Lord Berkeley:

Moved by Lord Berkeley

1: Clause 1, page 2, line 5, leave out subsection (7) and insert—“(7) For the purposes of this Part, a vehicle that travels autonomously does so “safely and legally” if a human driver, who drove in the same manner while undertaking a practical test of driving skills and behaviour in accordance with the Motor Vehicles (Driving Licences) Regulations 1999, would pass that test with no faults recorded by the examiner.(7A) The Secretary of State may by statutory instrument replace the definition of “safely and legally” in subsection (7) with a quantified measure of the risk per mile travelled of relevant incidents as defined in section 39, taking account of data gathered through the performance of the duties mentioned in sections 38 (general monitoring duty) and 39 (duty with respect to incidents with potential regulatory consequences).”Member’s explanatory statementThis amendment replaces the definition of “safely and legally” for the purpose of the self-driving test with a requirement that an autonomous vehicle should drive to a standard such that a human would pass the test with no faults recorded. It also allows for this definition to be replaced once suitable data becomes available as a result of sections 38 and 39.

My Lords, these amendments are all about road safety. Of course, it is a very important subject, which we discussed at length at Committee. Many of the comments made by noble Lords will have been reflected in what I am about to say and in what the Minister said. The Minister has some amendments and I have a couple of amendments in this grouping.

We are all struggling to come up with a definition of “road safety”—which will probably stand for many years—that will enable us to avoid the fear that automatic vehicles will by definition be less safe because they will run into more people. It is a very difficult and challenging subject. My view, and I am very grateful to Cycling UK and other groups for helping with this work, is that we need a step change in road safety. The risks of death or injury on our roads are significantly higher than for life in general, or indeed for other types of transport networks, such as rail. Particularly, pedestrians, people who cycle and other non-motorised road users bear a disproportionate brunt of this risk. I think that this will be a worry all the way through.

I was very interested to hear from Cycling UK and the Parliamentary Advisory Council for Transport Safety that they tried to follow up the work the Law Commission did in this regard—and did it very well. They came up with two options for trying to improve the definition. The first defined the standard required in terms of what would be required for a human driver to pass a driving test with no faults recorded by the examiner. The second was to quantify the risk of a collision or traffic infraction, possibly per something like 1 billion kilometres travelled.

I came to the conclusion that the first one was probably better, which is what is in my Amendment 1. This says basically that the vehicle should be driven—remotely, but driven—

“in the same manner while undertaking a practical test of driving skills and behaviour in accordance with the Motor Vehicles (Driving Licences) Regulations 1999, would pass that test with no faults”.

I think that is quite a good one. It would allow the Secretary of State to change it by statutory instrument if he or she thought that was a good idea.

The Minister will speak to his amendment, which I think is an improvement. It is a question of having a debate on these things. Although I do not think we will finish it today, I hope we can make some progress on the right way forward to make sure that road safety is not reduced; in other words, it needs to be improved.

There are two other amendments that go with this. First, Amendment 2 in my name relates to the types of locations or circumstances where these criteria are met. It is very different being on a motorway from being on a road in a congested town or in the countryside, and it is important that the principles that are applied should have the option of being different for each one.

Secondly, Amendment 4 says simply that we should aim for something a lot better than “better”. Whether

“significantly better for all road users” is the right wording is something that we can debate. I think “significantly” is important, and it is really important that it applies to all road users, which includes pedestrians, cyclists, children, older people, disabled people, and so on.

With that short introduction to the road safety issue in the Bill, I beg to move.

My Lords, I repeat the declarations of interest that I have made in the past.

I applaud the principles behind the suggestions made by the noble Lord, Lord Berkeley. However, there is a difficulty in coming up with new regulations that are different from elsewhere in the world, and I am afraid that “significantly” falls into that trap. It would make it a lot harder for international companies to work out exactly what was meant by these words. There is no established case law on these matters.

We all know that there are problems with existing human drivers, and we should expect that all autonomous vehicles turn out to be dramatically better than human beings. We should not look for circumstances where humans monitor computers but rather the other way around; computers will be better than humans at this. A lot of people suggest that car insurance will actually reduce when the number of autonomous vehicles increases. So I am afraid that I can only applaud the amendment produced by my noble friend the Minister and reject those proposed by the noble Lord, Lord Berkeley.

I hope the House will forgive me, but these various amendments on safety prompt me to ask the Minister about something that has not featured much in our discussions: the issue of hacking into self-driving vehicles—SDVs. It was touched on peripherally during the debate on data protection in Committee but not really highlighted as a major safety concern, which is why I thought I would bring it up now.

