The hype of driverless cars has been strong in recent years. However, a recent article by transport expert Christian Wolmar suggests that things might not be progressing as well as the earlier hype suggested. Wolmar visited an autonomous vehicle exhibition in Germany and some of his observations are really interesting. This starts with a reference to the typical hype that surrounds driverless cars:
There is something religious about the fervour with which adherents to the driverless credo practise their faith and promise us a new kingdom. Their proselytising has indeed convinced many. Politicians are making outlandish statements, such as Jesse Norman’s two weeks ago, that ‘Our entire use of roads is to be revolutionised by autonomous vehicles’, and pouring large sums — a promised £180 million so far — into bizarre research projects such as the development of strange robot cars slower than a Reliant Robin and allowed only on pavements in Milton Keynes.
The public, too, has been won over. Ask the average Joe in your local or even Basil in his club, and they will sing the praises of a technology that will apparently end carnage on the roads and allow them to check their Twitter accounts while being driven to work.
However, it seems that when you dig behind the scenes a bit, as Wolmar does, things aren’t going as well as the hyped up media releases would have you believe.
Surprisingly, I met more doomsayers than purveyors of the autonomous driving dream. The starkest warning came from Tim Mackey, who styles himself ‘senior technical evangelist’ for Black Duck Software, a company that specialises in security issues around autonomous vehicles. He believes there will be a seminal event that will stop all the players in the industry in their tracks. ‘We have had it in other areas of computing, such as the big data hacks and security lapses,’ he said, ‘and it will happen in relation to autonomous cars. At the moment, none of the big players are thinking properly about security aspects and then they will be forced to.’ He pointed to a video showing on another stand in which a man was calling up a car from a garage using a phone app: ‘That sort of thing is just too easy to hack. There’s all sorts of software put into cars that is old and easy to access. We just have to hope that the wake-up call will be minor and not kill anyone.’ Indeed, in a test a few years ago, hackers were able to get hold of a car’s steering and braking systems and Mackey is convinced that criminals will one day use the same method.
It’s not just security risks that provide some of the upcoming challenges to speedy advancements of driverless technology. Recent high-profile accidents have made people worried, but at the same time the vehicle manufacturers (called OEMs now, for some reason) are stuck in a situation where shareholders are demanding huge investment in these new technologies, but the pay-off is a long way away:
More widely, there was a general expectation these suppliers were riding the crest of a wave that will hit the rocks soon. While there is no doubting the scale of this industry, with billions being invested every year, none of the OEMs has yet made a penny from selling a driverless car. This money, benefiting these exhibitors, is therefore a punt, a high-stakes bet there is a pot of gold at the end of the rainbow. One, Johannes, told me: ‘I see a pattern like the dotcom boom. At some point, people are going to realise that the day when they start to get returns for their investment is far off, if ever. Then they will start pulling out and who knows how bad it will get. But the clever money will move somewhere else.’ The bad publicity caused by a couple of deaths in Tesla cars while its autopilot was engaged and by the Uber fatality may be seen as the start of public disenchantment with the concept.
Overall, the impression was that things are progressing slower than the hype would make you believe.
Of the 20 or so exhibitors I spoke to, not a single one believed autonomous cars would be on our roads within a decade. There are a myriad problems, ranging from insurance issues to the limitations of the technology and the resistance of the public to travelling in them. Rather than swallowing the fatuous statements from politicians about how driverless cars are going to change our lives, we need a sober assessment of their potential benefits, if any, and, crucially, of their downsides.
Another example of how much progress is required before we should be willing to let these things loose on our streets is captured by this recent tweet, which discusses the limitations of Toyota’s pedestrian detection systems:
Some situations where Toyota's automatic braking technology might not avoid crashing into a pedestrian. https://t.co/5mg03KHXi3 pic.twitter.com/jYWmZmIBd9
— Timothy B. Lee (@binarybits) June 7, 2018
So, basically most pedestrians will be squashed by these things.
