Q We now resume our public sitting and will hear evidence from the autonomous driving insurance group, TRL—the transport research laboratory—and the Association of British Insurers. I remind all hon. Members that questions should be limited to matters within the scope of the Bill and that we must stick to the timings in the sittings motion agreed by the Committee. For this session we only have until 10.30 am. Will the witnesses please introduce themselves?
This is a question about insurance, but it is open to any of the witnesses to answer. When automated vehicles and conventional vehicles share our roads, will questions of who is liable for accidents become more complicatedQ ?
I do not think they will become more complicated, because I think the information that should be made available from the autonomous vehicle will make it much easier to establish what has happened. If you think of the sensors that are involved in getting the vehicle around safely, there are traditional cameras, lidar, radar, ultrasound and all those sorts of things; that will give a much more complete picture than we currently have. A lot of insurance claims at the current time are based on different opinions with very little evidence to substantiate them. We still send people out to measure skid marks in the road, for instance; so we will be moving to a much clearer but more granular position. There will be a lot more data, so I suppose it will be more complex in that way, but I think, in terms of establishing who is responsible, things should be clearer.
I would add that it is quite important that we establish with these technologies that that capture of the data that David has described is a requirement of the vehicles. That really builds on the kinds of data that are captured by vehicles today but which are not necessarily available for investigators when it comes to investigating road traffic accidents, which could be very useful for in-depth investigations, in some cases. Therefore, as vehicles become more complex, with a greater ability to capture external data in the moments before a collision, we believe that it is very important that those vehicles are able to preserve that information and make it available to the appropriate and authorised investigators, in terms of understanding what has happened during that incident sequence.
The capability is there. I think we are then drifting into data that motor manufacturers would not necessarily want to share with third parties. They would argue that maybe that driving information is something that they could use for different business purposes. There is currently a big debate in the telematics market about whether there will still be a future for separate telematics boxes being fitted in these vehicles to provide insurance and other solutions when the vehicles are being driven manually; but certainly there would be the capability to record that information.
Q Can I ask about the transition from an automated vehicle to manual? There is a time lag; research has been done into that. At what point is the driver responsible, therefore becoming the insured party, during the transition from automated vehicle to the driver taking over?
My view on that would be that when the transition is from the driver to the car, the driver has to be responsible for what is happening to some degree throughout the whole of that transition phase. Once they have actually got confirmation that the car is in autonomous mode, that is the point when they are no longer responsible. In reverse, when the car is transitioning back to the driver, the same applies, but the driver is not responsible until they have taken full control of the vehicle. I think that is the easiest way to deal with that.
We are involved in a number of the Government-backed consortia. There is Venturer in Bristol; the first trials that were carried out with the Venturer vehicles and in the simulator were with regard to the handover. There are two elements that need to be decided on. I agree with Ben that you should not make somebody responsible until they have fully taken control, whether that is the machine or a human being, but nobody has really worked through that. The other aspect is about making sure that the vehicle has controls that do not try to hand over too quickly. As insurers, one of the things we are very concerned about is that handover. People may be surprised at how long it actually takes a human who has been disengaged to get up to speed, so to speak, so that they are alert enough to be able to drive the vehicle safely. That is why it will take a while for European vehicle manufacturing regulations to catch up, but there will be regulations that require minimum periods and indicators and signalling during that handover phase, because that is essential for keeping these safe.
A key point is that while there are lots of data that other parties—police, investigators—might want, insurers are clear that we only want the data when a collision has occurred to confirm whether the car was in automated mode or not. I do not think we are looking to use the Bill as a way of grabbing loads and loads of data and tracking cars from A to B.
Q I have one last question. If you have all your data from an accident from the software in the vehicle, which tells you that the vehicle performed perfectly, but there was an accident with a driver who is driving the vehicle manually, is the assumption from the insurance company that you are not going to pay out because everything functioned perfectly with the machine?