I sat on the House’s Science and Technology Committee when it produced its report on automated vehicles some five or six years ago—I am afraid the doldrums of Covid blur my account of time. I remember that during that committee’s investigation, we spent some time discussing in detail the question of hacking into these vehicles, and I felt it only right that it should feature in our discussions on safety today.

We all know how easy it is for someone, or some group of someones, to hack into our computers from a distance, and it could be a criminals or, worse, an enemy state. Why should it not be the same with an SDV? I raised this subject with Waymo and others, but I have to say that I was not convinced by its assurances that it could not happen. We all know that both at Microsoft and here in Parliament it takes a team of experts, sometimes working around the clock, to keep all our devices free from hackers, and an SDV will just be another device.

I was going to bring this matter up when the noble Lord, Lord Lucas, who is not in his place, had an excellent amendment on the obvious necessity for our emergency services to be able to talk to or even control SDVs in certain circumstances. Sadly, however, I could not be here on the 10 January. I was going to say that if it is too easy for a policeman, an ambulance driver or a fireman to get sufficient access to control an SDV, I feel sure that it will not be impossible for someone with malicious intent to get hold of whatever device or code that makes this possible. Could it be that stealing a car will become easier, and that a suicide bomber will now no longer need to commit suicide but just hack into someone else’s car or an SDV for hire and drive it into a crowd or the gates of Parliament, for example? Or maybe you could commit murder by getting control of a car and driving it into your intended victim. It is also entirely possible that no one would know who had done it, because it had been done from a considerable distance—maybe from the other side of the world.

I do not know whether any of your Lordships have seen a series called “Vigil”, one of these television thriller fictions, in which an armed remote-control drone was captured remotely and used to create death, destruction and mayhem on British soil. However, no one knew who was controlling it, which was the essence of the whodunnit plot. Incidentally, it turned out that it was being controlled all the way from the Middle East. I am afraid my thoughts leapt—rather melodramatically, I admit—from that fiction to the reality of what we are trying to achieve here with the Bill.

I am sure there are technical solutions to all these issues, and the whiz-kids on either side of the good-versus-evil divide will continuously compete with one other to win the war of control. It occurred to me, for instance, that perhaps all policemen should be issued with a zapper that brings to a dead halt any SDV that appears to be behaving dangerously. That may be too drastic a solution but, believe me, we will need some solution. My point is that we are entering a brave new world, and we need to properly think through all the problems we are going to encounter. We particularly need to ensure that SDVs become an accepted and safe reality.

I did not want our debate on the safety of these vehicles or the future to pass without a serious commitment from government to being always on the alert to controlling or at least minimising this safety problem. Therefore, by way of a question, I would like reassurance from the Minister that before companies can be licensed to produce SDVs, there will be checks, monitoring and even the holding of emergency real-life exercises with the police to test against what they would do if a dangerous hacker got control of a vehicle.

Will the Government commit to ongoing vigilance over the licensing process, the manufacturers, the operators, the car hire companies, the taxi services and the so-called Uber 2s, and so on, to minimise the dangers from malicious hackers? I realise, of course, that all this vigilance will not eradicate the danger of hacking into such self-driving devices. It is clear that we are unlikely to ever see the end of people trying to get into our other devices, our banking services and the like, but I hope that ongoing vigilance will at least minimise this particular safety risk.

My Lords, following on from the noble Lord, Lord Cameron, I remind the House that I raised national security and people hacking into the system at Second Reading. Group 5 today deals with data protection issues; careful control of data is one way in which to make it more difficult for outside forces to hack into it. However, if you present a complete picture of every road and road sign in Britain to people who are able to drive around the UK, then you are opening a very big picture to the world. There will be people who want to take advantage of that in a way which could be hugely damaging.

I thank the noble Lord, Lord Berkeley, for his amendments. We had a vigorous debate in Committee about issues of safety. I do not know whether the definition produced in government Amendment 3 is absolutely the last word on the topic, but the Government have moved a long way. I thank the Minister for that amendment, which is an advance and improvement on the original. As the noble Lord, Lord Borwick, said, we need to take into account issues associated with international definitions. Government Amendment 7 is also important as a step forward, because it gives this House an important role at a key point when that statement of safety principles is issued.

The Minister will be pleased to know that I took his advice and went to visit Wayve in King’s Cross. Wayve is a local company which is developing a driverless car—an automated vehicle. I went for quite a long drive around the streets of King’s Cross and can report that I found it surprisingly relaxing. I did not expect to be relaxed but I was. I mention this because one key point was made to me during that drive, as we overtook a cyclist very carefully. The key point was that these cars will always be programmed to drive legally; that is a great deal better than you and me as, from time to time, we lapse from the highest standards. Some people out there drive in a way which does not follow the law—they wilfully drive too fast or inconsiderately, and so on.