Make no mistake, there are some important potential benefits from driverless technology. Being able to squeeze much more out of existing roads means we won’t need to widen roads or build more of them. Plus safety systems should get better over time and substantially improve the safety of the transport system. But, as I discussed recently, there are huge unresolved issues with this technology – not just technical barriers that need to be overcome, but also proactive regulation to ensure driverless cars actually make our streets safer and better. Maybe it is a good thing that the technology is progressing slower than expected.
Processing...
Might also want to have a look at the NTSB report on the Uber murder of a pedestrian: https://www.ntsb.gov/news/press-releases/Pages/NR20180524.aspx.
In particular the lines that say the emergency braking system was disabled because of too many false positives
I think it was Bill Gates who said something along the lines that people over-estimate how much change will occur in 5 years but under-estimate how much change will occur in 10 years. Driverless cars will happen at some point and it will be better than what we are doing now – remember prefect is the enemy of the good. My guess is that in 5 – 10 years the change will occur. Many of the technologies that are needed for driverless cars such as 5g will be in place. It won’t happen overnight but it will happen.
As an engineer trained in embedded system design and having 20-odd years experience in IT, I disagree strongly.
The reason: It’s not about the physical technology (5G, embedded computing, etc). It’s about the software, things like security and how to separate signal from noise (is that a pedestrian, or a cutout?). Then you have issues like convincing the public that you should trust the car when it definitely _will_ kill people in some circumstances – Ya can’nae break the laws of physics, Jim.
Regarding the killing of people due to physics, the UBER car would still have killed or seriously injured that pedestrian even with breaking enabled. If their algorithm was set in such a manner that it considered the risk of that particular pedestrian running out, the car would be considerably slower than a horse and cart. That in itself isn’t a problem, however who would buy such a vehicle? People are too selfish, or impatient, or under external time pressure – They’d buy a non autonomous vehicle, or drive in hands-on mode only. Or (shock, horror) take public transport 😉
As for security: https://edition.cnn.com/2015/05/17/us/fbi-hacker-flight-computer-systems/
In a recent interview, Chris advised that the situation hasn’t yet improved substantially. Cars are even less secure than airplanes and it’s only a matter of time before they become attractive targets for miscreants.
The security risks alone should be enough to but general use autonomous vehicles in the “too hard” basket. That said, specific use AVs have a place… Just that place is in controlled environments. Heck, even current vehicles give me concerns around their hackability.
You raise some good practical issues but these are not impossible to figure out – software is developing in this area very fast with the amount of data being collected. Google has already tested completely autonomous cars on public roads without incident. However, each time a driverless car crashes it is world headlines, but that is only because it novel.
However, I think the statement of the perfect being the enemy of the good is apt. All of us trust cars now and hundreds in NZ die each year yet we tolerate that risk. I think driverless cars will crash but I think if can reduce our road-toll then it is worth it.
Security?People already use vehicles to commit terrorist acts now so there is the potential to actually make cars more secure i.e. impossible to deliberately crash or go over the speed limit.
But I think your concerns are mostly practical and, therefore, these can be figured out. The day will come when there will be only dirverless cars on the road – and that time will come a lot sooner than later.
“All of us trust cars now…” this is true to a large extent although while I’ll happily drive around town (when I’m not on my bike or PT) I do think twice about long journeys on our pool quality state highways and rural roads. Generally our small family fly for anything that would be longer than a couple hours drive and that’s easy to justify when considering time, costs and risk. Bring on regional rapid rail so more longer journeys can be time competitive while avoiding our roads.
People risk being killed in human operated vehicles because they feel as if they’re in control. When they’re placing their lives in the hands of a machine, the bar of trust is set much much higher.
All problems can’t be resolved – If “resolved” also includes isolating the cause of the problem and working around it.