No, that is the opposite of what the Bill is trying to achieve. There will be accidents on the roads where nobody is to blame, as there are now. If you can have an accident with a human driver where nobody is to blame, you can have that with an automated vehicle. For instance, a vehicle is driving carefully down a road but there is some black ice and it skids off and takes out a bus queue—I know that is a bit of a dramatic scenario—but everything has functioned perfectly. The Bill makes it clear that is an accident—injury has been caused by the autonomous vehicle—and it would be paid for by the insurer. In that circumstance there is unlikely to be any recovery from the motor manufacturer, but the whole point of the Bill is to give the general public the confidence that if somebody is injured, we do not have to worry about whether we are going to claim that the software was defective. If somebody is injured by an automated vehicle, there will be virtually a strict liability on the insurer and we will deal with that claim.
Q Ben, some of us have obviously been round this track before with the previous incarnation of the Bill, and you will know that since then we have made some changes to it. One of the changes was the agreement to define a list of automated vehicles to create clarity. One of the criticisms of the earlier incarnation of the Bill was that that was not clear enough. Do you think we have made progress there?
Yes, I think so. I think the definition that you have used in the Bill is clear. To me, it is pretty unambiguous that we are talking about cars that are being entirely driven themselves. I anticipate that there will be a pretty detailed consultation on how you actually draw up the list of vehicles and define what is and is not an automated vehicle. We are obviously very keen to be involved in that and to provide views. Within the industry and within the Association of British Insurers’ work, we have made a bit of progress in working out what we think the criteria for an automated car are, and those are views that we definitely want to feed in. So, yes.
Q I take that as read, Chairman. There was a lot of debate on Second Reading about the possible safety benefits of automated vehicles, and questions were asked about what that meant for insurance premiums, for example. What is your view on that? Clearly it is early days, but do you anticipate that safer vehicles will lead to easier insurance and lower premiums?
Yes, I think it is very clear. We have a very competitive market for insurance. If we see claims costs coming down, which much safer vehicles would definitely do, we would be looking at a similar effect on insurance premiums. We cannot say exactly what will happen until we have seen the cars in real life.
Yes, it is very welcomed by the industry. I think it is very clear that the legislation and broadly the development of automated driving are something that insurers are genuinely enthusiastic about. In terms of the work we do in the ABI, it is one of the areas where we get the most engagement and interest from our members.
Off the back of the Minister’s question, clause 1(1) defines “autonomous”, or rather the vehicles that could be classed as “safely driving themselves”. That does not seem very tight to me. Does the definition not need to be tighter? Does it not give the Secretary of State an awful lot of power? I do not know whether you have a copy of the Bill in front of you. Clause 1(1) states that autonomous vehiclesQ
“are in the Secretary of State’s opinion designed or adapted to be capable, in at least some circumstances or situations, of safely driving themselves.”
The clause allows the Secretary of State to come up with a list of vehicles that he or she thinks are capable of being driven safely without being operated manually at all. The definition does not seem tight to me.
I would make two points on that. On the one hand, we would obviously want to see robust and good consultation on how that list is put together. We would want it to be transparent and we would want the opportunity as an industry to feed into that. The wording does have an advantage in that it clearly states “safely driving themselves”. One of our views is that we want a clear and unambiguous distinction between cars that are completely hands-off—maybe not for the whole of the journey, but for parts of the journey—versus cars where the manufacturer might be saying, “You can do a lot with the automated functions, but you need to be there hovering over the steering wheel as a backstop.” We do not want those things to be blurred, and the definition in the Bill does that.
If I can make one further point, being on the list is clear—there is a definition—but there will also be a role for insurers to play in thinking about, “We have a claims history and car A is brilliant and has a really good safety record, while car B might not be a very good functioning car, but it has got itself on to the list.” Insurers will want to take a view on that in terms of how they approach those vehicles in offering products.