Another point was made to me, because during that drive, first, we had a very indecisive elderly lady wondering whether she was going to cross at a zebra crossing and, secondly, we had that cyclist. Of course, those users are always going to be there, because even when we have totally driverless cars, which will be decades on, we are still going to have human nature intervening, so this is a very complex issue.

I thank the noble Lord, Lord Berkeley, for his contribution. I also thank the Minister for the steps forward that we have made in improving the definition and the role of this House in the statement of safety principles.

My Lords, I think this group has two subgroups. There is the subgroup of amendments in the name of the noble Lord, Lord Davies, and my noble friend Lord Berkeley’s subgroup. I am afraid to tell my noble friend that we will support the Davies subgroup and not the Berkeley subgroup.

There are many reasons for this, ending with a very pragmatic one. First, the proposals from the noble Lord, Lord Davies, are structurally sound as they separate the roles of Clause 1 and Clause 2. Clause 1, as it will stand after these amendments, in essence says, among other things, that there shall be a safety standard. The clause is headed “Basic concepts”. Clause 2 attempts to address what that safety standard shall be.

We believe that government Amendment 3 is right. It is a very sound definition of “safe enough”. It is built around the well-crafted concept of

“careful and competent human drivers”.

It is today’s standard at its best. It is today’s standard after, as is set out in the commissioners’ report, eliminating the distracted, the drowsy, the drunk, the drugged and the disqualified. It is a high standard but not an infinite standard. It recognises that there has to be a limitation, otherwise the whole pursuit of a standard that is not defined becomes impossible.

It passes what I consider to be the death test. One of these vehicles is going to kill somebody. It is inevitable; the sheer volume of events will mean that something will go wrong. It is at that moment that you have to be able to respond to public opinion, have a standard that is easy for people to understand and defend it. I know this because I have been in that position when running a railway. The 1974 Act that applies to railways demands a standard: that the risk is as low as reasonably practical. It is one of the most brilliant pieces of legislation ever passed. Its impact on safety in this country has been enormous. Its impact on construction and railways, and its crossover impact on nuclear, have served this country well. I believe that this standard, which involves being as safe as a careful and competent driver, is the natural equivalent.

I also note that the law commissions produced three answers. Since they took three years or something to come to these three answers, it seems a pretty good idea to pick one of them. They were options A, B and C. Option C is, in my view, clearly rejected by these amendments. That option was to be

“overall, safer than the average human driver”.

The average human driver includes this wonderful list of distracted, drowsy, drunk, drugged and disqualified drivers. The world is a better place for eliminating them. Option B was

“as safe as a human driver who does not cause a fault accident”.

That is so ill defined that even the law commissions gave up on it. Option A is this one:

“as safe as a competent and careful human driver”.

It passes that test in a way that, when the experts set about turning this into regulations, I believe it will be feasible for them to achieve.

We also support government Amendment 7, which is a compromise. It ensures that Parliament—the importance of Parliament is very much brought out in the supporting documentation—has a positive involvement with the initial statement of safety principles. It also assures us that there will be a negative involvement with subsequent revisions. That is a balance, and we can support that.

I am afraid that government Amendments 3 and 7 have a rather unique advantage that we should not ignore: the name on them is the Minister’s, that of the noble Lord, Lord Davies. But, with the greatest respect to him, if you rub out “Lord Davies” and look under that name, you see “His Majesty’s Government”. Their majority in the other place means that these two amendments will become law—a piece of law that will guide this industry well.

I turn to an issue that is not so directly involved but needs to be there to tidy things up: the principles relating to equality and fairness. What does this mean in this environment? This too is set out in the law commissions’ report. In essence it means that an autonomous vehicle does not come at the expense of any particular group of road users. The policy scoping notes say:

“Government is likely to include a safety principle relating to equality and fairness”.

That is not there at the moment, but I am delighted to be advised by the Minister that this will be changed from “likely to include” to “will include”. This emphasis is particularly important for pedestrians, who must not be sacrificed to achieve the introduction of automated vehicles.

My Lords, I will speak in support of Amendments 1 and 4 in the name of the noble Lord, Lord Berkeley. We dealt with safety a lot in Committee, and it is paramount. This is the most important part of the Bill. I became an enthusiast about automated vehicles because I turned up to a briefing. Most people you talk to are ambivalent at best, and there is a sort of dystopian “Blade Runner” worry about faceless terminator drones.