IOW: Resolving the killing machine issue means segregating the AV from situations in which it would be possible for it to kill. This is why for non-AVs we have rules around what can travel on a motorway, this is why we have median barriers and pedestrian fences.
Aversion to premature death makes either the car or it’s operating environment uninviting at best. Trucks and special vehicles will adopt autonomy before widespread public use… Well, after the initial hump.
The hump: AVs are nice luxury toys and, as such, the playthings of the affluent. The early adopters will buy many, then the accidents happen and/or people develop tall poppy syndrome towards those adopters (think of the Prius backlash in it’s early days) and their market penetration will decline. Will it rebound? Perhaps, however (and I admit I may be lacking creativity here) I can’t see how without creating an environment that is low on livability.
I wish I had your faith in technology to resolve all issues, however having lived and breathed tech since the mid 80’s (long before it became a career) I can’t share your optimism. Perhaps it’s the curse of the engineer – To become more and more cynical as the days wear on. 🙂
All problems _can_ be resolved…. Grrr.
As someone who has assumed airplanes are pretty secure but has no technical expertise, can you explain to me this comment ‘Cars are even less secure than airplanes’? It seems to also imply planes are not that secure!
Well, there is still a large aircraft missing from Malaysian Airlines, with hundreds of people on, and no one has seen it for years – no one knows whether it is pilot error, pilot on purpose, machinery malfunction, etc. I’d say that’s a fairly large insecurity right there…
Even if this aircraft has been electronically hijacked it is one aircraft. There are around 100,000 commercial flights a day, if one aircraft has been hijacked out of the millions of flights in the computerised aircraft era then I would say they must be pretty secure.
From the article: “Roberts said he used a modified Ethernet cable to connect his laptop to an electronic box underneath his seat that controls the entertainment system. From there, he hacked into the airplane’s computer nerve center, the document cites Roberts as telling the FBI.”
More recently, around cars: https://www.theregister.co.uk/2018/07/04/plane_hacker_roberts_interview/
In short, there hasn’t been any black hat action yet (that we know of) because it’s technically risky or tricky. That said, security through obscurity and basic firewalls only is no longer a satisfactory way to do things.
The second article explains in short why cars are insecure. A better link: https://www.extremetech.com/extreme/212251-todays-car-hacks-bmw-chrysler-mercedes-benz-on-ios – The ubiquitous Samy Kamkar showed how his OwnStar device was adapted to get into the car via their remote iOS apps. He targeted BMW Remote, Mercedes-Benz mbrace, and Chrysler Uconnect services, all on Apple iOS.
The flaw, according to Kamkar, is the automaker’s almost childlike faith that the certificate on the remote server is valid, disregarding what kind of network provides the connection and whether it’s a legitimate remote server or one that just says it is
OwnStar is a small Raspberry Pi PC with wireless connections in a portable carry case. The hacker puts it near the vulnerable car. When the owner issues a remote unlock or other command from an iPhone or other iOS device such as remote start, OwnStar gloms onto the exchange and grabs the logon credentials. OwnStar then mimics the owner’s remote device to access the car and has access to all the remote functions.
As an aircraft avionic engineer I can assure you that is utter bollocks. In flight entertainment is a completely independent system to the flight control/navigation avionics found in commercial aircraft. There is no way whatsoever he did what he perported to do.
The technology behind electric cars has been around for ages. The first commercial lithium ion battery was in 1991. And yet they still make up a tiny fraction of the vehicle fleet.
That is because their raw price new, i.e. not subsidised by the taxpayer, is for most people unaffordable and commercially uneconomic. And unless there are substantial changes not much is going to change.
Fiat Chrysler are now seriously looking at manufacture of EV’s but aimed at the very high end of the market.
In fairness the new price of a petrol vehicle is for most people uneconomic, which is why they buy second hand. The most common EV in NZ at the moment is the Nissan Leaf, a second hand import.
