HoC 85mm(Green).tif


Transport Committee

Oral evidence: Self-driving vehicles, HC 519

Wednesday 8 March 2023

Ordered by the House of Commons to be published on 8 March 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 233260


II: Ed Houghton, Head of Research and Service Design, DG Cities; Simon Morgan, Chair of Traffic Signs Panel, Institute of Highway Engineers; and Christian Wolmar, author.

Written evidence from witnesses:

DG Cities

Institute of Highway Engineers

Examination of Witnesses

Witnesses: Ed Houghton, Simon Morgan and Christian Wolmar.

Q233       Chair: Welcome to our second panel. For the purposes of our records, could I ask you to introduce yourselves and your organisation?

Christian Wolmar: I am Christian Wolmar. I am a writer and broadcaster on transport affairs, and I am the author of Driverless Cars: On a Road to Nowhere?

Simon Morgan: I am Simon Morgan. I am here representing the Institute of Highway Engineers, which is a professional body whose membership is mainly road engineers working in local government, basically responsible for the safety, improvement, quality and environment of the road layout and infrastructurepeople who are very interested in how their work will relate to self-driving vehicles. I am also chair of the BSI committee that is responsible for traffic sign standards in the UK. I sit on various other BSI committees responsible for road infrastructure.

All of that is voluntary. In my day job, I chair a company called Buchanan Computing which advises local and national Governments on traffic orders for road signs and road markings. It is quasi-legal in that we get involved with regulation and individual traffic orders for particular locations. We advise national Governments, and have done so recently with Transport Scotland on their new traffic sign regulations. We are doing the same thing for the Welsh Government at the moment.

Ed Houghton: Good morning. I am Ed Houghton, head of research and service design for DG Cities, an urban innovation company that specialises in the deployment of artificial intelligence and IoT in urban spaces.

Q234       Chair: Christian, you have been quite vocal in the past on the issue of self-driving or automated vehicles. I use that term in the sense of everything from individual private cars to the automated buses and delivery robots that we were hearing about from our first panel. Could you summarise for the Committee what you see as the advantages and disadvantages of greater use of these vehicles?

Christian Wolmar: Let me pose a question: how many of you came by driverless car this morning? I suspect that nobody did, and my driverless bicycle doesn’t exist either. I am making a serious point, because we have been at this for a long time. This is not a new technology; it is a technology that started with the American defence agency about 15 years ago. There have been constant efforts by driverless car technology enthusiasts and companies to suggest that this is just around the corner.

One of the things I am worried about is that your Committee seems to be set up to suggest that this is going to happen inevitably and what we have to do is adapt to ways of bringing it into use, whereas I think we have to go back a step and ask: what is the feasibility of this? Is it desirable? What is it for, and is it at all likely to come about?

Let me end my opening remarks on one point. The idea that computers see the world as we do is completely wrong. If you close your eyes and open them again, you immediately see a vast array of detail that we are able to adapt to, whereas the computer will interpret a car, for example, as a blue box. Any car is a blue box; any human being is a pink cylinder. It sees a completely different world. I am worried that that is not understood, and there is the idea that, because humans are pretty good at driving, we can replicate it automatically—I can give some more detail about that later—and I am not sure what that is all about.

Q235       Chair: It is a perfectly fair challenge to us. My next question follows on from the last question I asked our witness from Stagecoach. They have a bus that will be automated in how it functions, but currently it has a human being at the wheel so that they can intervene and take over if something happens. You can buy a car today that has very sophisticated driver-assist technology and can automatically brake if it senses that you are too near the car in frontit has cruise control and all those facilitiesbut, again, the human is ultimately in charge and responsible.

What we are exploring as part of this inquiry is whether and how we can move to the next stage, where ultimately the machine is in control. Where would your thoughts be, given where technology is and the technology that could be there to have automated control? How do we square the circle of where the human stops being responsible?

Christian Wolmar: It is very good that you distinguish between driver-assist technology and driverlessness. It is notable that a speaker in your previous panel was very focused on the fundamental problem, which is that you cannot have a half-driverless car. There has to be a moment when you rely completely on the technology. That is the difference between level 3 and level 4. I have some suspicion that this might be insuperable. Obviously, it is difficult to talk about technology being insuperable, but several auto manufacturers have basically given up on developing level 3 because of that transition moment, which is so problematic. As was mentioned, there are different suggestions that the moment of transition should take two seconds or 40 seconds, but the idea is that you cannot have semi-driverless vehicles. Therefore, you have to have totally driverless vehicles.

I have been studying this and looking at all the developments for the last five or six years since I took it up. What is funny about it is that you get all these trials, and the newspapers say that driverless technology will be implemented in so and so. This all started off when I became interested in an Evening Standard article about 10 years ago that said we would certainly have driverless cars in London by now. Your chap from Stagecoach in the last panel highlighted precisely that problem. They will have an operator. They want an operator not only to control the vehicle in case of emergency but because people want somebody on board the bus.