Q Thank you very much for that answer. Clause 4 describes vehicle software updates. It concerns me that—if this is a vehicle that is completely autonomous—the onus is put on the passenger, if you will, to update the software. They have to agree manually to update software. Is that the right position? Should it not be automatically updated?
I think the onus to do those software updates should definitely be on the manufacturer. They should ensure that the system works, and I think that links back to part 1. The Bill says that where a wilfully negligent person deliberately ignores what the manufacturer wants them to do and finds a way around the manufacturer’s systems and still takes the car on the road, it would be unreasonable to say that that person is still a victim. I think you need this protection in the Bill, but you also need robust measures to ensure that people cannot override the safety-critical updates.
I think “safety-critical” is the key phrase. That places a strong incentive on the manufacturer to say, “If the update is safety-critical, you have to ensure that the driver knows.” We have got to be absolutely clear that there is a distinction between “nice to have” upgrades, that perhaps involve a slight improvement in the maps functionality or something like that, and an upgrade where if you do not have it, the car is potentially unsafe and we have a problem.
Q Finally, while I have you here, the current situation for vehicle insurance is driver-centric. The driver of the vehicle is the insured person. The Bill provides for a situation where the vehicle is insured. Is that definitely the way we should be going? Should there be some more provisions?
I think what we have in the Bill is the right way. When these cars first come to road, most users/drivers will probably use the automated function for 10% or 20% of the journey. That is why we want to keep to a system with an all-in-one approach. The Government have described this as
“a rolling programme of regulatory reform”,
so if we really move to having cars without steering wheels or genuinely A-to-B autonomous cars we probably need to look again at what the right approach to insurance is, but I think the technology is a long way from that.
We think it definitely should apply. I know that there have been discussions between the MIB and Department officials about the correct way to do that, and it will be interesting to see how the Committee approaches it. My understanding is that the reason for not having the MIB scheme in the Bill is that it is not in the Road Traffic Act 1988 either, so the existing system is not directly in primary legislation. I think the MIB will be assured so long as the Government confirm that it is still the ultimate fund of last resort, which it definitely should be. It does not necessarily need to be in the Bill, but we would like absolute clarity on how it will work.
“updates that the insured person knows, or ought reasonably to know, are safety-critical.”
That strikes me as very woolly. I would be grateful for your opinion on where the balance should lie. I accept that if someone has wilfully not installed updates or overwritten them, or something, they become liable. However, if the manufacturer has sent through an update, but the person has not taken it to the garage or downloaded the software—or whatever—at what point do they become liable? I have an update waiting for my iPhone, but I have not got round to doing it. That is not safety-critical, but is there a parallel? Do we need a tighter definition than “ought reasonably to know”?
It is interesting that you mention the iPhone, because that is exactly the debate that we had in our early discussions. Currently, for most things you buy, you have the right to refuse a software update. You are allowed not to get round to doing your iPhone update; you can continue to bypass it. Our view was that when we are talking about a tonne of metal travelling at high speed on the road, people should lose that right, because it would enable them to take risks with other people’s lives. We think the updates should be implemented straight away, because we see them as being improvements. As for whether they are safety-critical or not, it would be a damn sight easier if all updates had to be implemented immediately and the responsibility fell on the manufacturer, but then you are drifting into trying to impose something in UK legislation that some European territories and motor manufacturers have probably not really thought through yet.
The idea of saying that people have to install safety-critical updates immediately is something that we recommend. As for the detail of how it should be dealt with in the Bill, I have to plead ignorance, but the reason for pushing for it is that we honestly believe that if a manufacturer has updated the software, it is to make the vehicle perform better. These are not iPhones that can only annoy other people; these are vehicles that can kill other people. Those updates should be mandatory in whatever way we can make them so.
Q If you were rushing out of the house at 9 o’clock and you got an email from the manufacturer that said, “Version 1.7 is now available and you need to install it”, and you thought, “I’m running late for a meeting—I’ll do it at the end of the day”, would that fall foul?
