There was a news article a week or two back saying that driverless cars currently under test in California had been involved in four collisions. This sounded bad until you dug into the details and it turned out that in each and every case it was a human driver at fault. As Nassim Taleb points out there is no such thing as confirmatory evidence, but this in no way falsifies my theory that driverless cars are already safer than their human-directed equivalent.
This makes me think that the driverless car revolution is on the way and is going to take place far sooner than most of us think. Yes, there are legal issues to be resolved. Yes, government will drag its feet. Yes, there will be horrible accidents of the sort only computers can cause. Yes, there will be a transitional period of mixed human and computer driving. But it will happen and it will – over all – be better. But given it is going to happen I wonder what it will be like? For instance:
- Will cars continue to be user-owned? Will we even have “our” exclusive cars or instead use cars in the same way we use taxis today?
- Could this make micro-cars more attractive?
- Will styling continue to be so important?
- Is there anything to prevent a speed-limit of 120mph, or higher, on motorways? If so, what future inter-city trains?
- Will this advantage electric cars?
- If buses can self-drive is there any future for commuter trains?
- If cars can drive themselves to and from our doorsteps will we still need driveways?
- Is this good or bad news for Uber?
- What will cabins be like without the need for a driver and a steering wheel?
- Will there be implications for the layout of vehicles?
- How soon will it become illegal to drive a car on the public highway?
It’s going to be fascinating to watch.
Hopefully they’ll look better than this.
Why do these driverless cars have to be such poxy little golf carts? Make them look like a 1959 Cadillac El Dorado and people might start to get interested.
@JohnK, these are test models which will likely have been designed to be relatively cheap to construct, and possibly to modify. I expect (at least, I hope) future driverless cars will be considerably more stylish.
The Tesla Model S all-wheel drive has some driverless tech in it. I think we might see a lot of higher-end vehicles equipped with driverless tech so that they can self-drive in more predictable environments such as motorways and dual-carriageways while the driver will still be able to take over on remote roads (where data services are unavailable or spotty, or the SatNav data inaccurate [lots of SatNav data is inaccurate, this could prevent the driverless tech working optimally). The prospect of faster, safer journeys on motorways is great.
Will you be able to hop in, have it take you to th pub, sink six pints and then have it take you home – I‘m afraid I doubt it.
Would you pop your seven year old daughter in one, alone, and have her delivered to school?
We need driver-less trains first, that don’t go on strike and don’t take corners at high speed, this has been do-able for decades, but human self-interest has prevented it, I’d forecast the same for cars.
Runcie, I thought of that too. However vested interests have a much more powerful effect in the nationalized/corporatized railway companies than in the much free-er market of private cars. Road haulage interests like URTU will almost certainly agitate against driverless tech for a long time but as the tech becomes authorized for use by private vehicles it will eventually become impossible for road haulage unions to continue effective resistance.
One wonders what the test drive will be like.
Sorry, I remain unconvinced of this happening, at least in my lifetime.
Too many years with electronics and computers prevents me from having that type of complete faith in those things when it comes to my life !
This whole idea / project is being pursued / driven, for the most part, by a bunch of wide-eyed dreamers who are so caught up in the idea of SOMEHOW doing this that they getting tunnel vision without consideration of consequences, ramifications etc.!
First, one of the problems, at least on the ‘west’ side of the pond, with poor driving / accidents, is that we no longer teach people HOW to DRIVE ! If junior can get dad’s land yacht around the block without running into anything, he gets a drivers license and is turned loose on the rest of us. There is no situational awareness, acceleration/braking/handling ‘realities’, avoidance maneuvering etc. etc. taught anywhere except for competition driving schools. And don’t get me started on texting / distractions. Drivers, especially young ones, try to do far too much with far too little experience so that the act of operating the car is about 4th on their priority list ! I actually saw a young woman, one morning, on an interstate, at about 75 MPH applying eye makeup and eating !! Wonder if she even wondered where that eyelash curler would end up if she were to suddenly experience sever, unexpected deceleration !!
Second, we bundle up kids with knee / elbow pads, gloves, helmets, on and on, to ride a bicycle for crying out loud. Let them skin a few knees, break a bone now and then and they learn, to a degree, not to do STUPID things with mechanical devices. Then we put these children who have NEVER hurt themselves in a car with anti-lock brakes, airbags everywhere including the glove box and tell them, in so many words, ‘it’s almost impossible to get killed in this technological safety cocoon’. Then we wonder WHY they drive the way they do !!