They are going to introduce all of this incredibly expensive technology. The technology is very expensive, with all the LIDARs, sensors and whatever, so what are they trying to do if you still need a human on board? Training somebody to drive a car or even a bus is not that difficult, and it doesn’t take long, so where is the saving? What is the point of this technology? That is one of the fundamental questions that must be asked. For example, why are the Government spending some £250 million on encouraging this sort of technology when it does not seem to solve any particular problem?

Q236       Chair: I have a couple of follow-ups. First, it is not for me to speak for Stagecoach, but they would argue that having their bus captain, who is there to assist passengers, is better value for passengers than having the driver behind the wheel and having to focus almost 100% on driving. Secondly, how would you respond to the point that with humans driving, accidents, injuries and fatalities happen every day of the year? Is it not the case that automating some of it will help reduce the number of accidents?

Christian Wolmar: Again, that was addressed very well by the previous panel. The first question you have to try to answer, which will be difficult for your Committee—it would be difficult for anyone—is: what level of safety are you expecting? Are you expecting a driverless car to be as safe, a little bit less safe or more safe than existing human drivers? Existing human drivers are better than you might think. We are pretty good at it. We can deal with lots and lots of situations. The statistics for serious accidents are easy to look up, but by and large most people end up back home safely from their journeys.

First, you have to decide what level of safety it would be necessary to impose on driverless cars. Does it have to be better than humans? I think it does. Then you get into all sorts of difficulties about how you make it better. Currently, in most of the trials the operator takes over at some point when is an emergency. The rate of takeover is relatively high, particularly in difficult situations. In my book I mention the Holborn problem. Try to get a driverless car to go through Holborn at 6 o'clock in the evening. It is virtually impossible. Somehow, you have to build into the vehicle that it can drive through very difficult situations with lots of pedestrians, cars, cyclists and whatever, and do it safely. I think that might be insuperable because, as in Isaac Asimov’s rules of robotics, the car must not be programmed to injure somebody. Therefore, you have to programme it to stop when people walk in front of it and the like. That makes it very difficult to work out how you can have that level of safety and yet run a car that is at all reasonable in traffic without it having to stop all the time.

Q237       Chair: I have one last question before I hand over to colleagues. Do you see a way of segregating the autonomous vehicle market? You have given the example of the difficulties of taking a car through central London at rush hour, but a very different case would be going along, say, the M6 in Cumbria, if I was driving up to see family in Scotland. Would you see a potential for them operating there? There isn’t a segregated lane, but you have one road going in one direction.

Christian Wolmar: That was addressed very well by the previous panel. The problem is that, if I was going to start my journey from my house in Holloway, where it is very crowded, and then go up the M1 and M6 to Scotland, I would be able to have an autonomous vehicle for some of the journey, but would it be the same vehicle? Is it really worth the hassle? I do not envisage a situation where it would be possible to do that without having to have—I don’t know—two different cars, or people taking over at different points or whatever.

It is important to remember that this started off as a nirvana and a dream for various technical people in America, whose names pop up at different times, of completely replacing the way that cars operate. They presented this picture. I have been at conferences where I have debated against people who were proselytising in that way. They presented it as a completely different world. Your grandparents would be able to be taken to hospital in a driverless car; your kids would be able to be taken there; you would be able to go into the office and send the driverless car home, and so on. It would be an entirely driverless car world. That is not now going to happen. They have admitted that it will not happen.

Then you have to think about the use cases where it is worth while. Obviously, there are airports, airside or whatever; there are some very limited uses, but otherwise is it really worth all the hassle and cost of developing this technology just to help you drive up the M6 a bit?

Chair: Fair point.

Q238       Jack Brereton: How realistic do you think it is that the technology is going to become more dominant, or the dominant technology, in the future?

Christian Wolmar: I am not a Luddite. I use my mobile phone as much as anybody. I no longer watch television; I only watch box sets and so on. I am not a Luddite, but on this one I think the technologists bit off more than they could chew. In presenting that vision of a world of driverlessness and a completely different approach to transport, they made a big mistake. If they had said, “We’re going to develop a technology that will make things a bit safer for you and will enable you to swap lanes driving up motorways, even enable you to have automatic braking systems in certain cases, and so on,that would have been fine and we could see an incremental improvement.

Jack Brereton: We have some of that already.

Christian Wolmar: We have some of that already, and that is all fine, but taking the leap, particularly the leap to level 4, where the car can drive itself all the time, even within a constrained area, might prove to be impossible in rain, at night-time and in places where there are lots of pedestriansin every use case. That might prove to be impossible. I might be laughed at in 20 years time, like the people who originally thought that trains would kill all the sheep when they drove through. I might be wrong, but having studied this for several years, and seen the lack of progress and the fact that every trial still has an operator, remote control or whatever—there are a few robo-taxis out there, but they are very limited—I cannot see it progressing to a state where it changes our lives.

Q239       Jack Brereton: Do either of the other witnesses have any views on how realistic it is that this technology will become more dominant?

Ed Houghton: On some of the points raised so far, I agree about the complexity of the challenge, but I think we fall into the trap of not thinking about consumer choice and consumer perceptions of drivingthe experience of driving and the view of driving as a personal pursuit. The previous witnesses were very useful in setting the scene for how services can be designed around the user. To add to what you have already heard, those users may not necessarily know exactly what they want at this stage, so within industry we have to spend a lot more time speaking to consumers about these utopian ideas.