I think the Bill is trying to allow for some delay, but a reasonable delay, and does not want people to deliberately and unnecessarily stall. If an update is coming through—if they have found a fatal flaw in the software that is likely to make your vehicle veer off the road—my view is that that vehicle should be immobile until the software update is implemented. The motor manufacturers would be able to build that into their technology and machines if they wanted to.
My understanding is that that sounds very good in principle, but how do you define that extent? Many upgrades might have a degree of safety- critical improvements in their nature. How would you define the seriousness of the upgrade?
Clause 4(6)(b) is a definition—that feels to me like it means that it is unsafe to use. If you started saying at this stage a car must be immobilised, we would potentially be legislating for things that we do not know the manufacturers will do in every circumstance. There might be times when the car could move. It might be safe to move it at 20 miles per hour or so—I am just speculating. Is it right to put it in the Bill at this stage? I would definitely say that it is something that needs to be carefully defined and thought about when you create the list of automated vehicles. I know we keep coming back to “the list is everything”, but I think the list is the mechanism by which many of the potential problems of the Bill will get solved.
I am interested in the collection, retention and distribution of data and also the issue that has just been raised by my colleagues about the software prescription; whether it will be prescribed that software must be updated and all the problems that might ensue from thatQ .
Let me lay out an example and ask your view of the Bill. If somebody switches to manual from automated and is involved in an accident while in excess of 30 miles per hour. What happens next? How much of that data becomes available in the case that ensues? For instance, I presume that speed would be used, but what about the on-board cameras or anything else? How much of this data will be kept, retained and used from the functions of the vehicle for a case in which there is an accident with a driver in manual mode? Does the Bill provide a robust framework for accidents and insurance claims and what about road safety? Will it enhance road safety or are we stopping at legitimate information for insurance companies? Should the Bill also include data made available so that road safety is improved?
The basis of the question is around the availability of data. My technical background is in forensic accident investigation and in order to investigate accidents—to get to the root cause—we need to start at the before-accident period and understand as much as we can. We are limited today to things such as skid marks, as David referred to earlier, as the tools to reconstruct those accidents. The kinds of data that are potentially available from electronic vehicles increase the amount of data significantly. With the cameras, radar, lidar—light detection and ranging—and ultrasonic sensors we can get a very clear picture of what was going on around the vehicle at the time of an incident. When we look at the consequences of an incident, we can put the two together and have a very clear understanding from establishing liability and whether that indicates that the vehicle in some way behaved unreasonably—or that the driver, pedestrian or cyclist that it was interacting with behaved unreasonably given the context of the situation. That gives us the information that would allow us to make a determination on liability. I think that is critical to insurers, to police investigating such incidents and to road safety in the future.
To advance the future legislation on autonomous vehicles, we will need a method to understand what is going wrong in the real world. We will also need a method to use that information to improve our understanding of vehicle functioning in the real world and how that can be improved by manufacturers or by legislators applying the right tools to ensure that vehicle performance is improved over time.
If I could add the insurance perspective on that, for what we need to do for this Bill—to establish whether the car was in automated or manual mode—we need a fairly limited amount of data. You mentioned speed, but we do not necessarily need speed to do that. We just need to know whether it was in automated mode. There are potentially lots of other uses for car data for the police and for accident investigators. In a disputed claim with contradictory evidence in court, you could find it a lot easier to solve cases with data, but I would draw a distinction between the data that insurers need to make this Bill operational and the data from cars that would be useful to understand claims. That might be a valid concern for vehicles not covered by this Bill; as cars get more technically sophisticated with more assisted functions, you might want to understand more about how it works for any car. I think whether it is reasonable to ask for data is still best managed via a judge.
It is also important, if we want the data, that manufacturers record it. My understanding is that at the moment, if you hit a pedestrian in an accident, you will not necessarily trigger an airbag so the data that the car keeps on a rolling basis are not automatically recorded or stored and they would not be available. As part of the work to define an automated car, we need more clarity about what data are recorded and stored and about the process to ensure that the data are sent to the right people at the right time. An insurer is one party that would want some of the data.