That’s my point exactly. Even if driverless cars were possible in 10 years, how long will it be before they are affordable? I know it is a bit different because we can probably already manufacture all of the required components in bulk, but what car company isn’t going to try and charge a premium for them while they can.
That’s kind of why I think the most promising business model for AVs is car sharing, not personal ownership. Uber and Lyft are the most well known companies trying to go this route. Why would I buy a car and have to pay for insurance, maintenance, a parking space… when a driverless taxi will be so cheap?
“… they will sing the praises of a technology that will apparently end carnage on the roads and allow them to check their Twitter accounts while being driven to work…”
AKA “A Bus”.
AKA “A Bus”.
You think that people don’t get killed on buses?
fatality rate 0.11 per billion miles of travel. It’s not never, but it’s pretty close, compared to 7.3 driving a car.
http://election.princeton.edu/wp-content/uploads/2015/05/transportation-fatalities.jpg
The last generation of Toyota cars didn’t even stop automatically for pedestrians. So progress is being made. Getting to the point of trusting the car to drive completely on its own is significantly different though. What is happening already is driver assistance technology that makes driving substantially easier is available in almost all modern cars. This will reduce accidents and allow people to travel more easily in their car (which I know is a bad thing on this blog).
So over time, cars will do more driving without needing much driver input. Eventually that will morph into the car driving completely by itself. But because of the number of variables involved, that will take longer
I only advantage I can see is you don’t have to park them somewhere. I could drive myself to a meeting and then get the car to drive around the block for an hour until i need it, then drive myself home. It isn’t going to help traffic congestion but it will get us past all the maximum parking rules.
And there’s a huge demographic who would use them exactly like this.
Let’s face it, driverless cars are a solution looking for a problem.
Congestion and deaths aren’t problems?
But driverless cars aren’t necessarily going to solve either congestion or deaths. There is a good chance they will make congestion much worse – and the list of pedestrian fatalities shown above, indicates that pedestrian death rate is still going to be high too. Basically, every single thing on that list, is what pedestrians do.
In that case it is a bad solution to a real problem rather than a solution looking for a problem.
@ average human – https://www.engineersaustralia.org.au/News/driverless-cars-could-lead-increased-congestion – While basic levels of automation could have a small positive impact in helping to ease traffic congestion, higher levels of automation could have a detrimental effect on congestion when the user adoption rate is low.
Jan 6 this year from BBC: The Department for Transport predicted a “decline in network performance” once one in four cars become driverless.
However, should driverless vehicles make up between 50% and 75% of cars, DfT researchers say they will reduce congestion.
Driverless cars are a solution, looking for a carpark.
I’m also of the view autonomous technology will not be ready for a while. On the other hand, once the tech is ready then the transition will be very short as the benefits and efficiencies are compelling. So short-term pessimist; medium-term optimist i guess?
The way I see it is this: these cars will NEVER satisfactorily identify risks, in particular jaywalking pedestrians. The companies are BS’ing you with their software claims in order to make their product acceptable to the public. So what’s coming is severe restrictions on pedestrians, cyclists and horse owners.
As a pedestrian, you will not be permitted to cross a street anywhere except at light controlled crossings, and heavy on-the-spot fines will apply. In heavily pedestrianized streets this will probably be re-enforced by fencing off the footpath from the street. Pedestrians, you will be facing longer walks. You may have to walk several hundred metres and then double back the same amount just to cross the street.
Cyclists will be restricted to designated lanes. If no lane exists, or if you come to an intersection, you will be forced by law to dismount and walk your bike to a bike lane.
Horse riders: you will not be permitted to walk your horse anywhere near a road, and certainly not along a grass verge. To go anywhere, by law you will have to truck your horse to an off road site.
Motorcyclists will not be affected as they will presumably carry the same “awareness and identification” software that the cars will carry.
These cars are about to take over the streets again, and the restrictions I have mentioned will give the illusion of greater safety, but it will be at the cost of convenience.