I think consumers want the idea of a utopia for their towns and cities, and I think that consumers and members of the public want road safety to increase. They also know that using the road as a pedestrian, a cyclist, a user of an e-scooter or a driver is one of the least safe things they can do on a daily basis. We have become incredibly comfortable with a very unsafe pursuit for many people, particularly vulnerable people.

My addition to what has been said is to focus on what consumers want and need, and to understand a little more about the potential for additional technologies around the autonomous vehicle that could be improved by the type of technology we are talking about. The connectedness of vehicles in particular is something humans are incredibly bad at. We are very bad at understanding how other road users behave and engage with one another. Connected technology has the potential to reduce a lot of the errors that come from our perception of the spaces around us. There are challenges in that, and telling that story to consumers is something we need to focus on. It does not help that a lot of the stories we read paint a utopia without really articulating that—pardon the pun—we are on a journey to that utopia. We have to think about it in those ways for consumers to understand it in better detail.

Simon Morgan: The institute I represent doesn’t take a view on whether it is feasible or in what timescale, but we have strong views about what is necessary for self-driving vehicles, at levels 3, 4 or whatever, to be safe and workable on the UK road system. It may come up in later questions, but we are very interested in safety, for example. We say that anything that makes roads safer is a great benefit, because the toll of casualties and fatalities on UK roads is still unacceptably high.

If it can be proved that the use of this technology improves road safety, we are all for it, and we would wish to promote it so that it is available as soon as possible; but it obviously has to be provable and we have yet to see any evidence that that is going to happen. We would say that, rather than just aiming for autonomous vehicles to emulate a competent, careful human driver, they should exceed that. If the promises are that one of the reasons for the technology is to reduce road casualties, shouldn’t it be 50% or 75% better than a human driver, not just equal to a competent human driver?

Q240       Jack Brereton: Witnesses in previous sessions have very much focused on the fact that level 5 remains completely theoretical at this stage. Do you think it will continue to remain theoretical?

Simon Morgan: I don’t think the institute has a view on that. We are interested in the road infrastructure. Our members have many different individual views, and we do not take a collective view on the subject. We don’t have a view on whether or not it is feasible.

Christian Wolmar: I have already said that, but one of the important things is that there might be an attempt to adapt the environment so that level 5 becomes feasible. In other words, local authorities will be expected to make sure all their road signs are properly painted and there are balises to enable vehicles to be interconnected, or whatever. They might have to spend a vast amount of money doing that, and I think that should be resisted.

Q241       Jack Brereton: Why do you think there has been such a focus on this technology? There has been huge hype around it being the solution to so many things. If, as you say, it is not really viable in many cases, why has there been such a focus on it?

Christian Wolmar: If you look at its origins, it came from Silicon Valley in California. Some very bright people with a lot of surplus cash, because they work for technologies that develop monopolies and, therefore, they make super-profits, thought this was the next big idea. They are quite powerful people with literally hundreds of millions, billions, behind them that they can spend on playing with this technology. As you know, America is very car-oriented, so the idea was generated that this could become the next big thing for motor vehicles, and it was backed up by a lot of media hype.

Almost every article I read about it presents a case that is not quite true. It says that self-driving cars are going to be introduced in different places and we are going to have this experiment and so on, and it is going to change our lives in lots of ways. It is quite a sexy subject for the media. It is the combination of a lot of footloose capital, some very bright techie people pushing it, and the media being eager for stories of how we are all going to be in self-driving cars in three years’ time. It is always three years’ time.

Q242       Grahame Morris: I am quite interested in something Ed said about consumer choice. The Transport Select Committee in 2018—I thought it was two years ago, but I have just looked it up, and it was actually five years ago—did a very interesting inquiry into mobility as a service, MaaS. I did not know what it was when we started the inquiry, but I thought it was a fascinating concept. I don’t know, Mr Wolmar, if that is a utopian idea. What is your view on the concept of mobility as a service and the role of self-driving autonomous vehicles?

Christian Wolmar: I really do not want to come across as a Luddite, but I have looked at that idea and, as you say, it was five years ago. Mobility as a service is a concept that has not been brought forward in many places. The idea is fundamentally flawed. The idea is that you wake up in the morning and you have one app, and you think, “Oh, today I might hire a car,” or, “Today, I’ll take the bus,” or, “Today, I might cycle,” and you use the app to decide your different transport modes.

Q243       Grahame Morris: You think it is a flawed idea?

Christian Wolmar: It is pretty unnecessary in this day and age. We have lots of different apps for lots of different things. I don’t think many people wake up in the morning not having decided what mode of transport they are going to use. It is another of those technological fixes for the fundamental problem of road congestion.

Q244       Grahame Morris: There is the other side for town planners and so on, Mr Morgan: reducing the number of cars in the built environment and having to plan for six-lane motorways. If this was the utopian direction, there are all sorts of advantages to it.