At least over here, these ‘developers’ are going to run into a legal reality. ANY and ALL injuries / deaths are going to be blamed, whether true or not, on the ‘automation’. There is no such thing as an ‘accident’ – someone &/or something MUST be responsible ( blamed ) and therefore someone or something MUST be forced to pay ! This will lead to the producers of said devices to be sued literally out of business or so close to it that the idea will be forever tarnished. Don’t believe me ? Talk to some people in the general aviation industry ( Piper, Cessna Beechcraft etc. ).
This whole idea will be supported and encouraged by governments the world over and we may very well be forced into this situation even with its faults. After all, you really think governments give a damn about a few deaths or injuries ? And besides, what a wonderful way to better keep track of and control over at least some segment of these unruly oafs who are only useful for paying taxes and voting !!
Just to give one example, for those of you who actually trust the world of computers.
Years ago ( more years than I care to think about ) I worked on a series of text processing programs ( used for publishing, page format, fonts etc etc ). The programs had gotten so large and complicated that a change to a single line of code required a man-year of testing before it was ever released !! You want to know something ? Bugs in that code are STILL being found and fixed today !!
So you go right ahead and strap yourself into that rubberized roller skate and go flying down the road with the complete knowledge and total conviction that absolutely nothing can go wrong go wrong go wrong go wrong …………………
It occurs to me that a great deal of the styling of modern cars – not to mention the weight – is bound up in safety measures: crumple zones, side-impact protection, bonnets (US=hood) of a certain shape. But with cars that are almost impossible to crash lots of those things can be dispensed with. Yes, you can have your Cadillac Eldorado if, that is, you still plan on owning/possessing your own car.
I have no doubt that driverless cars will find all sorts of new ways to kill their occupants and others. But the over all total will be smaller. Much smaller.
The real test will come when we have a neighborhood full of these things and no human drivers.
As to reliability. Aerospace (aircraft) manages this quite well. Everything important is at least triple. Or the left wing flap can serve if the right wing flap is disabled.
Depends on how much you are willing to spend for how much reliability.
One question you didn’t ask is:
– Will one of the passengers have to be awake and alert enough to take over driving the car if something (GPS, etc.) breaks down?
– Would you get a DUI if you’re drunk as hell and your driver-less car gets into an accident?
One of the great things a driver-less cars would do is make commuting to and from work suck less. In the morning you could maybe doze off, or eat breakfast, or check e-mails, or get ready for the days work. On the drive home, the same thing, plus maybe pop open a cold one. However, if you’re required, by law, to constantly monitor what you’re car is doing, and being ready (and sober enough) to take over in case things start to go bad, it wouldn’t be much less of a strain than in the old days of driving yourself.
I sincerely hope that Jerry is correct or even better over-estimating the capacity of computerised crap.
There is nothing whatsoever fascinating about watching tyranny come to pass. Indeed a car-crash in slow motion would be a very apt metaphor.
The political scum will know where you are at every moment, be able to switch your car off (“Minority Report” style) or lock the doors and have it cart you to jail. You will travel where and when they let you and if something is going on they don’t want you involved in or don’t want you to see–the system will be “locked down” (jailhouse speak–itself the trappings of tyranny) or will inform you that there is too much traffic in those areas–“Please try again later”.
“How soon will it become illegal to drive a car on a public highway?” It is time people curbed their foolish enthusiasm for this kind of techno-junk and recognised freedom as being worth more than all the tyranny-enhancing toys the future might bring.
All of you guys who think that the software will work flawlessly –
Are you still restarting windows? Do you still get multiple regular upgrades to fix crap software and operating systems? Do the upgrades sometimes work less well than their predecessors? Have you been hacked? Gotten a virus? Blue screen of death? Does your browser freeze regularly? How often do you get “this program has stopped responding” messages?
And life-critical self driving software will be better how?
I write the stuff and while software is improving, we have a long, long way to go before I ride in a self controlled car in a western Canadian blizzard on a mountain road.
But you can go for me. I’ll watch.
No Fred, I use Linux. MS Windows is not a good example of how software can be.
GM was promising the world driverless cars as far back as the 1930s.
Of course trolly cars, trams and railways served much the same purpose as diverless cars.
