My speech on driverless cars at the Transportation Research Board, Washington DC, 15/1/19

There is something extraordinary happening out there in relation to driverless cars. But it is not quite what you think – or have been led to believe.

The spending on the development of autonomous vehicles is quite unprecedented. I am not sure in the history of transport whether so much has been spent on a single concept.

We’ve had railway booms, and explosions of road building but nothing on this scale.

The Brookings Institute estimated that between 2014 and 2017, $80 bn had been spent on developing the technology but of course since much is carried out by private companies or simply off the books we have no idea of the total amount. Other estimates therefore suggest $200 bn has been spent so far. Suffice to say, it would probably fund public transit for a year in the top 50 US cities, at a guess. Or build numerous subway systems and a new tunnel into Manhattan.

Clearly, this money is being spent in search of a Holy Grail. But what is it precisely?

The promises are indeed ambitious.

The advent of driverless cars will eliminate road deaths, reduce congestion, free up car parking space, cut pollution and liberate people from the drudgery of sitting behind the wheel. They will provide a new freedom for the blind, the infirm and the young who will be empowered to use cars without being accompanied for the first time. Listen to Parnell Diggs, the director of government affairs for the National Federation of the Blind:

‘We anxiously anticipate the day that all blind people will have the opportunity to drive independently and we believe that autonomous vehicles will make this day possible’

This utopian view of the future of transportation has been swallowed wholesale by the media. I have lost count of the number of articles I have read in the media which start off ‘the driverless car revolution is coming’. Indeed, the public discourse has been so influenced that according to research by Thatcham, a body funded by UK insurers, ’71 per cent of drivers around the world believe they can purchase a self-driving car right now’.

Let us look at some of the hype that has created this

The most outlandish was produced by a  so called think tank called Rethinkx produced a report in 2017 that said:

‘95 percent of U.S. passenger miles traveled will be served by on-demand Autonomous Electric Vehicles owned by companies providing Transport as a Service. As fewer cars travel more miles, the number of passenger vehicles on American roads will drop from 247 million in 2020 to 44 million in 2030.

This is not poor research, or mistaken analysis. It is a lie, simple as that, a breathtaking piece of dishonesty which the author must know is simply not feasible. I need not waste time in showing you why.

But there is more. A consultant called David Galland wrote in Forbes magazine in March 2017

 ‘There will be 10 million self-driving cars on the roads by 2020, with one in four being self driving by 2030’

This was repeated in an analysis by a consultancy called Dreamit which quoted

‘experts estimate that 10 million vehicles will be on the road by 2020’

…and this sort of thing pops up almost daily in my inbox, poorly researched articles than requoted as being the findings of ‘experts’ with the implication that they have carried out meticulous studies to reach their conclusions. If only.

But these experts have been wrong time and again:

In 2012 Google’s Sergey Brin said ‘Driverless cars would be available for Google employees within a year and they will be on the market commercially by 2018’

Well Sergey might be one of the world’s richest men and he has put billions into the development of these vehicles, but boy, did he get that one wrong.

Such promises are still being made: In a similar vein, Uber’s Travis Kalanick said in 2015 that Uber’s entire fleet would be driverless by 2030.

And, folks, you will soon be getting your pizza in an autonomous vehicle. According to a widely covered news report in August 2017:

‘Domino’s has begun testing pizza deliveries using driverless cars. The research is being carried out with Ford, with driverless deliveries tested over the next few weeks in the US city of Ann Arbor, Michigan’.

Why would anyone want that? The truth is the cars have an operator but in any case why would you want your pizza delivered in a car you have to go out on the street to reach in the rain or snow?

Bored, perhaps, with just predicting the future, the media has now taken to covering more obscure aspects of this revolution. NBC reported that a magazine called Annals of Tourism Research, suggested that people would have more sex because of autonomous:

‘With the relative privacy of a car, and no need to pay attention to the road, sex on the go will likely be common with the widespread adoption of self-driving vehicles’.

Let’s see, therefore, where the tech and auto companies have got so far with their efforts to provide us with pizza and sex:

The technology

It is nowhere near ready. Obviously, the most publicised event of 2018 was the death of a woman killed by a driverless car being tested by Uber. That led to a suspension of all the company’s operations.

But that was not the only such setback. Volvo, having promised the introduction of driverless cars by 2018, announced it would be another four years before they were on the streets.

Tesla delayed its highly publicised plan to have an autonomous car drive from coast to coast. Originally scheduled for January 2017, it was supposed to happen last year but now it’s unclear when it will happen.