Simon Morgan: One of the main advantages of mobility as a service is a greater degree of integration between traffic modes and being able reliably to make a journey that involves a bus and then a train, and then perhaps hiring a bicycle for the last bit or whatever, and to be able to book that and know it is all going to happen and all the different links of the chain will be in place when you need them. That is still, as Mr Wolmar says, quite slow in actually appearing anywhere. I agree that it is less likely that people are going to want software to tell them what mode to use. They are going to want the integration and reliability that it implies.

Ed Houghton: I came here today using Citymapper, which gave me very many different modes that I could use to make the journey. It did not tell me that it was snowing, but it at least gave me a good degree of choice and flexibility, and that is the key to mobility as a service, and where we see usership and ownership being a very different model in the future, particularly for those who live in urban environments. Individuals and families no longer opting to own their own vehicle, but moving towards a shared vehicle. Given living costs, given the cost of travel and given that many of us may no longer need to commute to the office, it makes sense to see the model of vehicle ownership changing. That then raises the question: why should I have a personal vehicle outside my house that I use once a week to go to Sainsbury’s when, actually, Sainsbury’s can deliver to me?

We have an opportunity to think about this in a different way, and consumers, through our surveys, are telling us exactly the same thing. At DG Cities, we have conducted research with consumers for Project Endeavour and for D-RISK, which were two Innovate UK-funded autonomous vehicle projects. We conducted service design processes with them where we came up with ideas. We ran workshops and developed solutions to some of the issues that they are facing. Through that, it was clear that there is a massive opportunity to tackle some of the barriers they face, particularly vulnerable or elderly people who may have disabilities and cannot access transport today, or who may even fear using transport today because of the quality and standards that they currently experience. We may have erred towards not thinking about the utopia of level 5, but on the road towards level 5 there are real opportunities to radically improve the quality of our transport system and to help consumers make much better choices.

Theoretically, mobility as a service is a possibility that is coming into practice now. We see it emerging through new apps that are coming online all the time. With self-driving, it is going to become part of the way that we use different services and modes. Level 5 will not necessarily happen in three years. I agree that three years is what we always read in the press. If you talk to the Oxboticas of the world, they will probably say it is more likely to be 10 or 15 years. Somebody else may say it is more like 30. The gradual automation of our ways of using different modes of travel will continue, and we need to engage consumers at every step of the way to do that. It is very important that they are part and parcel of that process.

Q245       Mr Bradshaw: Christian, can you find a historical parallel between the development of transport technology and the practice of your rich, clever people spending a lot of money and pouring it down the drain?

Christian Wolmar: No, it is odd. I have spent a lot of time thinking about why this obsession has carried on for so long. It is waning a bit, it must be said. My comparison with past times is slightly more mundane. If you look at pictures of Parliament Square in 1920, you see bicycles, you see rather more horses, admittedly, than anything else, you see trucks, you see buses and you see trams; by and large, if you look out there today you see the same methods of transport.

A lot of the technical developments in transport that we have been promised such as hyperloop, mobility as a service, maglev and other technologies—a lot of the technological promises in transport—have not materialised. If you remember, we were going to get to Australia in an hour and a half in rockets and so on. We are, essentially, moving in much the same way as we did in 1920. I do not know why those guys thought they could completely change it. As I stress, it is the footloose capital that did it.

Q246       Mr Bradshaw: Ed, you talked about the importance of consumer desires and behaviour in the context of people who drive. The implication of what you said, as I took it, is that there are some people who quite enjoy driving and the autonomy it gives them. There has been a big change, hasn’t there, generationally, in that a lot of young people—I am not young, but I am in that camp—would rather not waste time driving if they can be taken somewhere, either by public transport or in an autonomous vehicle, or have something delivered to them, which means they can use that time productively? That is a big change in human behaviour, which is going to have an impact on this, surely.

Ed Houghton: It is. Societally, we place driving licences, gaining the freedom of driving, as almost a part of our identity. We progress as we age. The vehicle becomes part of how we gain freedom. For many that has been the way. Particularly for those in rural or semi-urban environments, part of identity is gaining the freedom of having a vehicle and doing their driving test at 17. Then in later life, we see it as part of identity in reducing freedoms and no longer being able to drive.

The challenge that industry and you, as political leaders, face is that you have to disconnect the vehicle, the object, from somebody’s identity, but we see that happening. Young people now no longer see it as a valuable commodity to own. They would much rather be more sustainable and think about their emissions. They would much rather spend their money and their time doing things that are more pleasurable and interesting to them. That is not to say that is the way for everybody, and that is why choice is important.

There is a very large population of people whose identity is in the vehicle they own. They gain pleasure from driving. They enjoy sharing their experiences with others in driving clubs. They may be part of a club where they have a particular interest in a particular make of vehicle. That is also incredibly important, so we should not lose it, which is why it is about choice. It is about flexibility and enabling consumers to make a choice of whether they want the freedom of gaining a driving licence, or whether they would prioritise other interests. They would rather gain pleasure in being more sustainable. Maybe that is more interesting and important to them. It is a real challenge, but we are also seeing younger people being far more aware of the importance of being sustainable and of the damage that motor vehicles and car ownership do to the environment. That is now incredibly important to many people.