Safety needs to be beyond reproach when bad things happen. As the noble Lord, Lord Tunnicliffe, said, bad things will happen—deaths will happen. We need to be able to face people and say that we did the best we possibly could. The noble Lord, Lord Tunnicliffe, said this needs to be easy to understand and define; that is absolutely right, but it needs to be equivalent to, or better than, a driver who does the best in a driving test. That does not sound too high to me.

Amendment 4 mentions “significantly” improving road safety. The noble Lord, Lord Borwick, said that we should expect all autonomous vehicles to be better than human drivers, but what if they are not? We need to hold them to account. This would make the whole thing easier to sell to a sceptical public, as opposed to the government amendment. I am not a lawyer, but I do not see why trying to make things significantly better would deter players from joining the market. The industry will spend money on this only when it sees a momentum shift in public opinion, which is why safety is so important and why these amendments are so important.

My Lords, we begin once again with the question of safety. I am grateful to colleagues across the House for their constructive engagement on this issue. The Government’s position remains that the safety standard is best articulated in statutory guidance, with the benefit of consultation. This is the most appropriate way of assessing the public’s attitude to risk, which in turn is the only objective answer to the question of “How safe is safe enough?”. This rationale was set out by the law commissions and is not one from which we intend to deviate.

Nevertheless, I have reflected on our discussions in recent weeks and recognise the strength of feeling on this subject. This is a novel area, with an uncertain future. It is therefore reasonable that Parliament should expect to set the parameters within which the safety standard will be defined. To that end, I have tabled government Amendments 3 and 7. This will establish the “careful and competent driver” standard as the minimum level of road safety that the statement of safety principles should look to achieve—in effect, cementing our safety ambition into law. It will also guarantee a substantive debate in Parliament on the first iteration of those principles.

As I have said previously, the “careful and competent” standard is considerably higher than that of the average driver. This means the objective of a significant improvement in road safety is now baked in from the beginning. Further, I recognise the desire to clarify that this improvement in safety applies to all road users. I can therefore confirm that the statement of safety principles will include an explicit principle on equality and fairness. This could include, for example, a declaration that overall safety benefits should not come at the expense of any particular group of road users. Further detail could then specify that training datasets must be representative of different sectors of society. The exact framing will of course be shaped by consultation.

More broadly, I reiterate the point I made in Committee that references in the Bill to “road safety” do indeed already apply to all road users. This is also the case in existing road safety legislation, where offences such as dangerous driving are concerned with the safety of all road users; this includes, but is not limited to, pedestrians, cyclists, horse riders, motorcyclists and disabled people.

For these reasons, I believe the intent of Amendment 4 is now provided for. Indeed, our proposed Amendment 3 achieves this without the ambiguity created by relative terms such as “significantly better”.

Regarding Amendment 2, Clause 1(3) already establishes that safety is to be assessed in relation to location and circumstances. The safety considerations and appropriate assessment methodologies will vary depending on the location, circumstances, use case and road users in question. It is more appropriate that these details be defined in approval and authorisation requirements, rather than the statement of safety principles.

The first part of Amendment 1 would effectively apply a minimum safety standard equivalent to that of a novice human driver who has just passed their test. The practical limitations of human driving tests constrain the monitoring and assessment of each new driver’s performance to a short time window. These limitations do not apply to self-driving vehicles. We can assess performance in multitudes of situations, including rare ones, and across thousands of miles of driving. We therefore believe safety is best assessed by a combination of real-world, track and virtual testing.

More pertinently, the amendment looks to redefine the phrase “safely and legally” in purely statistical terms. Doing so would contradict the law commissions’ basic principle that these concepts are ultimately defined by public acceptance and public confidence. As I said at the outset, we do not believe it wise to deviate from this principle. I hope that, with the additional assurances of government Amendments 3 and 7, the noble Lord, Lord Berkeley, will agree with me on that point.

Before I conclude, I will briefly address the security point raised by the noble Lord, Lord Cameron of Dillington. Cyber and national security sit at the very heart of our plans to bring self-driving vehicles to UK roads. Vehicles with automatic systems will be subject to detailed technical cybersecurity assessment as part of the well-established type approval process. This will include assessment to ensure vehicles continue to be cyber resilient throughout their lifetime. Before a company can be authorised as a self-driving entity, it must meet requirements relating to good repute, which will include consideration for cybersecurity. We will, of course, be working with the police and the security services to enable this.

My Lords, I am grateful to all noble Lords who have taken part in this short debate. It has been a very interesting series of contributions on the subject of safety, which we will go on debating for a very long time. The Minister, as we know, has moved and made improvements. I will study carefully what he said in his response, because I detect some further studies that may come in future guidance, or something like that. For the moment, I beg leave to withdraw the amendment.

Amendment 1 withdrawn.

Clause 2: Statement of safety principles

Amendment 2 not moved.