But the government undermined them – with price controls (Los Angeles once had the largest mass transit system in the world – but price controls destroyed it in the 1930s).
And by subsidised road building.
Get rid of the subsidised road building and road repairs.
And get rid of the web of regulations on railways and so on.
And transport would become more rational.
I feel the hot drooling breath of bureaucrats salivating at the prospect of eliminating signal lights in cities, and improving commute times, and so being able to prevent anyone anywhere from disobeying any of their stupid road rules, as well as all the social engineering they could do with this power. Will the highway speeds go up as a consequence? No, that would be inefficient use of fuel, contributing to climate confusion. They will slow us down, say how many cars may go along which roads. Right now you have to get an appointment to hike a trail in California. They’ll find a way to pull the same stunt with cars, closing roads through land needed for migration of some rare creature we haven’t discovered yet. Quad erat demonstrandum. Et cetera. Ad infinitum.
Fred, keep in mind that Windows (and other horror-story examples) are extremely complicated programs designed to do a nearly infinite number of very different things. Auto driving software, by contrast, would be focused on one single thing and would probably do it very well. That’s especially true if it were restricted to limited-access highways which had guidance sensors physically embedded into the roadway (to minimize reliance on satellites). We already have auto-pilots which handle most of the commercial flight, with the human pilot taking over only on takeoff and landing (or in emergencies). I see driverless cars using much the same approach, with the human driver taking over on local roads. (Which, unfortunately, would eliminate the possibility of your car driving you home after a long night at the pub!)
And I expect the first application of this technology will be in long-haul commercial vehicles, which will eliminate much of the problem of driver fatigue (and the excessive regulations adopted in response to it). Personally, I would be much happier to drive on highways along with driverless trucks than with human-operated ones.
I say, START with driverless school busses. Then move on to driverless “official gub’mint business” limousines/SUVs. Maybe THEN graduate to “entertainment industry” tractor trailers/tour busses/mobile homes.(caravans?)
If it crashes as often as windows then heaven help us
Assuming these work safely (a big assumption) I still see problems. It would be very hard to use one from a remote house location if it wasn’t owned and already stored at home. It would also be difficult to persuade them to go off-road, and some vehicles need to go off-road. Then the question of what happens when a teen with a laptop hacks it to go twice as fast or it has to swap road sides for a french holiday journey? Also they must have some method to ‘manually’ instruct them to move around – say in a garage for maintenance or in a car park, how does that work if they have no steering gear at all?
Yes. And no. Depends. Here in the majority of America, we’ll still have our cars. In the big cities and those parts of Europe where car ownership is rarer, this will reinforce that trend.
Nope. If anything it’ll make larger cars more attractive. I can have a small office suite or keep the comforts of home at hand since I don’t have to find a place to park a large van.
Absolutely
Yes, political interference. Same as it does now to limit interstates to 70 mph even on laser straight stretches.
Not at all. Electric cars will get better, but they don’t have any particular advantage just because they’re driverless. If anything, until they can match FF vehicles for range they’ll be at a disadvantage – I’d prefer to take most trips in my driverless car rather than fly and an electric car doesn’t have the range to compete.
Nope.
Yes, because I will still own my own car and won’t want to pay to garage it elsewhere if I have a home. It can reduce the need for *parking lots*, but not driveways I think.
For *Uber* – good news. They can just pivot to providing links to carshare providers. For *Uber drivers* – they better get a self-driving car and start renting it out.
More group-centric – inward facing seats, reclining, drinks cabinet. Rolling parties.
See #2 and #9.
Dunno. In America, probably a long time. And a long fight against it being mandated once the government realizes they can control movement easier if everyone is in a self-driving car. In Europe? It will probably parallel the final outlawing of cash. Consider that a canary and within 10 years of it happening (if it does).
Yes we will still own personal cars, especially in more suburban and rural areas where a wait for an Autonomous Vehicle (AV) cab would be inconvenient. Plus knowing the previous passenger did not vomit in the seat is always a selling point.
In areas with lots of commuting, yes, for cab fleets where carrying cargo is less important.
For personally owned cars, certainly
Only the presence of unpredictable human operators.
Maybe, if the AV cabs can auto-recharge. If my Roomba can, so can they
Good question. I suggest no, but only after years of political posturing and wrangling. Railroads will largly relegated to freight use.