Q247       Mr Bradshaw: There is a generational divide. Do you detect any other differences, social or economic, in public attitudes towards automation?

Ed Houghton: In our research on the Project Endeavour programme, we ran service design workshops with communities, and we talked about the concept of a ride-sharing taxi. Within that, we talked about using the vehicle at different times of day through different environments, and there was a definite gender divide in people’s perceptions of safety. We see that as an issue in many other aspects of transport. Women were particularly uncomfortable with the concept of ride-sharing autonomous vehicles.

That raises the question of what you can do as a developer of the vehicle to help women to feel safe to use it at a certain time of day in a certain environment. What can we do to educate and inform the wider public about safer behaviour when using the road? How can we educate people to make the right decisions and to call out things when they occur? We can rely on the technology to improve safety, and that is very important, but we also see that there is an education piece about using transport safely. The gender divide is particularly important when it comes to safety.

The second key trend that emerges quite often, not only in this space of research but in wider transport research, is that as we get older we become far more confident in our driving ability. It is not necessarily that we are better drivers, but we believe we are better drivers. Elderly people in our surveys were far less likely to use or adopt a self-driving vehicle service until the point when they lose their freedom, and then we saw a massive increase in those responding that they are very interested in a self-driving service. That is all about choice. It is all about understanding at which point in the journey of an individual they are likely to use a service.

Those trends are important when we think about consumer engagement, service design and wider policy and regulation, and that we do not see it as a one-size-fits-all service, but we make sure that consumers and their different identities are reflected.

Q248       Mr Bradshaw: As it is young people who are more open to automation, and young people are the future, demands and appetite for automation will grow, one would assume, but there is still a role for active government in educating to address some of the other issues and problems that you just referred to.

Ed Houghton: There is certainly a role for education, information sharing and myth busting on quite a lot that is published on the utopia of self-driving vehicles, and helping individuals to understand their role in using the service as a responsible user. That is very important. There certainly is a lot more education and information sharing that is needed. That is not only with physical trials. It is unfeasible to do a physical trial in every single town and city across the country—it is economically very challenging—but there are ways that you can inform and educate through virtual reality.

With Project Endeavour, we developed a film of a live vehicle driving around Oxford. We then developed a virtual reality offer where you could look at that and experience sitting in the vehicle from the comfort of your own home. That was really useful because we could distribute it far and wide to schools, local employers and families, and they could sign up online and download it. That is the kind of information and education that should be done. It is not only classroom-based; it is wider, where we can share more information and different modes.

Q249       Paul Howell: I want to follow up on a couple of points, Ed, before I come to the point I was going to raise. In the decision-making space, there are not just the differentials you talked about, but regional differentials. I am a north-east MP. We are a bit further from the cities. I am looking at my phone at the moment; there is another bus route being cancelled. This is about accessibility for more people. Is there a space for some of the more automated vehicles meaning that more people can retain the capacity to get to work, to the shops, to friends and to social events, leisure, and so on?

Ed Houghton: There certainly is regional variation in the way you would design services. It is good to see that the most recent round of funding in innovation is looking at projects across the country. Specifically in the north, there are a few projects that are really important and will demonstrate locational differences. Looking at attitudes and perceptions of different use cases is very important.

It is a real challenge because, within those environments, the type or mode of transport that you use will be about accessibility for you and choice for you, and many do not have the choice that you would have in a city like London. I expect that service providers would go into those kinds of environments later down the line. The Stagecoaches of the world may prioritise other environments to introduce new technologies. It is not going to be that you open the door one day and everything is automated wherever you are in the country. It is certainly a transition that will occur over many years.

Q250       Paul Howell: I want to go back to the safety agenda, particularly to look at, as Simon touched on earlier, what level of safety should be expected compared with the standard competent driver. Is it the same? Is it 50% better? I want to talk to the other two gentlemen for a while, Christian, because this is based on the assumption that the project is coming, as opposed to whether it is going to come or not, which is more your conversation. I will continue with Ed and then I will come to Simon. Then I want to digress a little bit.

Ed Houghton: Very briefly, we asked that question in our interviews. We ran a trial with participants who sat in a vehicle that drove around the borough of Greenwich. It was completely autonomous with a safety driver in the front seat. Afterwards, we interviewed them and asked them several questions. One of the questions was, “If we took out the safety driver, would you still do the trial? Almost every single person said no, they wouldn’t. It places the safety driver as critical to how people currently view the technology.

The second question we asked was exactly your point: “How much safer does it need to be for you to want to use it over the longer term? People said, “It needs to be twice as safe or 10 times as safe for me to use it.” The best response we had was, “It needs to be Lewis Hamilton for me to sit in that vehicle. It needs to be F1 driver quality to be able to deal with whatever happens on the road. That is the level of expectation that consumers have. That is how they see it. It cannot be just as good as what they currently experience. It has to be the best they could ever experience for them to be able to trust it.

Simon Morgan: We are interested in the actual level of safety, not so much people’s expectations, although there is evidence that whenever people are not in control they expect a greater level of safety. There is not as much outcry as there ought to be about the carnage on the roads, but when you get a rail accident or an air accident, or even a coach accident on the road, it is massive news and there is a massive outcry, quite rightly, notwithstanding the fact that those modes, rail and air particularly, are hundreds of times safer per kilometre travelled.