This is the primary problem. You can see it happening in the States already as many states have passed laws that restrict cities from regulating driverless cars, which will likely translate to them being inclined to install barriers to keep the pedestrians from getting slaughtered, and corral them to their friendly driverless Uber kiosk that will conveniently take them across the street.
I see your points, but I don’t think NZ would ever get like that (certainly hope not anyway).
Can you imagine the costs of fencing off every road?
Not every road. But we actually have these fences already on some intersections.
NZ is already like that. And especially Auckland. Just the fines are missing.
You’re absolutely correct.
American roads, mainly highways, straight, lane delineated, are very different from roads in many NZ cities – ie Wellington’s Kelburn, Brooklyn, Ngaio, Aro Valley etc – I’d like to see an autonymous car try and tackle Devon St in Wellington. Fairly sure it will just give up and shut down all systems…
The vast majority of American roads and streets and not interstates or other highways, they are regular roads like ours.
Also the vast majority of roads and streets in NZ are not like Devon St.
My point, jezza, is not that all streets are like Devon St (which has to be seen to be believed) but that America is largely ruled by the grid. Almost an entire continent of straight lines, at 90 degrees to each other, and then curly bits in the cup de sacs, all largely flat as a pancake. It is the sort of thing that a computer-based system can cope with easily, but the rest of the world is not necessarily like that. The rest of the world – especially the old world in Europe, and the hilly bits like NZ, Greece, Peru etc – are unlikely to work well with autonomous cars.
It was an attempt to show that America is designing things for its rules, in its conditions, and that won’t necessarily work in other parts of the world. Do you get what I mean? Or shall I just go and bash my head against another brick wall for a while?
Yeah, that’s not true about the US. It’s a big country and has thousands of miles of old pre-automobile routes and windy mountainous bits too. Please don’t bash your head against any walls, but please do appreciate that your premise is flawed in that particular case.
This is a great post and discussion. The issues will be raised and debated in a workshop in Auckland on 2 August. Here is a link: https://www.eventbrite.com/e/preparing-cities-for-a-transportation-revolution-auckland-tickets-47661593108
In terms of the hype cycle, AVs have passed the “peak of inflated expectations and now are entering the decline into the trough of disillusionment.
Xerox was a chemical company that invested all it’s profits for 20 years into developing the 914. It was mortgaged to the hilt and close to bankruptcy by the time it released the 914. That first copier turned the tiny company into one of the most profitable companies in the world at the time.
Getting AV’s to work well will take time and lots of investment. I don’t expect to see them for sale for another several years at best.
Well, you can buy a 9/10 autonomous Tesla right now in NZ. Probably sooner than “several years” away.
I’m talking about a true Level 4/5 AV. Teslas are like Level 2 autonomy.
“Being able to squeeze much more out of existing roads means we won’t need to widen roads or build more of them” – doesn’t this only happen once all cars are driveless? Then we can get rid of lanes and traffic lights etc. But while us idiot humans are driving I can’t see any throughput improvements – they will need to drive as stupidly as us.
So lets say driverless cars were for sale in 10 years time, and lets say it takes another 10 years before they are affordable, it is then another 14 years before the average NZ will own one, and probably another 6 years before there is a big enough majority to ban humans from driving. That’s 40 years away.
Mmmm, exactly. We could make some fairly well-educated guesses about what changes to our energy patterns and enviro-political stability we’ll see in 40 years time, and I don’t think that planning on continuing to transport people around in individual cars as a common thing is particularly wise. The level of techno-optimism shown as our civilisation declines is one of the biggest sources of amusement for me.
Actually, the opposite is probably true. Mixing a few autonomous or semi-autonomous cars into a vehicle fleet still mostly driven by humans will reduce speeds (the AVs always obey the legal speed limit, forcing the rest of the traffic to do likewise). So we’ll need to widen roads and build more of them.