Yes. For foot access from the road to your doorstep, or for personally owned AV’s, or for your bicycle. You may think only a narrow path is needed for most of those, but one you go to the expense of a 3 foot wide path, adding a few more feet for an AV is a good hedge
Very good. Uber is testing an AV in Pittsburgh right now, and they are on record as saying they will offer AV service as soon as it is legal. IMO, it has always been clear that they are using Human drivers to collect data on volume and traffic that they might expect for a fleet of Uber-owned AV cabs.
Initially they will be very like current cars, but in the future they will look like a circular sofa so everyone can face everyone else. Think Heinliens Job: A comedy of Justice.
Clearly
I suspect decades, if ever. But we will likely see huge penalty increases for misbehaving, and having a collision.
Will cars continue to be user-owned? Will we even have “our” exclusive cars or instead use cars in the same way we use taxis today?
Why are these people trying to pry our cars away from us?
Could this make micro-cars more attractive?
Why are these people obsessed in cramming us into golf-carts?
Will styling continue to be so important?
Why do these people want to ensure our public *and* private property is depressing and ugly, and want to drive all art from the world?
Is there anything to prevent a speed-limit of 120mph, or higher, on motorways? If so, what future inter-city trains?
Yes: The other cars on the road. Our highways are already too crowded to do much over 70. Part of this is that the trucks have lower speed limits than the cars, part of it is there are twice as many drivers on the road than their were a generation ago.
Will this advantage electric cars?
Greater energy density (by a factor of 5 or so – it has been getting better) and/or totalitarian political control over our driving choices will advantage electric cars.
If buses can self-drive is there any future for commuter trains?
If cars can drive themselves to and from our doorsteps will we still need driveways?
Why are these *same* people obsessed with robbing us of our yards. Why do we see articles claiming that wanting a single-bedroom house/apartment to yourself is selfish?
Is this good or bad news for Uber?
What will cabins be like without the need for a driver and a steering wheel?
Very dangerous in the inevitable event that you have to take personal control.
No Fred, I use Linux. MS Windows is not a good example of how software can be.
There’s a reason why Linux hasn’t gained more than 2% user-share like it swears it will any day now.:-P. When your terminal is barfing at you about package X wanting last year’s version of libwhydoyoueverchangethis.so, and when you downgrade you can’t open Firefox anymore to find the arcane terminal commands required to stop yum from panicking: rather than melting down in Lovecraftian terror, we crack open another computer (yes, we have them – probably more than two) and grind through it. Normal users lose their sanity. I still have to walk acquaintances of mine through how to find their data files in Outlook.
When you want to compile a library, only to find out that the maintainer only sort of half-assed an install script because he had better things to do than make sure all the symbols were ending up in the object files, and if we needed to, we could always go through the makefile to get the gist and do it ourselves …
PS – I love Linux. I work with it a lot. I have several linux computers that I play with. But I also have a busted installation of redhat 5 on my student workstation that won’t update anything because it’s out of memory in the boot partition and can’t store the latest kernel, which for some reason everything newer than 2008 depends on. (*I* didn’t set it up. I wanted to install Mint. :-P)
The human brain has billions of neurons and tens of trillions of synapses. Our internal representation of the external world is made of billions of values being turned over in parallel at 10-100 Hz. Our senses have very good fidelity: our eyes automatically adjust their focus and aperture, our ears also contain internal gyroscopes, our cerebellum is running control algorithms beneath our conscious attention that allow us to do something like drive (or ice-skate, or hang-glide – all learned unnatural behaviors) without even paying conscious attention to the details.
Our computers hold maybe a few MB of data (maybe GB, though if you’re talking about current embedded stuff, forget it), and process them in serial at 1-2 GHz. They see the world through relatively grainy cameras. They fake actually knowing where they are the same way we do, by synthesizing tons of cues about the nature of our surroundings, because they have GPS sattelites telling them their location. We *could* keep bolting more sensors with more data on, but then you have to *do* something with that data, and modern computers suck very badly at synthesizing it all into reasonable judgements about what is *out there*. They see the world through maybe a dozen pins. We see the world through millions of neuron thick connectors.
I remember sitting in a presentation once where the presenter was discussing an aircraft accident. The pitot probe clogged, so the avionics couldn’t figure out what the altitude was. Because it thought it was too high, it *would not let the pilot deploy the landing gear*. The pilot eventually had to skid it along the runway into another hangar without his gear because the computer thought it knew better than the pilot.