The actual level of safety, trying to reduce the toll on the roads, is the one that interests us. We do not have enough expertise to say which level of autonomy is likely to deliver that. Even simple driver assistance, as is available now, should hopefully reduce road casualties, but we are not sure of any research or evidence that points to that. Our argument is let’s actually see some evidence of it, rather than simply saying to the public, “What do you think?”, because, of course, people do not know at this stage.

Even though we are talking about the carnage on the roads, it is, as Christian says, extremely rare to have a road accident considering all the kilometres travelled. People, on the whole, manage incredibly well. It is actually quite a high standard that autonomous vehicles have to reach and exceed to beat a good human driver. Nevertheless, multiplied by all the millions of kilometres travelled by drivers on the roads, it comes to a sizeable and unacceptable figure of casualties.

Q251       Paul Howell: I think we can all empathise with the fact that, if I am sitting in the driver seat, the car is clearly safer than when my wife is sitting in it—

Simon Morgan: Yes.

Paul Howell: —and vice versa in terms of our perception. It is not necessarily reality. I am not saying I am a better driver in any way, shape or form, but the difference between perception and reality is important. We have to make sure that, when we are putting legislation in place, it is more about reality than perception. That is the point I think you are making, and I support that.

In terms of safety in the space of use, if you look at the impact of an incident in a town or a village, it is probably going to be a small incident with one or two people who are possibly seriously hurt, whereas if it is an incident on a motorway there will probably be multiple vehicles and things that catch the headlines. Conversely, when we are talking about this technology, it seems as though the technology is more likely to be developable in a motorway-type environment where it could be applied, as opposed to being in the busy environment of a town centre.

Is there some space where we could end up talking too much about a utopian future that is never going to be delivered, and we try to legislate for that instead of trying to deliver incremental changes so that you have, as an easy example, a situation where a lorry driver gets on to a motorway, switches on the automotive things that control his vehicle until he leaves the motorway, and therefore you have increased safety in that high-speed environment and reduced the risk of a multiple incident, or do we wait for everything? Is there a way through that? I am not sure whether the vision of utopia is stopping the legislative process for incremental gains.

Simon Morgan: I have difficultly answering that question. What can be achieved with different levels of autonomy is not my area of expertise. I am interested in how whatever level is proposed or is in actual effect interacts with the road infrastructure. I do not know how feasible it is.

Ed Houghton: If we are thinking about it as a transition over time, it would make a lot of sense to focus on different types of urban, semi-urban or rural space. The interesting thing about this area is the way in which the data will help to create the wider system learning. How we support developers to create, essentially, a database of experiences that you can use to train vehicles across the system is a very interesting question.

We have done work on Project D-RISK, which is a driving test, essentially, for self-driving algorithms. It says, for your example, that there will be types of urban environment or rural environment that are highly complex and for which we need to build a very robust dataset to be able to manage. There are others where you are far less likely to have high risk because they are far less complex. That is where we could probably deploy a technology much sooner. That kind of incremental view, the maturity of the technology and the maturity of the database, is what I think will enable that transition to happen.

The D-RISK project has been valuable because it has allowed us not only to think about the wider database of what we call edge cases—high risk, high-impact accidents that can be very severe—for training the vehicle and the AI, but it is also a very good way of communicating what is happening in the vehicle to the consumer. When we talk to the consumer about a driving test for self-driving technology, it is very powerful and helps them to understand that the system is learning, and that is very important. It is a good way of articulating to consumers the potential of the technology.

Chair: Can we have quicker questions and answers? I am conscious of time.

Christian Wolmar: Can I make a very quick point? Your typology is very good: looking at the nirvana might get in the way of sensible legislation. The only problem with your example is that, of course, motorways are already much safer, and therefore they are not really where we need to focus road safety issues.

Paul Howell: No, but it might be where we can, as opposed to where we cannot, and that might be where the point goes. I could digress for a while, Chair, but you are obviously pushing me to move on to other things, so I will revert to you.

Chair: If we have time at the end, you can come back, but there are one or two other areas we want to cover.

Q252       Mr Bradshaw: Chair, forgive me, but I have to go for Question Time, so I may not get through my section, but I am sure you will pick up if I don’t. I want to drill down, Simon, to some of the physical infrastructure that you think is necessary to push this whole agenda forward.

Simon Morgan: That is our main concern and interest, and in many ways we welcome the discussion on autonomous vehicles and the push for it to move forward, because it bears very closely on issues close to our members’ heartsnamely, the quality of the roads and the signs and markings on them.

You are probably all aware that there is an enormous maintenance backlog. It has been estimated at £12 billion, and that just relates to the structure of the road itself—the road pavement. You can probably add half that again if you want to add lighting, signs, road markings and so on. Most of our local authority members feel constrained by budgets, and they cannot do all the road maintenance they would like to do. They are aware that there are road markings that are badly worn and need replacing, and signs that are far too old and are no longer reflective at night or are hidden behind trees that need cutting.

