On the face of it, there’s only good news here. Autonomous cars are very much the coming thing, and they represent a potential future for motoring that’s cleaner, greener and, most importantly, safer. With the machines taking control of motion, you and I will be able to sit back, read a book, tweet, chat, watch a film, or even get some work done while the organised electrons take the driving strain.
The MobilityX conference, which opens in Dublin on Thursday before switching – at the invitation of Mayo County Council – to an intensive TED Talks-style session on Achill Island on Friday, has been established to try and entice autonomous technology companies to set up shop here in Ireland. With our legacy of high-tech industry and investment, and a huge global market predicted for autonomous vehicles, Ireland could be poised to reap great benefits from both the safety and fiduciary sides of autonomy.
That’s certainly the hope of Philip McNamara, vice president of Voxpro and organiser of MobilityX.
"I think with [just] 160 road deaths in Ireland last year, we've done well, over the past few years, but there's only so much that airbags and safety belts can do. So from all the research, Google is saying that its autonomous cars are 10 times safer than the best drivers, and 40 times safer than teenagers, so bringing in legislation that would allow autonomous cars to start testing here would be a very positive thing, a very positive signal.
"You have companies such as Google's Waymo, companies who are testing in Arizona and so on, because there is legislation there allowing them to do so. If there were legislation here, allowing the same, perhaps out in the west of Ireland, somewhere like Mayo where we're holding our weekend event, that would be really positive, and would encourage more companies to set up here.
“I think there certainly are wet places in the US and the rest of the world, but I think that there are a lot of testing conditions here – wind, rain, whatnot – plus we’re close to the major German headquarters, which is a big advantage. I was talking to the guys at Google and they said that they need to test their cars in the US first, because it’s simpler to do it, they have dry weather and straight roads, but eventually they do need to be able to test in places like Ireland, and there are certainly benefits to testing here, even down to silly things such as what happens when a sheep wanders out into the road. All of these crazy things do happen, and Ireland would be a great space for it. You can’t just test in the nice, warm places, and you need somewhere where there’s public support for the testing. I think we have always been open towards new tech and open towards innovative companies.”
Already happening
To an extent, it's already happening. Jaguar Land Rover is in the process of setting up a new research and development base, dedicated to future autonomous and electric vehicle technology, in Shannon, while Valeo, the giant French car component and technology company, already has a base in Tuam, and is working with the Lero team at NUI Galway on next-generation autonomous vehicle sensors.
One of the big questions is just how far away autonomous vehicles are from being an everyday reality, with estimates ranging from five minutes to never, depending upon to whom it is you're speaking. Bruno Fernandez is co-founder and chief technical officer of Nexar, a company which makes cloud-connected dash cameras, part of the vast sensor suite that autonomous cars will need. He will be a key speaker at the MobilityX conference, and tells The Irish Times that while partial autonomy is already here, the sci-fi, fully robotic car is still some time away.
“From simple warnings, to lane-keeping, adaptive cruise control, highway autopilot, the tech is here, and every year new features are incrementally becoming available,” says Fernandez. “In contrast, full automation, meaning an autonomous vehicle [AV] with no steering-wheel, where the human does not need to ever take control, and that can perform any driving task, in any environment, is still decades away from becoming a reality and mass-market.
“Creating a fully autonomous, reliable driving agent is still complex, expensive, and years away. There are several reasons: lack of those variations and edge cases, limitations in the sensors, lack of cost-effective computational resources that one can actually put in a production car at the scale of millions of vehicle units, and even the actual algorithms and math behind autonomous driving, AVs are still a huge undertaking.”
In every new technology there's going to be situations where the tech fails
One has to wonder, though, whether autonomy is becoming something of a technological millstone around the car industry's neck, and one of its own making. The vast rush towards autonomy has, generally, been viewed (and publicised) as the biggest safety measure, ever. Machines make fewer mistakes than humans, ergo a machine that drives a car will be safer than a human driver. But that has been called into doubt in recent weeks, after the death of a pedestrian in Arizona, following a collision involving a prototype autonomous car being tested by ride-hailing service Uber, and the death of a driver in a Tesla Model X, who perished in a fiery accident while the car was allegedly under partial computer control.
Unfortunate necessity
McNamara sees these incidents as an unfortunate necessity, though. “In every new technology there’s going to be situations where the tech fails. If you go back to the early airliners, there were terrible tragedies because the technology was evolving,” he says. “And with any new tech, there’s always a fear factor, at first. There are going to be pioneers, such as Uber, where there will be tragedies. Unfortunately, that did happen in Arizona, but people will learn from it. It’s terrible that it took an accident to make it happen, but the sensors will get better, the cars will get safer. People will be highly motivated to make sure it doesn’t happen again, but it’s inevitable that this sort of thing will happen, as the technology gets better over time.”
Others are far, far less convinced of the efficacy of autonomous tech. Andrew Clews is a motoring writer and podcaster, and something of a sceptic of robotic cars, and he says that there's a fundamental error in the software that most are using for autonomous car control.
“Autonomous vehicles are the solution to a problem no one knew we had. Yet they have ignited the imagination of the public thanks to a mix of – and I’ll be polite here – over-optimistic marketing and complete lack of understanding of what the implications are,” he says.