Even Waymo, which is way ahead technically, has struggled Waymo’s self-driving cars struggled with basic infrastructure comprehension and driving tasks, such as making an unprotected left turn or stopping at traffic signals designed to control the flow of traffic from a ramp onto a road. Its much touted launch of a taxi service in Phoenix has turned out to be a damp squib. It is available only to those already signed up to its pilot programme and there will still be an operator overseeing the car – and an engineer

There are a few shuttle buses – about 100 have been sold by a French company – which drive on pre-organised routes – and according to passengers, go terribly slowly. One passenger who used a bus in Syon in Switzerland told me a typical story that there was a lorry discharging a load blocking the street but the car could not overtake it and so it sat there for 15 minutes. Similarly, in Las Vegas earlier this year, a bus was struck by a reversing lorry because it had no way of taking evasive action or sounding its horn.

And of course there is the very wide variety of tests being carried out in the US, mostly with an operator aboard. But nothing that can be considered remotely as a driverless car of Level 4, defined as where there is no need for anyone to pay attention to the road.

The big problem is, in fact, Level 3 where the car drives itself much of the time but requires the driver to stay alert and take over if something goes wrong. Ford began testing Level 3 vehicles and found that the drivers dozed off as they had nothing to do. They tried putting in two test drivers but they, too, dozed off.

Michael DeKort, an aerospace engineer turned whistleblower wrote recently:

Handover cannot be made safe no matter what monitoring and notification system is used. That is because enough time cannot be provided to regain proper situational awareness in critical scenarios.’

So could this be solved by Artificial Intelligence? DeKort suggests not because of the sheer amount of learning that would have to take place: ‘thousands of accident scenarios will have to be run thousands of times over to train the AI on those scenarios’ – and until it learns, there will be lots of deaths.

AI is often touted as the solution to all these problems, but in fact many people in the industry are sceptical about it. AI is really machine learning, and it is a process that is non-linear. The computer analysts do not know quite how an AI solution is reached. There is no clear mathematical formula. Therefore each car could be different. And if one goes rogue or fails to learn, say, how it should react to a particular hazard and say, drives off a cliff, then how can anyone be sure that lots of other cars of the same type will not do the same.

This brings us nearly into the minefield of the legal, liability and regulatory issues.

Personally, I think the trolley question – who do you kill, the driver by going over a cliff or the kids standing in the road – is not really the crucial question. The more interesting issue is who is liable in an accident. Would it be the software developer (though AI adds another layer of complexity), the manufacturer, the owner, or the person in the vehicle?

Moreover, there is the huge issue of hacking. The large number of computers and the code they use on board are sure to be a key vulnerability. These cars will have a dozen computers or processing units, each vulnerable. What if you get in your car one morning and it says ‘send me 500 bucks or the brakes might fail’.

The way the driverless cars are being developed adds to the problem. Rather than cooperating to produce the technology, there is a massive battle going on to be the first to reach the market with a product. This means that there is enormous secrecy about the software and therefore if something goes wrong, it will be very difficult for the regulators to uncover what happened. 

The public image took a massive blow in 2018

 There has not only been the death of Elaine Herzberg in Arizona but several incidents involving Tesla cars in autopilot mode such as one smashing into a police car and another driving into a highway barrier killing the driver

This has increased public concern. There have been numerous surveys on the acceptability of driverless cars and they all point to a great reluctance to use them. For example, in a recent UK survey, only 29 per cent would be happy to be picked up by a driverless car, and when it comes to trusting autonomous driving technology with the safety of a person’s child, that number dives to 17 per cent.

It is though, the promise to bring about a transport revolution that the driverless car industry is finding most difficult to fulfil.

In order to achieve its enormous reduction in road deaths combined with less congestion and other environmental benefits, the world will have to be dominated by driverless, electric, share use cars.

Electric is possibly desirable, provided the power is obtained sustainably but there are still issues around the availability of the raw materials. Persuading people to buy electric cars has proved an uphill task as people have worried about the range of the vehicles. In the UK, the numbers are still miniscule.  There’s fewer than 200,000 electric or hybrid cars in a fleet of 30 million. If even 10 per cent of the fleet were purely electric there would be huge problems in providing sufficient batteries. And these lithium batteries are highly toxic and not necessarily environmentally sustainable. Moreover creating sufficient charging points is proving expensive.

Leaving that aside, even if the technical problems mentioned already could be overcome, the notion of shared use seems fanciful.

There is no evidence that driverless cars will mean that personal ownership would come to an end. People want to have full time access to their cars even if they do not use them for 10 per cent of the time.