Anything that helps an autonomous vehicle, partly or fully autonomous, to navigate the road also helps the human driver, so we are terribly enthusiastic about any initiatives in this area that enable those working in local government and on trunk roads to have greater resources to improve the road network for the benefit of everybody. That is one aspect; the physical signs and markings that are already important to human drivers and to the driver assistance technologies that are already available.

There is also the issue of how you know the rules of the road. A key example is speed limits. It is only a short time away that cars manufactured in Europe will be required to have an in-car display of the current speed limit, but how do you find that out in the UK? You can’t. If you have the very latest satnav and you buy a dataset from Ordnance Survey, it is only about 85% accurate. You can have a car with forward-facing cameras that attempt to pick up the signs, but they do not add a great deal of reliability.

There is a big issue for Government in making the rules of the road available electronically in various ways, for the benefit of both human drivers and any autonomous or driver assistance technology. That is, to be fair to the Department for Transport, moving forward, and there are initiatives on setting up data standards and so on, but we are still quite a few years away. It needs to be UK-wide, whereas the Department for Transport’s remit focuses on England; it needs to bring it in for the devolved Administrations as well.

That will be needed to help not just with speed limits but with, “Is this a bus-only street?”, or, “Can I make a right turn here?” All that needs to be available in some sort of machine-readable form, not just relying on picking up the traffic signs.

Q253       Mr Bradshaw: A huge investment in something that one of the other panellists thinks is a bit utopian.

Simon Morgan: All of this investment will benefit human drivers, so it would not be wasted, even if autonomy is slower in coming than we think.

Q254       Ruth Cadbury: Simon, what will be needed to ensure that self-driving and conventional vehicles can share the roads?

Simon Morgan: It is an interesting question. It is obviously for the technology developers to work out what they need to do to programme their vehicles to identify other road users—self-driving vehicles or horses, pedestrians and cyclists—who will never be connected to any sort of network. We hope they have algorithms that, with their various cameras and LIDAR devices, can pick up the other vehicles and road users and make appropriate decisions to avoid collisions, and possibly be clever enough to try to anticipate what other drivers are trying to do. We rely a lot on clues and signals from other drivers, people making hand signals and flashing their headlamps, and things that are probably unofficial signals. Even just a nod or a smile from another driver is something on which we rely a lot in urban areas, but that, clearly, will be absent in the case of high levels of autonomy. It is up to the providers of the technology to try to replicate or find some substitute for that.

Ruth Cadbury: I seem to remember my driving instructor when I was 17 saying, “Don’t wave people on.”

Ed Houghton: It is about education and information for other road users on what to expect, how to behave and how they expect the vehicle will manoeuvre. It raises a question about how we mark which vehicles have this technology and which don’t, and whether that is a useful thing to do for wider acceptance. There is a piece of work going on at the University of Nottingham at the moment, I believe, that is looking at how you display whether a vehicle has that capability and what the impact of it is on other road users. As well as all the infrastructure changes, there is also education and information for other road users.

Christian Wolmar: This has been a big issue in America in places where there is a high concentration of autonomous vehicles. Some local residents are driven absolutely mad by the fact they drive very conservatively. They are trained to be extra careful and often stop unexpectedly. That is quite a big issue. The idea of a mixed park of autonomous and conventional cars is a very difficult issue to crack.

Ruth Cadbury: And that is a driving culture that is more conservative than we are here.

Christian Wolmar: Absolutely, yes.

Q255       Ruth Cadbury: To what extent is the underinvestment in basic stuff like white lines and signs already a challenge and likely to be a significant challenge? Would I be right in saying that, if we are to have higher levels of autonomous vehicles, we need 100% good signage, good road markings and clear kerbs everywhere they are going to be used?

Simon Morgan: I would say so. On main roads divided into lanes, it is obviously essential, even for current driver assistance technologies, that the lane lines are clear and visible in all weather conditions. A survey by the Road Safety Markings Association came up with the fact that about 50% of road markings are seriously below standard even on trunk roads and main roads, so, yes, it is already a current issue. Vertical signs are perhaps less so, if there can be a way of providing the equivalent information electronically to vehicles, but that is still many years away, in my judgment.

Q256       Ruth Cadbury: We have covered driving and parking laws, but is it clear how driving and parking laws will be policed with self-driving vehicles?

Simon Morgan: We think it is not clear, and we are very worried about it for two reasons. Members of the IHE are very concerned about road safety, and, of course, people following the rules of the road is crucial. Our members are in authorities that do civil enforcement, mainly of parking controls like waiting and loading. Increasingly, since it became available in the middle of last year, authorities can apply for a designation orderauthorities in London have been doing it for 15 years or soto enforce moving traffic restrictions themselves, such as banned turns and box junctions.

Our members are quite concerned that no attention seems to have been given to what will happen about civil penalties with autonomous vehicles. Quite a lot has been said about civil liability in the event of a collision and the equivalent of criminal sanctions for a human driver, but nothing much about how an authority would recoup the money it is used to receiving for a vehicle that stops in a box junction. If the vehicle is autonomous, whose fault is it? Is it the fault of the autonomy or the human driver? That is not as simple as you think in a box junction. It ought to be whoever is in control of the car at that point.