Imagine if your toaster killed 160 people in Ireland every year – it would be legislated out of existence
“AVs are being developed with the use of artificial intelligence and machine learning, to much fanfare. But we shouldn’t be excited, only fearful. Machine Learning, or ML, does not follow safety-critical software creation standards. ML does not allow for all the code involved to be inspected, check how the system is wired together and demonstrate it will behave correctly for all possible situations. Speaking to experts in the field, when some claim ML is being developed for safety systems, they are dismissive that the equivalent of an electronic zap when it does anything naughty shows it will do the right thing at the right time. We are being asked to believe, yes believe, that it will work, for there isn’t and cannot be any proof via this method, which is totally at odds with safety-critical software standards. To create software following the safety-critical standards will be incredibly slow, expensive and complicated.”
Sinister aspect
There is another, potentially more sinister aspect to autonomous vehicles, and that is the rationale behind their development in the first place. AVs were first developed by the US military, as a way of delivering supplies to the battlefield without the need for valuable manpower to be tied up in a logistics effort. It didn't take long for the Silicon Valley experts who did the early work on the tech to realise that there was the potential for a mountain of money to be made. The Boston Consulting Group reckons that the global market for autonomous tech will top $42 billion by 2025, and some estimates have put the market as high as $5 trillion by 2050. That's a huge amount of money, and firms from traditional car makers to Google, Apple and Uber are hungry to get their share. A considerable slice of that revenue could well come from piping advertising through to those people, sitting in traffic in their autonomous cars, who now don't have any driving on which to concentrate. That should trigger, at the very least, concerns about being constantly bombarded with unwanted ads and, at the worst, worries about the potential scale of another Facebook privacy scandal.
Again, Philips waves such concerns away. "I think that if you look at it, you have 35,000 people dying every year in the US in road accidents. Everyone has this household appliance that is your vehicle, and there's a huge chance you could die in it. Imagine if your toaster killed 160 people in Ireland every year – it would be legislated out of existence," he says. "A car is like a home appliance, and we get in it every day and use it to go to work. And yet it's extremely dangerous. So I don't think it's about trying to sell us advertising, although that will come, it will be created, but I think the safety is the biggest issue. Plus it's incredibly unproductive to sit in your car in bumper-to-bumper traffic, for an hour and a half, or two hours, every day. If you could put that resource towards working, or even watching a movie, or being on the phone safely, that's hugely more beneficial towards society. Plus there's the enormous cost of each road death or injury. So no, I don't think it's an issue that they're going to sell us more ads. I think road deaths are a massive societal cost that somehow we've come to accept. We'll look back in 10 or 20 years time, and say, 'wow, I can't believe we let 18-year-olds drive these things at 100km/h on public roads.'"
It remains to be seen if safety, rather than money-making, will retain its primacy amongst autonomous vehicle developers, but one thing is more or less certain: robot cars will be a driver of unemployment. Frost & Sullivan, one of the world’s major business consultancies and analytic companies, is suggesting that the rise of the robotic car could cut the cost of motoring by a factor of 12 or more, but at a huge cost of lost jobs.
Volatile
According to the research, the per-mile cost of motoring is enormously volatile, but an average cost of say, hiring a taxi works out at around $12 per mile, or €10 (€6.68 per kilometre if you want to be more precise). Frost & Sullivan says that this could be cut to as little as $1 per mile (€0.89 per mile, or €0.55 per km). That’s more or less equivalent to the cost of most public transport systems in major cities, and therefore brings motoring within the reach of a far greater number of people.
Sadly, though, it’s bad news from an employment perspective. While some of the saving comes from the whole concept of shared ownership, and the amortisation of the cost of the vehicle across several users rather than one, and more savings come from artificially intelligent navigation systems which can avoid traffic snarl-ups, most of the cost reduction is because robots don’t need wages or downtime.
“The biggest potential for decreasing the price per mile is provided by autonomous systems replacing drivers, which will make it possible to save the 30 to 40 per cent of the cost we pay for the ride,” Frost & Sullivan’s Arunprasad Nandakumar says.
If we, as a society, were really interested in reducing accidents and fatalities, we'd plough money into driver training
That's to say nothing of the potential impact on the haulage industry. Cargo transport currently represents around 4.5 per cent of the total employment figures of the European Union, and of that more than one quarter are employed in road transport and haulage. That's millions of jobs on the line, and a report published by the International Transport Forum (ITF) in Leipzig has suggested that 50-70 per cent of those jobs could be gone by 2030. José Viegas, secretary-general of the ITF, said that "driverless trucks could be a regular presence on many roads within the next 10 years. Self-driving trucks already operate in controlled environments like ports or mines. Trials on public roads are under way in many regions including the United States and the European Union. Manufacturers are investing heavily into automation, and many governments are actively reviewing their regulations. Preparing now for potential negative social impact of job losses will mitigate the risks in case a rapid transition occurs."
Dire warnings
Given the recent dire warnings issued by the World Bank, which suggested that employment rights and benefits would have to be slashed and burned to allow humans to compete with machines for work, perhaps autonomy is not a technology that, morally speaking, we should be pursuing at all.
Andrew Clews reckons that making money is really the only motivation behind autonomy, and that we should be far more careful about how we approach it.
“If we, as a society, were really interested in reducing accidents and fatalities, we’d plough money into driver training. We’d make it much more difficult to get a licence,” he says. “We’d have regular compulsory retraining. That is all far cheaper and provides a quicker reduction in road accidents, injuries and deaths. This is a very complicated issue, with many facets that need grown-up discussion to take place. No more simplistic ‘everything will be wonderful’ without explaining what it will actually take to get there.
“As politicians in the US have recently found, if you do not understand the issues and implications, you could end up looking foolish for statements and actions you’ve taken to allow AVs on the roads in the rush to chase the cash they promise.”