Q I am interested in your final comment that there needs to be discussion about what data are kept—perhaps the Bill does not go into that in detail. Road safety is important, and some of the data could be used to improve it. I have concerns, therefore, and I ask the question again that you partially answered: does the Bill provide a robust framework for improved road safety through data retention, collection and distribution, particularly with insurance companies? You answered by saying that perhaps it does not, but I wonder what the other two think.
My view is that the Bill undoubtedly aids road safety because it will encourage the use of safer vehicles on the roads, but in terms of data, no—the Bill does not have a robust framework for provision, storage and transmission of those data. I think that is partly because of the stage that we are at. Some things are contentious and some are not. Data sharing is really contentious, whether because of general data protection regulation or because motor manufacturers are concerned about infringement of their intellectual property. We are very keen for there to be some clarity about the storage and transmission of data, the form that data are transmitted in so that they are useful, and the speed of transmission—there is no point us getting the data three months later. That is not in the Bill.
When we had the original discussions, we talked about data. We were still forming our opinions about what data would be required—as I say, that is very contentious. Our view was that it was better to support a Bill that would be part of a rolling programme of legislation and acknowledge that more needed to be done on that data piece than to delay it. We feel that delaying connected and autonomous vehicles hitting our roads would have a negative impact on road safety.
MyQ first question requires just a yes or no answer, and I am interested to know whether the answer is the same from all of you. As has been mentioned, in clause 2(1)(b) there is a reference to the vehicle being insured, which is a new concept, and at the end of clause 2(1), as David Williams said, strict liability of the insurer is established. Do you read that combination as meaning that strict liability attaches to the insurer of the vehicle?
Q My second question: do you agree that it is perfectly possible that the person who is the passenger/driver, who is in the vehicle, might have a different insurer?
There is a chance, I suppose, but I do not think that we would have a dual insurance situation, because the other insurance would insure the actions of that individual, whereas the autonomous policy would cover the actions of the autonomous vehicle. If the vehicle is operating autonomously, it is not being controlled by that driver and therefore they would have no liability.
Q Indeed. My third question: do you accept that strict liability would operate for the insurer of the vehicle, even if the driver had inappropriately transferred control of the vehicle to the vehicle?
Q Yes, so the driver is sitting there driving the vehicle, and he does whatever it is that you do to switch it to automated use, but does so under inappropriate circumstances. Do you accept that under clause 2(1), as it is written at the moment, strict liability would nevertheless attach to the insurer of the vehicle and not to the insurer of the driver?
I think so, because if the vehicle is operating autonomously, strict liability applies. If it is about to crash into a wall and he has flicked the vehicle into autonomous mode, but it has not had the opportunity to take control, we come back to one of the earlier questions—
That is my take on it, but I would say that it is difficult for me to imagine circumstances where doing that would be inappropriate and where someone would still be able to switch a car into autonomous mode.
Q But if I have understood you, that would entirely rely on whether, as a matter of fact, the technology had that effect. In other words, there is nothing in the Bill to prevent that situation.
But we want the man in the street to know that if a vehicle is operating autonomously, compensation will be available. That is why there is strict liability. We might not like the particular scenario, if we can think of one that might happen, but I agree: my interpretation is that strict liability would apply.
IfQ we go back to control of data, Ben, you said that more clarity is needed on what data will be kept and shared. David, you said that the Bill does not look like a robust enough framework, but that you would not want it to be held up. Surely if it is believed that the Bill does not provide a robust enough framework we should look at making amendments, rather than holding the Bill up. Are there amendments that we as parliamentarians should consider?
I am not aware of the planned timetable. There are two aspects: first, the vehicle has to get on the list and insurers then need to decide whether they will insure those vehicles. If, for some reason, a motor manufacturer decides they are either not capable of making or are not going to make any of that information available even if it ends up on the list, it will struggle to get insurance in the UK market.