The lesson the presenter took away from this is “clearly we need more intelligent avionics on our aircraft, able to combine *two or more* signals to determine its true situation.” When someone asked why the pilot didn’t have an override, the presenter was baffled about why we would *want* that.
If it all works driverless cars come about slowly as cars get more and more automated. there will be many advantages from almost no crashes and the elderly will keep their mobility. As accident rates fall insurance premiums will collapse on automated cars. Meanwhile the non automated pool of insurees will shrink. The market will make it impossible for the young to afford insurance and the younger generations will no longer learn to drive manual cars.
Proponents of automated cars wax lyrical about auto cabs and never needing to park again or own a car. Or that people will live further out, or pubic transport will die etc.
But their are fundamental problems off all of this. unless battery technology changes people will still use petrol engines. automated cars will still need to be fuelled and serviced, probably to higher standard than normal cars. No manufacturer would leave itself open to a lawsuit of part failing and causing a crash while the car is driving itself.
Your mileage starts to collapse over 70 mph, so while it is technically possible to travel at 120 mph, who could afford to?
So the idea that your car will drop you off at work and go home again is fantasy and so is the idea that it could park on a residential street near your destination and pick you up when needed.
While cars could travel closer together on motorways on city streets they would need to mix with non motorised traffic. The streets are only so big. Cities barely move now what if they are full of empty cars trying to mover themselves around between home and work or providing a school run.
What will matter is how much a taxi drivers wage is part of the total cost of ownership. If it’s high then many city dwellers could give up cars for on demand services.
But think through the consequences like all form of public transport for work related purposes vast amounts of capital is spent trying to support the peak travel times. The solution for automated cabs is variable pricing.
So in busier cities public transport will thrive, especially if buses can be automated too. In Britain with it’s high transport fares where most bus services are already commercial the economics of bus travel could be transformed with much lower costs.
Train travel is likely to survive in most forms in Britain. Trains can carry large numbers of people. The London commuter belt is pretty much secure, unless anyone fancies spending a hundred billion of so bulldozing a motorway network through inner London.
On current Network rail finances intercity rail and the London commuter rail network cover their costs or makes a bit of profit. It’s the rural routes far from big cities that require the big subsidies.
Of the £4 billion of so difference between spending and ticket sales on UK rail, £ 2 billion is accounted for by Scotland and £1 billion by Wales. So for Metropolitan England rail use will probably continue to grow.
Alex
May 22, 2015 at 7:56 pm
Well, I use Forth. http://spacetimepro.blogspot.com/2014/12/fish-forth-v15-lpc1115-stm32f4xx.html
It is really good for these kinds of hardware intensive applications. It encourages well tested short snippets (the way programming should be done). And if it is your chip’s native language aka the RTX2010 (on the Philae comet mission) it is really fast and really low power.
Part of the reason we are in the software mess we are in today is that every single compiler is a two stack (or more) machine and yet the vast majority of our chips are at best one and a half stack machines. And in lots of others the second stack has to be totally emulated in software.
This is stupid beyond belief given that a second stack adds very few transistors and those few transistors are incredibly cheap. An up down counter and 16 or 32 words of RAM would do very nicely. More or less a duplication of what every processor already has.
This is the future of long distance train travel:
http://www.cnbc.com/id/102458347
Assuming we can put an end to these sorts of Gestapo tactics.
http://www.theatlantic.com/politics/archive/2015/05/how-the-dea-harasses-amtrak-passengers/393230/
cavehobbit
May 22, 2015 at 10:39 pm
Re: recharging car batteries. The power grid does quite nicely with kilowatt – use it anyway you like – loads. Megawatt loads need coordination with the grid operator. This means a “Smart” grid. You have now opened your grid up to hackers. Uh. Oh.
The trucking industry will likely be the first to use driver-less vehicles for commercial purposes on public roads. Some mines in Australia are already using driver-less truck on private roads. The savings from being able to operate such large expensive vehicles 24/7 without driver breaks are considerable. Also, the trucking industry could first open up routes between hubs connected only by motorways and this will be an easier sell for the regulators.
Since my knee-jerk contribution above, I have been thinking more… and finding myself in an ironical situation. As a technologist I have long shaken my head at people who oppose new technology. Yet since Google’s involvement and evident success in this particular field, I have become extremely wary of the consequences of driver-less vehicles.