Most people overestimate what they can do in one year and underestimate what they can do in ten years
True autonomous vehicles may still be decades away, but in the meantime the same technology can offer some much simpler improvements. One possibility that should be of interest in NZ: a car that cannot be driven on the wrong side of the road. If the driver steers the vehicle into a lane where it’s going in the wrong direction, the software automatically sounds the horn, vibrates the driver’s seat, turns on the hazard flashers, etc. If the driver persists, the car automatically pulls over and stops.
Would be pretty easy for the cars to integrate with the traffic light system and pull on the brakes if you try to run a red light.
Also could stop speeding, driving without headlights, not stopping at a stop sign, drink driving, pretty much everything.
Also overtaking on blind corners, tailgating, failing to give way, etc. etc. The list of such features is quite long, once you start thinking about it. All much easier to implement than full vehicle autonomy, so perhaps we can expect to see it a lot sooner.
It would be great if the writer could have contextualised this article by providing evidence for this supposed “hype”. No one that I know or have heard of has, for example, argued that driverless cars would be on the market within the next 10 years. This piece is built on a strawman!
I’ve read several articles saying exactly that. People saying no new petrol cars after 2025, or saying : Our generation will be the last to actually buy a car etc.
One of the benefits of the hype and massive investment in to Driverless Cars is the opportunity to incrementally improve the current range of new ICE and Electric Cars.
In my lifetime I have seen improvements to cars that are now ubiquitous and often mandatory like:
Retractable Seatbelts
Air Bags
Side Air Bags
ABS
Traction Control
Crumple Zones
Reversing Camera’s
and the new Driverless Car investment will allow for the rollout of
Lane Control
Automatic Braking
Adaptive Cruise Control
All way before full autonomous cars are common.
I think in my lifetime the car fleet will be semi-autonomous so for example those who need to drive on a holiday weekend can virtually connect their car to one in front and drive in an easy convoy.
My own headline for this would have been “Driverless Cars Progressing Slowly AS Expected”; those who are in the industry looking into this issue are very aware of the technical, political and societal issues that have to be addressed before Level 4 AVs are even a reasonable proportion of the landscape (it’s also likely to be truck and bus fleets before Mr Jones’ personal commuter car). And yet the focus on AVs at the Trpt Ministry level continues to be perplexingly high – same for EVs (also struggling to hit 1% of the fleet anytime soon).
Personally, if we are looking at what is actually going to have a meaningful impact on our transport patterns in the coming decade (e.g. the timeframe of the current GPS) I’d be paying far more attention to Mobility as a Service, and the need to support the various transport offerings that this might entail (PT, bike share, car share, Uber, etc) and the payment/information systems required to make it work well.
“So, basically most pedestrians will be squashed by these things.”
An egregious non-sequitur. The list is from a 2015 US-spec Toyota Camry and refers to the enhanced reversing camera features fitted to a regular human-operated car (page 312).
I think AVs will eventually have to be imposed by legislation, not public demand.
There is the view that public will fall in behind them due to improved safety. I find this view completely bogus, as the widespread resistance in NZ to current viable public safety measures (e.g. lowering open road speed limits) suggests that there is a tacit acceptance of the current riskiness of car travel.
That is, the average person considers that the current risk of death or injury is outweighed by the utility gained by being able to drive in their current manner.
Once you remove the human aspect, autonomous cars will be very hard to adjust to. They will often flail aimlessly, act seemingly erratically and ultra-conservatively. They’ll crawl through areas with narrow roadways, and/or a lot of pedestrians, not using their intimidatory factor to proceed through as a human driver would. When they get on the road, the way they drive will frustrate non-AV car users, likely leading to abuse and a reluctance to drive again – capping usage before it can really get off the ground. You’re effectively making your car a “grandma” driver – you can see that they’ll be hated by the average punter.
AV’s aren’t really a matter of technology, they’re a social issue.