My Lords, I am delighted to follow the noble Earl, Lord Selborne, because it gives me an opportunity to thank him for his leadership of the committee, which he did with his customary firmness, urbanity and unfailing courtesy. He had to deal with some fairly unruly committee members at times, which he did with great dexterity.
As the noble Earl has already pointed out, the public interest in autonomous vehicles focuses largely on road vehicles and surface transport. This is understandable, but in many ways this is probably not the main area of change in the immediate future: seaways and the air tend to be less congested and more firmly under external traffic management control. I suspect that in both those areas we will see seriously autonomous vehicles in the not too distant future: this looks like an opportunity for air freight and sea freight in particular.
As the noble Earl, Lord Selborne, pointed out, the Society of Motor Manufacturers and Traders produced the rather useful table, which we reproduce in our report, with the levels 1 to 5, and level 0 being zero automation. In fact, as far as surface vehicles and motor vehicles are concerned, we have been living with what I would describe as creeping automation for quite a long time. With the first vehicle that I ever owned, as a student, one had to adjust by hand not only the fuel mixture but the timing, and of course double declutching was standard. These are things which have moved into our vehicles and made them easier to drive—almost imperceptibly—and we simply take them for granted.
Anyone who has had the opportunity to drive one of the relatively modern vehicles that are around today will find themselves with all sorts of driver-assisting technologies which we did not dream of 10 years ago. We have already mentioned automatic parking; there is also automatic switching on of your lights when the external light drops to a particular level and automatic switching on of the windscreen wipers—a whole range of activities which make driving easier. Particularly valuable, I think, is the so-called lane assist, where sensors sense the track and alert the driver if he drives out of a marked lane on the road. These things are coming. Levels 1, 2 and 3 of the series that the noble Earl, Lord Selborne, described represent increasing levels of automation, using these and probably other devices and assists that we have not yet thought of or certainly are not aware of, with less and less driver involvement.
It is worth pointing out that modern passenger aircraft are probably somewhere between levels 2 and 3. It is well known that passenger aircraft travel on autopilot for long distances. Indeed, it is perfectly possible for modern aircraft to land automatically; take-off has a few more complications. But we are very much in those areas.
The existence of these levels perhaps suggests that the differences between them are more or less equidistant, of similar length, and we just progress from one to the other. But in fact when one moves from level 3 to level 4 there is a big jump in the technology requirement, and certainly an even bigger one when one moves to level 5. Level 4 allows a vehicle to travel under a certain degree of direction. Fundamentally, this means that its application on the surface will be in the automated control of shuttle vehicles or public transport vehicles travelling along well-defined routes—trams and light rail are the obvious easy applications. It would be perfectly possible for there to be movement on marked highways where there were clear lines or electronic markers of other kinds, but these would be limited and the vehicles would not be able to move outside them.
Level 5—default full autonomy—is another kettle of fish. There you are talking about the vehicle taking full responsibility for everything—in so far as a vehicle can be responsible for anything, and that raises some interesting legal and insurance questions. Here I think we have prodigious challenges of data processing and data management. I was discussing this recently with a friend who works in artificial intelligence and he said that, for full autonomy in one of these vehicles, you have to have the ability to process images and interpret those images in real time, like a human being can do, which means exceedingly quickly. He said that if we look at the computational capabilities that are available today, you would need about as much power to drive a computer that can do that as you would need to propel the vehicle. That is not to say that this is not going to happen, but it emphasises that it will not happen tomorrow.