I love to drive; I love the feeling of freedom, the manifest control and feeling “one with the machine”. I fear the heavy hand of the state to take this joy from me, in dribs and drabs. Yet I recognize the parallel to the man who railed against not being able to ride his horse down Main Street, in this new-fangled age of automobiles.
So I ask myself, is there a difference here, in this situation, that ought to make us pause at the advent of this technology? Certainly the introduction of the sewing machine was an easier case. It had its cultural consequences, but the action of using one (or choosing not to) happens in private, be it home or business. The state has no calling there.
Now this has to do with transportation, and anarcho-capitalist dreams of purely private roads notwithstanding, land transportation largely involves publicly controlled roads, and so some manner of compromise and rules collectively determined.
And so all I can say is that there may be a great many wonderful consequences of this new tech, and I will embrace them gladly. I will hope in parallel with that acceptance that the Joy of Driving will be such a value to all that the state will not overreach to the complete destruction of that Joy.
With the lyrics of Rush’s Red Barchetta echoing in my mind…
I’m gonna be lazy and repeat what I said last time this topic came up on here.
I don’t see the future of driverless cars coming any time soon, mainly for reasons of cost. What it comes down to is the fact that reliability is expensive, and the levels of reliability required for driverless cars would be along the lines of those seen in the airline industry. Expensively reliable systems are cost effective for mass transport machines like aeroplanes because the same system can transport several hundred people at the same time, but to put such a system into a machine for 4 people? Something would need to change.
I know quite a lot about reliability. In the design of an oil and gas installation there are many, many automated actions controlling the process and shutting down the plant in the event of an emergency. The devices which govern the actions (which includes the entire “loop” from initiating the action to executing it) are rated according to what is known as a Safety Integrity Level, or SIL. A SIL-1 system will fail every 100 times it is asked to do something; SIL-2 every 1,000 times; SIL-3 every 10,000 times, SIL-4 every 100,000 times. For non-critical actions you use a SIL-1 or -2, but for anything which will shut down your plant in the event of a gas leak or fire you will need a SIL-4. The reason we don’t install SIL-4 on everything is because a SIL-3 or -4 system is staggeringly expensive compared to a SIL-1 or -2. As I said, the whole “loop” needs to be rated: your gas detectors, your Process Logic Controller (PLC), your valve actuators all need to be of the same rating.
Then you have redundancy: anything which can shut your plant down should do so on the basis of 2 out of 3 voting. So you put up 3 gas detectors (all of which are SIL-4) and the PLC will only initiate the action if no less than 2 gas detectors say the same thing. This prevents false positives, or spurious trips as they are known in the industry. I can tell you from experience that the SIL levels and redundancy levels required to achieve the reliability seen in aircraft, the nuclear industry and (snigger) the oil industry (stop laughing at the back!) is seriously expensive, and it is not simply a case of the price of electronics falling and the Chinese will take care of it all. This kit is high-end stuff, and rigorously tested, plus it needs to be maintained properly.
I’m not saying that SIL-4 levels of reliability will never be cheap, just that we are one hell of a long way from it being so now. And I would like to see the cost of a production car with SIL-4 levels of reliability if it were made now. I’d be guessing somewhere between $500k-$1m.
This thread at Tim Worstall’s place is worth reading too. As I said there:
Does anyone care that the motor car is the most loved invention of man? Or that I do not want to live for one day after it becomes illegal to drive mine?
Expensively reliable systems are cost effective for mass transport machines like aeroplanes because the same system can transport several hundred people at the same time, but to put such a system into a machine for 4 people? Something would need to change.
Mass production changes the equation. Design once – repeat 10 million times. Quantity has a quality all its own.
Many of the major systems in autos are currently “by wire”. So where are the costs? The sensor package and testing the software.
The costs are in the reliability assurance, and quantity does not imply quality. A lot of it is down to how the stuff is manufactured, particularly the PLCs. You’ll notice that despite the iPads and other electronic goods, we’re not flying around in Chinese airliners.
This stuff is already mass produced, and it still makes even oil companies balk at the costs.
Tim Newman
May 23, 2015 at 7:29 am
What you are leaving out is that if there is an incentive (mass production) the costs can come down quite rapidly.
In aerospace a LOT of the testing is currently “by hand”. The same for nuclear (although I have been out of that business for 50 years). You start getting the hands off and your costs start looking like material (at the smelter/refinery) + energy + profit. If you took the materials a car is made of and priced them (per kg) they start looking like a quote from the commodity exchange.