The idea that autonomy would result in shared use has no basis. What is the logic? At the moment, people have access to taxis, Uber, Zipcar and Hertz. In what way does autonomy change the equation?

The driverless car supporters say that it would be much cheaper. But the only gain is to dispense with the driver, who is probably on minimum wages. So why does the advent of autonomy mean that millions people would suddenly be happy to no longer own their own car. There is simply no logic to this proposition.

Of course some Millenials who live in city centres with good transit do not own cars. But what if you live in a suburb or a rural area? And have these guys ever tried to fit a child seat into a taxi? How long would that take every morning on the way to nursery? And what about those golf clubs that are safely in the trunk rather than in the garage?

This, in fact, the weakest point of the whole argument. Shared use is ot a corollary of autonomy. The idea that we will all be happy to share vehicles, abandoning our cherished Jaguars or Maseratis to be driven in featureless shared pods in which a drunken teenager might have vomited the night before, is plain crazy

So why are they suggesting it is. This is where the whole model falls apart. The supporters of this technology only put forward the idea of shared use because otherwise they know the idea would be unpopular. After all, if you were to own your autonomous vehicle, you would then use it like a conventional car – except some of the time it would be driving around empty, such as when you went to the airport or a station – you would send it home for your partner to use.

Therefore it would create more pollution, more congestion not less. Therefore the supporters of the technology had to find a way out to argue that their invention would be environmentally beneficial. But the proposition simply does not hold water.

The truth is car ownership could be greatly reduced now if people today were ready to risk always having an Uber or a Hertz self drive at hand but by and large they don’t or they use them in conjunction with retaining their car.

Knowing that driverless cars are unlikely to be able to work without problem in urban areas, there is a more dangerous prospect, and that is the idea that city authorities will have to adapt their environment to accommodate them.

There has been talk, for example, of making cyclists wear devices that warn of their presence, as driverless cars have had difficulty coping with their slightly anarchic behaviour

For driverless cars to achieve their aim of ending road accidents, two conditions have to be met:

  • first the cars have to be truly driverless, not merely fitted with driver aids
  • and secondly, that all cars have to be driverless and the roads would have to be cleared of everything else: pedestrians, cyclists, motorcyclists, dogs, horses

The implication is that the urban environment will have to be changed. Caroline Hutt, who calls herself the Mobility Programme Manager for Addison Lee, a UK taxi firm, wrote in the Evening Standard last year:

‘London needs to embrace the arrival of driverless car. Street design needs consideration — autonomous vehicles need a de-cluttered environment, with clear delineation between pedestrian and vehicle space. ‘

As she admits, that’s different from the London Mayor’s present “Healthy Streets” agenda which emphasises shared space and pedestrian priority.

So she wants to see ‘the Government convene a City Mobility Task Force to make strategic decisions about adopting autonomous vehicles. Changing habits will take time’

So  one could sum this up that in order to adopt a technology that we do not want, we have to ditch current policies around making city centres into healthy streets and ‘declutter the environment and put fences up between roads and pavements’

Her article highlights how today’s driverless car enthusiasts have got into trouble over their claims. There is a certain lack of logic about suggesting that enabling more people to use cars, such as the disabled, the blind, the infirm, and the demented and the young, will lead to less congestion.

The effect on public transit is obvious. If car travel is made easier for many people and the roads are adapted to smooth their path, what will be the role for buses and trams?

Perhaps some of these problems can be overcome, but it is maybe the simplest problems that are the most insuperable: driverless cars will have to be programme not to kill people. If someone stands in front of them, they will have to stop. Their range can be altered – say 5ft or 10 but not the fact that they cannot be allowed to hit someone knowingly.

This effectively opens up huge issues. ‘Bad people’ as Donald Trump calls them will have free rein to stop them. Driverless cars, therefore, can never be secure. You won’t be riding in one in the back streets of Baltimore, that’s for sure.

Anyone worried about security will never ride one. Prime Ministers will never ride one. And nor will the Queen. There is, too, a long list of other exclusions. Fire engines, the ambulance, the police, motor cyclists, vehicles which need priority – and so on.

Then there is a simpler issue. Imagine two driverless cars meeting each other on a single lane highway in the wilds of Somerset which I know well. If they are Level 5, as Waymo seeks, the passengers will not be able to intervene. So how will they sort themselves out?

A word, here, too about disabled and blind passengers

Waymo has explicitly sold this idea But just imagine, if these people are stuck in a broken down vehicle? Or if the vehicle takes them down a wrong turn? Or comes across an unexpected road works, an accident, or a landslide.