There are lots of things that do need further discussion. These vehicles are not really going to be on the road for a number of years, so setting out the UK’s intention from a headline regulatory view and commenting that data need to be available while we work on that is one thing. I am not fussed as to whether or not it is an amendment, but it would be sad if the amendment took two years to get through because the motor manufacturers’ lobby blocked it.
I would also point out that a lot of the technical side will be taken up at a UN/ECO global level, so it might not be feasible to define it in the Bill and then have to change it. The more sensible route might be to see how the technical discussions go at global level and ensure that the way the list operates is robust, rather than put it in the Bill.
Q The hon. Member for Hyndburn has talked about safety in relation to these projects looking at data protocols. Should that also encompass looking at safety and how the car was being driven or controlled, whether manually or in automatic mode? It has been said that insurance companies would only want data in the event of a collision, but it would be very tempting for insurance companies to want a lot more data. Even now, you can get an insurance policy for young drivers that involves fitting a black box into the car. The insurance company is obviously monitoring the operation of that car to keep the premiums down. Surely there is temptation to want more data and to do more. It may seem hypothetical, but if you use the internet then your history is used as an advertising and promotion tool. Therefore should there not be strict controls in terms of data control in the future, so that the data are not used?
I think there is a distinction to be made in relation to the data that the insurers would need as a condition of this Bill. The industry would love more data, as that helps with pricing. However, it is appropriate to ask what the insurance company needs and then to regulate that in order to make this Bill work. I refer to insurance companies, but actually it concerns what information the claimant would need for the purposes of verifying whether or not they have the right to make a claim. That is a key distinction. The more data that the insurers can potentially get on a commercial basis the better, but we recognise that there have to be controls on that.
I would add to that: as mentioned earlier, there is a difference between the limited amount of information that an insurer might require to understand whether the vehicle was being controlled by the vehicle or controlled by a driver, and information that could be beneficial from a road safety point of view that could also act as evidence from a capture and perspective point of view. This information will inform future policy at governmental level and potentially at legislative level. That is a more detailed source of data, and it would also be of the type that would assist more detailed investigations of what went wrong if an automated vehicle had an accident.
IQ declare an interest as chair of the all-party group on insurance and financial services. I would like to pick up on two points. Coming back to data: obviously, claims prices hinge on the quick sharing of data. In order to pay a claim, it will be necessary to know whether or not the car was in automated mode. Are there any current technical barriers between insurers and manufacturers that are going to delay that? Are there issues that you foresee causing problems?
We probably do not yet know enough about getting the data from the car to the insurance industry. Some work has started to be done via the Motor Insurance Bureau: as well as being the guarantee fund, they do a lot of data-sharing for the industry. We are confident that once we have data from a car, then the process of getting it to the insurer and settling the claim will be efficient. We would want confirmation that we can get it from the vehicle, but we have already started discussing that with the Society of Motor Manufacturers and Traders. That is something that can definitely be achieved within the timescales required.
Q May I make just one quick point? I echo some of the concerns colleagues have raised about clause 4(4)(b). The Bill talks about the damage suffered by an insured person arising from an accident occurring as a direct result of failure to install safety-critical updates. How would it be assessed whether the accident was a result of an installation not being made? Who would resolve those disputes? Can you see any problems around the insured potentially not seeing it as being a contributory factor to the accident? How would those disputes be resolved?
That relates directly to the point I have just made about the detail of the data. In that scenario, in order to resolve the question you would require a more detailed amount of data than purely who was in charge of the vehicle. It would be a question of what the variety of contributory factors to that collision were, what the vehicle systems saw and what they did in response to what they saw, and whether that can be related back to the functionality of the piece of software that was due for install. You would require a much more detailed set of data to resolve that question.