If a vehicle goes down a street that is marked “bus only,” and all the low-traffic neighbourhoods and school streets increasingly come into this category, whose fault is it? Is it the fault of the autonomy, in which case you send a bill to the company that wrote the software, or is it the fault of the human driver because they pressed a button that overrides it in some way and says, “I am allowed through that, so drive through there anyway”? There are lots of interesting legislative challenges that we do not think have been fully addressed.

Q257       Ruth Cadbury: There are a couple of questions that arise in terms of the road infrastructure. Given the investment is from the private sector now and probably in the future, where will the responsibility lie for bringing the road infrastructure up to scratch?

Simon Morgan: Interesting question.

Ruth Cadbury: That’s a good answer.

Simon Morgan: It is something for the Government to ponder. Could they make some additional levy? Of course, there are already various ways in which revenue from road traffic comes to Government, but it is not ring-fenced in any way; it is treated nowadays as general taxation. Is there any way that additional revenue would be raised specifically to improve road infrastructure for the benefit of autonomous vehicles? I am not sure whether it is possible, but it is something for the Government to think about.

Ruth Cadbury: It is on the record now.

Simon Morgan: Indeed. Local authorities would obviously like increases in their budgets to help them bring the roads up to the standards they would like.

Q258       Grahame Morris: I think the panellists were in for the previous panel. You may recall that, when we asked about the legislative framework, Lisa Johnson said it needed to be in place now. Mr Wolmar, is your vision less optimistic, that there isn’t such pressure to get the legislative framework in place following the White Paper? Dr Khastgir mentioned the developers, the investors, the potential to provide employment and to grow a whole new sector of high-skilled, well-paid labour. What is your view on the evidence we heard earlier?

Christian Wolmar: Lisa’s little robots at Starship are fine, but they are a completely different subject. In fact, they are a bit of a distraction from your considerations because they are little delivery vehicles that use both the pavement and roads. They are not really relevant, although one interesting point she made was that there is a problem of cars stopping for them and the robot not understanding, which raises issues.

On the legislation, it is interesting. The Law Commission report is something like 300 pages long, and I have looked at some of it. None of the fundamental questions has been answered. The fundamental questions are what safety level you want and who is responsible, as has just been raised. That is why the legislation is very difficult. I would like to see legislation as well, and I think it ought to be quite restrictive and directive, to ensure that safety considerations are right at the top of the agenda, but it seems to be very difficult to answer the fundamental questions. They need to be answered first before you get the legislation.

Q259       Grahame Morris: On levels of safety, the figures I have seen for June 2022 show that 1,760 people were killed in road traffic accidents, and in 2019—before the pandemic—if you add fatalities to those who were seriously injured, it was 29,804. That is not an insignificant number of people killed and seriously injured. Mr Houghton, what is your view on the legislative framework and the timetable we are looking at?

Ed Houghton: When we talk to consumers about assurance, regulation and legislation, having the right policy in place to emphasise safety was critical to them. Among the things we tested were driving licences for self-driving vehicles; demonstrators that show exactly how the vehicles are operating; MOTs on a regular basis, almost like a virtual MOT to demonstrate how the vehicle is making choices in its behaviour; and open-source access to help alleviate some of those challenges.

The legislation has a real balancing act not to stifle innovation, to ensure that industry can continue to progress, because it is progressing as we can see, and to assure the public that safety is paramount. Certainly, from our research with consumers, it is a top priority for them. They will not trust a vehicle provider. In the end, it is down to the Government, essentially, to give the consumer backing that a vehicle is safe to use. They are very aware that private organisations will be selling ideas to them, and they want to make sure there is a real mark of safety and trust that they can see.

Q260       Grahame Morris: Is there a role for the devolved Administrations and for local government? I am thinking of some of the elected Mayors who are particularly interested in transportfor example, Andy Burnham in Greater Manchester and Steve Rotheram. They might want to trial schemes. Are you working alongside them or encouraging those initiatives?

Simon Morgan: In the main, they are curtailed by national legislation. There is a limit to what they can do by means of local traffic orders. It is something on which there is a need for legislation, but it needs to be targeted at different areas rather than just picking up the whole autonomous vehicle issue and thinking, “Thats too big. We’ll leave that for a while.”

There are initiatives that are currently considering micro-mobility. Are we going to allow the wider use of electric scooters, which are currently just in trial areas and only for hire? If so, do we keep them off the footways? Talking of keeping vehicles off the footways, are we going to legislate to stop people parking partly on the footway? That is the sort of legislation that might pick up on the delivery pods that use the footways and that we heard about earlier from Starship. That is the sort of legislation they need, rather than anything bigger to do with the whole autonomous vehicle arena.

Grahame Morris: There are several issues on which there is cross-party consensus: pavement parking, electric vehicles and the legislative framework for self-driving vehicles. We desperately need a transport Bill. I hope the Government business managers are watching this and make time available.

Chair: I thank all three witnesses for their time and evidence today. These are fascinating subjects. We could spend much more time debating them than we have room for today. Thank you for your time. It has been really helpful and will usefully inform our report and recommendations.