Then there is employment. Driving is one of the world’s biggest sources of employment. Why are we trying to do those people out of gainful employment. I have not time to go into the issue of loss of employment here and it would, in any case, only come about if the vision, as represented by the tech and auto companies, comes true, which clearly I don’t believe will happen.

I have been struck by the contradictions in the Uber model. Currently Uber, the company that has never made a profit, gets if capital investment for free – from the drivers who have to pay for their cars. Yet, somehow it is seeking to dispense with drivers – how would they then be able to afford to provide the service. I asked an Uber executive this and he said that the OEMs – the car manufacturers – would lease the cars to them. But I can’t see why they would put up all that capital cost. It seems an unsustainable concept.

One might ask why, given these difficulties, the protagonists of this future – or dystopia as I would term it – are so desperate to sell it to us, using safety as their weapon. It is an unholy alliance between the desperate auto manufacturers and the tech companies who have exploited their monopoly positions to make super profits.

The car industry is having to face up to the reality that the damage caused by their products is being recognised as a key factor in all kinds of social negatives: pollution, environmental degradation, obesity, even inequality, road deaths and so on. They are terrified of becoming the cigarette companies of the future. That is why they are spending billions on R & D and each one is desperate to be the first to produce a viable product. As for the tech companies they have no idea what to do with their money – and see driverless cars as being an opportunity for more people to use their smartphones for longer.

Of course we should make use of technology – and some of it will make things safer. Fewer people now reverse into garage doors, and we are all the better for it.

There is a real danger that we are being distracted by this technology, that there is a  feeling being created that there is no point doing anything about the current situation because it will all be allright when we are all driverless.

That is to some extent happening already with widespread complacency about the increase in road deaths following years of sharp decline.  Yet, there are countless ways in which we can make our roads safer now. Indeed, we could even mandate that private cars are not allowed into city centres, and only taxis and minibuses would transport us. These are political decisions, and so would forcing us all to share driverless pods be a political decision.

I visited an autonomous car exhibition in Stuttgart and it was very interesting speaking to the exhibitors. I had expected I would be a bit like an atheist being introduced to the Pope but not at all. They were all as sceptical of the claims being made by the tech companies as I was. They were happy making money out of producing sensors, running test tracks or developing collision avoidance systems but they did not think that there would be any AVs driving freely on the roads in the near, even medium term, future.

So the extraordinary thing I mentioned at the beginning is not that this utopia – or dystopia as many see it – is happening tomorrow. It is, rather, that so many people have swallowed the hype and think that they will soon be in driverless cars.

I am not a conspiracy theorist, but in this instance, it is very clear what has happened. The hype is the key part of the business model. The more hype, the more investment money becomes available. Then maybe one or two companies develop a useful piece of software or a new bit of hardware and they will be sold for millions of dollars. All the other investors will lose their money but who cares? There will be more hype, more investment, more failure. But at some point it will come to a juddering halt.

And the cracks are beginning to appear. Look at what John Krafcik said at conference in November: ‘It’ll be decades before autonomous cars are widespread on the roads — and even then, they won’t be able to drive themselves in certain conditions.’

There is evidence that the supporters of this technology are becoming defensive. Last week a group of tech and auto companies announced that they were creating an organisation called Pave in order to get the public to better understand the advantages of the technology. Audi’s president of North American operations Mark Del Rosso said the group’s aim was to “educate policymakers” about how the “technical challenges of creating driverless vehicles are solvable”, and bring real advances in road safety.

The companies have realised that the public is getting doubts about the technology. About time. I have watched countless politicians be suckered in by the message that driverless cars will soon be on the roads. This is a dangerous fallacy and poses a big risk for supporters of transit. If local mayors and officials begin to believe the hype, they may well decide that investment in transit is a waste of money.

Transport researchers therefore have a big role to play in ensuring this does not happen. You need to warn the authorities that driverless cars are not the panacea that they are presented as. You need to be very wary of any request to start adapting the city streets to allow for driverless vehicles. You need to show that it is not technology that will solve congestion and pollution problems, but the recognition that cities are for people, not cars. And above all you need to be sceptical and wary of the promises that these modern snake oil salesmen are making.

I will leave you with the words of Michael DeKort: ‘The 20–30-year time period [for the introduction of driverless cars] isn’t remotely close. The real answer is that they will never get remotely close to finishing. Not much farther than the first base they are on now.

And just a final point. At the airport, they were checking for drugs – with some amazing new fangled machine, with a new type of sensor? Nope it was with a dog. There’s some things that we animals can do better than machines!

 

Shares