I have a very quick question, following on, I think, from the answers that were given to the Minister’s question, in which you all said you anticipated that, if this goes to plan, it will see safer vehicles and therefore a reduction in accidents, leading to a reduction in premiums, which is clearly a positive for all those paying them. What assessment has been made, or what is your view of, any likely impact of that on the insurance market and industry? As I understand it, car and vehicle insurance premiums to a degree underwrite other insurance policies across the industry—that is the way it is structured. What impact do you think significant reductions in premiums would have in terms of disruption of the insurance marketQ ?
Lots of work has been done on this by insurance companies and by market consultants, and they predict substantial reductions in the total premium pot. A couple of statistics—we think that 93% or 94% of accidents are caused by human error. I have driven in these machines; they are already much better drivers than most human beings. When we look at things like automated emergency braking systems—that is just one component of what will be the autonomous vehicle of the future—we know that they reduce accidents by 15% and injuries by 18%. So even if they cannot prevent the accident completely and absolutely, because they are braking better and faster there are fewer injuries.
We see a substantial impact. There will probably be a slight increase initially because you will have more expensive gadgets strapped around the periphery of vehicles, but once we see a higher proportion of these vehicles on the road, consultants predict a 50%-plus reduction in the total motor premium market. From our perspective, we are planning in that regard. The good thing is that it will not happen overnight, and therefore as we see motor premiums reduce we can move our staff and our capital on to other lines of business.
According to some of the figures I have seen, 63% of the adult population hold a valid driving licence, so by definition 37% of people currently do not. Of the 63% who do, many are precluded from driving because of health conditions or because they have lost confidence behind the wheel. My question is, first, do you think that this Bill will increase social mobility? Secondly, do you think it will increase car insurance volumes in your marketplace? Thirdly, you mentioned to my colleague that premiums will be lower for automated vehicles, Q so I want to seek reassurances about whether you think the Bill will reduce premiums for people who have mobility problems.
One of the consortia we are involved with, Flourish, is looking at cyber-risks and also at mobility, at segments of society that currently feel cut off—people, who perhaps are disabled, living in a rural area and not able to get out and about. That is one of the reasons we want this Bill to go ahead and are keen to support it. Absolutely, it will support that.
In terms of volumes of cars on the road, there are numerous different models. Overall, the view is that there would be fewer vehicles, because this will enable car sharing on a scale that has not previously been seen, but in terms of number of miles covered, there are diverging opinions. One thing that might happen is that, because it will be as easy to get a car even if you do not own one as it is to get a train or similar, more people will move to transport on the road, which will drive up the number of miles. There are other views that there will be an integrated transport network, meaning that more people use public transport because they are much more able to link into it than they are now. I think the jury is out in that regard.
It will absolutely reduce premiums. The other aspect is that even when we have a mixed car park of manual and automated vehicles, because 50% of those vehicles will be safer, although the premiums on manual vehicles will decrease less, they will be less exposed and involved in fewer accidents, so overall that will have a positive impact from a premiums perspective, even on manual vehicles, as the number of automated vehicles increases.
So what you are saying is that if the owner of a vehicle that is capable of being fully automated updates his or her software, has a vehicle with no warning lights displaying that there is something wrong and uses the vehicle in fully automatic mode in circumstances where it is proper to do so, there should be no liability on the owner, who is actually travelling as a passengerQ ?
Q So you are saying that in a scenario where a vehicle is in fully automated mode on a smart motorway, the Highways Agency suddenly drops the speed limit from 70 to 50 and the car in automated mode is slow enough in responding to be clocked by a police speed camera on a bridge, in those circumstances, the insurers of the vehicle should pay the speeding fine?
We are covering the liabilities. I think they are already there in the Road Traffic Act 1988, on insurers, but it would be extending those existing liabilities to the vehicle. I do not think we are responsible for criminal offences such as speeding now. I think you would have to find another way of—
You may need to look at whether or not there are additional criminal offences associated with automated cars. Certainly, this Bill does not compel insurers to pay speeding fines or any other. Ditto if an autonomous car parked illegally in a parking space. If it injured someone or damaged property, that would be the insurer’s responsibility; if it parked and received a parking fine, that would be the responsibility of the owner of the vehicle or another party. You may need to look at that in legislation, but I definitely do not think the Bill does that at the moment, and we would not support it if it did.
