vehicles than by humans. Over time that will become very small. In fact, it should be significantly better. We have requirements to disclose when roads are being dug up. There is lots of congestion information already fed into public networks. We are quite a long way from having autonomous vehicles available for operating in self-driving mode on local roads. It gives us a little bit of time, but there is a longer-term challenge without any doubt.

Q310       Ruth Cadbury: Of course, even technology-assisted vehicles now need decent lane markings. Along with potholes, there is not the capacity to re-mark the roads with white lines. They are fading and are often not seen. Do the vehicles pick up if the lines are worn away—

Jesse Norman: If the vehicle cannot detect lines, it will not engage. I will come back to this example because it is familiar. If you take BlueCruise, there are designated roads. It cannot be engaged on a road that has not already been designated, which means that it has passed the satisfaction and security case for infrastructure.

Q311       Ruth Cadbury: Obviously, it is easier to maintain and incorporate the technology on the strategic road network, which, presumably, will be where self-driving vehicles may well be used first, but you still have to get from the origin to the SRN road and off the other end.

Jesse Norman: That is true.

Q312       Ruth Cadbury: To go back to the digital infrastructure, who is responsible for rolling out and improving the digital coverage on our roads across the UK?

Jesse Norman: Do you mean 4G or 5G coverage?

Q313       Ruth Cadbury: The appropriate digital infrastructure that will be needed to support—

Jesse Norman: It will be mainly that and the cellular telephone companies, I imagine, in the first instance.

Q314       Ruth Cadbury: And in those areas where it is not viable for the phone companies, is it the local authority’s role or national Government’s role? Is it the vehicle manufacturer’s role?

Jesse Norman: I do not think we want national Government or local government running cellular technology infrastructure. We will have to have, by market forces, public pressure or national regulation, an adequate digital signal in all local areas where self-driving cars are going to be used. If they are going to be used or that is being planned, the pressure to make sure that the connectivity is there will be massive. There will be massive issues of equity and fairness raised. I speak with a bit of feeling because there are parts of Herefordshire where I cannot get a decent mobile signal, such as in Three Elms Road in Hereford, and that is in the middle of a city.

Q315       Ruth Cadbury: The UK is well behind other countries on that. You keep referring to self-driving vehicles as though they were private cars, but, of course, logistics will be a major user of autonomous vehicles. Unless we can get full digital roll-out, they are going to have to run a two-tier vehicle fleet.

Jesse Norman: I don’t think it is breaking any secrets, but it is fairly clear that part of the transition process will be the self-driving equivalent of driving your car to the motorway lane, where you can engage automatic lane keeping. In due course you will probably be driving your car through backstreets to get to the place where you can initiate self-driving. At that point, that technology will take over.

I anticipate that the key thing is that as you want to move into self-driving mode earlier and earlier, and as technology allows you to do that, so the pressure on digital infrastructure will become stronger and stronger. That is as it should be. If the effect of that is to put more pressure on the likes of the cellular telephone companiesVodafone, Three, and so on—then that is all well and good.

Colleagues have just told me that the road authorities may choose to develop their own networks in order to improve benefits. Of course, that is true. If a local authority wants to add additional services or to improve connectivity in a way that creates or improves infrastructure that facilitates self-driving vehicles, they are absolutely at liberty to do that. Connectivity is not necessary for everything.

Q316       Ruth Cadbury: In which case, will the Government still have a role to ensure compatibility across systems?

Jesse Norman: They probably will, yes. You can imagine a situation in which we will want to have national standards, in the way that ATE does with cycling and walking now. We want to have high-quality standards across the piece.

Q317       Ruth Cadbury: Finally, in real time these vehicles are going to need to know about other rules-of-the-road changes such as speed limits, bus-only streets, parking restrictions or restricted streets. What steps are you taking to ensure consistency and that that information is available electronically?

Jesse Norman: We have a national data strategy for the use of data. We are thinking very hard about the relationship between data and the use of AI. This is a very current aspect for us.

Q318       Mike Amesbury: The vision set out in the White Paper is to have a full regulatory framework by 2025. It is mid-2023 now. Isn’t it time to get a move on?

Jesse Norman: I think we are still in a very good place and ahead of the game, but we need to get legislation through. We should be perfectly clear about that. That is an absolute priority from my point of view. If we do that, we can certainly look to some self-driving applications being in place on that kind of schedule. I do not think 2025 is impossible at all, no.

Q319       Mike Amesbury: How advanced are the preparations for legislation?

Jesse Norman: The answer is highly.

Q320       Mike Amesbury: Highly?

Jesse Norman: Yes. We have done a lot of thinking about this.

Q321       Mike Amesbury: We took evidence from the likes of Starship Technologies, which the Chair will be very familiar with. In Finland and Japan, for example, they have got their framework in place just this past month. Do you feel that that is going to have an economic impact on investment in the UK as a result if we are falling behind our competitors?

Jesse Norman: No. Thank you for raising it, and it is—

Q322       Mike Amesbury: They say we are.

Jesse Norman: It is a very interesting question. They have been vociferous, with some reason, in making the case for what they have been able to achieve. It is also true that, as you know, although it raises questions about accessibility and disability, it is potentially highly advantageous to people who are vulnerable or have accessibility issues. I am not at all hostile to thinking about that. I do not think it is a settled situation at the moment, but it is one that we need to continue to engage with. I think I have made it clear to my office that I am very happy to do that.

Q323       Mike Amesbury: In the absence of legislation, what are you doing to help the industry?

Jesse Norman: There is a ton of stuff going on at the moment, as you know. You have had all that testimony. The companies continue to test in new user-case environments. There are all the different trials that we are running at the moment. I was out at the Living Lab the other day trying a Nissan autonomous vehicle. There are the trials we have just been discussing in Edinburgh. There are three or four others going on.

There is tremendous energy and drive in this focus. Our goal is to continue to do that, but we need to have legislation. We should be perfectly clear about that. The reason is that we have to get a legal liability regime, which is the core of the long-term competitive edge that the work we have done so far has conferred through the Law Commission. We have to get that on the statute book. That is the goal.

Q324       Chair: One last question from me. In the absence of legislation, are there other changes you could make by altering codes of practice that could help facilitate more advanced trials?

Jesse Norman: We are still very much pressing for legislation. I am slightly reluctant to move away from that fairly single-minded drive, for obvious reasons. If, for reasons that we do not control, that is not possible, of course we will think very hard about alternatives. They may take the form of codes of practice. There may be other things we can do to assist the process, but we have not come this far over seven, eight or nine years in this area not to push the thing with as much maximum energy and drive as we possibly can.

Chair: Thank you. We certainly hope to see a transport Bill in the King’s Speech, because there are a good number of issues that require that legislation. For now, I thank you both for your time and evidence this morning.