Q But you agree that it would be wholly unjust for the passenger in those circumstances to have his licence endorsed and be fined when he is relying on the vehicle, which is slow to respond?
But in all honesty, if someone was on a smart motorway and had connected an autonomous vehicle, they would be more likely to notice the reduction in speed than you or I would. It is a hypothetical question. I think the point, from our perspective, is that the Bill does not compel insurers to pay these sorts of fines. Yes, there are some other legal aspects that need to be debated, but this is about extending the Road Traffic Act 1988 to provide protection in line with the RTA, not about other criminal offences.
Q As far as the data are concerned, you are saying, I think, that you would prefer clarity. You would prefer some rules to be laid down about what data should be kept and who has access to them. Is that right?
This is very much part of the research and development that industry is doing right now, but the expectation on manufacturers providing access to an automated control system would be that, in that handover situation, the vehicle would be assessing the circumstances of the traffic and the road conditions surrounding it and would accept the handover only if it was able to respond appropriately to that traffic scenario.
Q So the vehicle is constantly taking in data on its environment—the surrounding area—and therefore is switching itself off, making it impossible for the driver to switch over to automated mode.
The vehicle would be expected to be aware of what is around it at all times, and during a process—as it was described earlier—of handover, whereby the person or the vehicle that is in control at a particular time will remain in control until the other half of the equation is ready to assume control, that readiness to assume control can be determined only by sensing what is around it in the specific scenario that the vehicle is driving in, and accepting that it is now able to assume that control in a safe manner.
This is a key point, because there will be many vehicles that can operate autonomously—initially, at least, only in certain environments, certain designed domains. For instance, I would imagine that the first ones that come to market will be able to operate on motorways and dual carriageways. [Interruption.] Exactly. Therefore, if you are travelling down Clapham High Street and you want to flick the vehicle into autonomous mode, it will not accept control.
Q Let me ask about transition, because the studies in Bristol suggested that vehicles slowed down significantly after the transition from automated mode to driver mode, because the driver was being excessively cautious or very cautious, which can lead to more dangerous circumstances. Is that an issue for insurers?
It is something we need to be aware of, which is why we asked Venturer to do handover first of all; I think guidance needs to be provided. I think it is less likely to be a safety risk and more likely to be a congestion risk, but the other aspect is that when we are doing these tests, we are deliberating doing on, off, on, off. In my vision of the future and, I think, the way motor manufacturers are designing vehicles, it will not be like that. It might be that you drive on the country roads because you enjoy that and then you hit the motorway and flick the vehicle into autonomous mode for the next couple of hours. But yes, we need to understand and provide appropriate training and guidance on the handover; that is something we still need to understand more about.
Q As insurance premiums come down, as predicted, in the future, is there a concern or risk that the Chancellor of the Exchequer will increase the Government’s take from insurance premiums, so that all the savings are not passed on to the driver?
Q You talked, in relation to the question from my hon. Friend the Member for North Cornwall about people who are disadvantaged not currently being able to access transport and so on, and subsequently, in answer to the question from the hon. Member for Eltham, dealt with predictable, standard routes. Where an autonomous vehicle is acting more like a bus in effect—it is on a standard journey—presumably it will be particularly appealing to those who currently cannot access transport of their own, because it will be, as you put it, in straightforward mode; even in a straightforward mode, many people currently cannot drive a car, because a visual impairment or a disability prevents their doing so. Is that a future you envisage?
I think there will be, in the same way as there are many variations even to the Uber model now, many variations to autonomous vehicles. I think the advantage will be that you will not have to stick your hand out to stop a bus; the vehicle could potentially come into your drive and then go back out and continue its journey.