This stuff is already mass produced, and it still makes even oil companies balk at the costs.
Thousands is not the same as 10s of millions.
Can you name me a single system or piece of equipment consisting of more than 20 parts and a PLC which is produced in the tens of millions and fails at a rate of less than 1 in 100,000? Perhaps a pocket calculator or Casio watch might meet the criteria, but not much else I suspect.
Plus, the oil and gas/aviation/nuclear processes are vastly more simple than that required to control a car driving down a road. So your starting costs are probably an order of magnitude higher before you’ve even taken into account volume savings.
It is curious to me that while Patrick hints at his expectation for the state to try and delay the full arrival of the driverless car, others here seem to assume that the state is likely to promote it. Me, I think both possibilities are equally likely.
As can also be seen from the comments here, some of us want this to happen, while others much less so* – which leads me to think that the same is going to be true among both politicians and their lobbyists. While in an environment free of government interference the different preferences would still be there, but would balance each other out in practice, eventually arriving at some kind of coexistence – in reality it will be politicians, their lobbyists and bureaucrats who will decide. The different preferences and interests will wage a war against each other, but it will be waged without the participation of the little people and at their expense.
The end result will be anyone’s guess: the driverless car may win or lose in the long run, but somehow I’m pretty sure that the outcome will not be the optimal one for all of us.
*My personal preference happens to go both ways: sometimes I like to drive and be in control, and some times I like to take a nap on the way there. Problem is that I doubt that our betters are going to grant me that kind of flexibility, at least not nearly to the full possible extent.
Tim Newman
May 23, 2015 at 9:49 am
I assume you are conversant with the electronics market. We are starting to see automotive rated parts. And their costs are not much different from unrated parts. And your point about calculators is an important one which you failed to recognize. Demand will call forth supply.
I agree that self driving cars will not be soon. But they will be sooner than you think.
Something else not mentioned so far. Car pools.
I assume you are conversant with the electronics market.
No, but I am very conversant with the design of automated detection and control systems, particularly with regards SIL levels and reliability of said systems, the procurement of the components, the testing requirements, and their associated costs. This comes from spending 3 years (2010-2013) being technically responsible for this stuff on an operating facility.
Rather than missing the point about calculators, you appear to have missed mine: you can make cheap, mass-produced and very reliable electronics but they tend to be extremely simple. I notice that you declined my request to provide an example of a mass-produced, reliable system of more than a handful of moving parts.
BTW when designing my (for now) low volume stuff my preference is for Automotive Rated Parts.
A nice little PDF about “Q” rated parts.
http://nepp.nasa.gov/workshops/etw2013/talks/Tue_June11_2013/0930_Sampson_Automotive%20Parts.pdf
OK. PCs have gotten quite reliable – if you don’t count software. And automotive electronics reliability is still improving.
As you point out simplicity helps. Quite a lot. Ease of testing helps. Quite a lot. And the languages currently in vogue are godawful in that respect.
Aside from the complexity, ugly syntax, and being at the mercy of compilers “C” is a one and a half stack language. You pay rather a lot for its stack thrashes. And guess what? Processor design has followed language design. Thus the proliferation of one and a half stack processors. The ARMs. I use them. I like the peripherals. But the half stack idea is a crippler.
But it does tend to keep out things like Forth which could simplify development by a factor of 10 in most cases. And by a factor of 1,000 in a few cases. Why? Because Forth promotes a different kind of thinking than “C” does. This kind of effect has been noted in natural language. Some things easy to express in one language are inexpressible in another. English “adopts”. And so it adapts. And every day there is more you can express in English.
And Forth? Well written it reads very close to English. That helps greatly with the thinking.
What a great comments thread! I particularly liked the contributions from cavehobbit, Myno, ams and MSimon.
ams, my point about Linux was not that it is some cure-all software merely that Windows is not a typical example of the reliability of software. Most desktop Linux distros are really unstable compared with *nix in commercial and industrial use yet considerably better than Windows. Even Windows, when well tuned and focused to specific tasks can stay in continuous use for months. Generic software supporting a diverse user-space is really difficult to keep lean and stable whereas tightly focused, rarely and carefully modified software can continue to run indefinitely. Also see MSimon’s point above on the architecture and electrical engineering of computers.
Driverless tech is really, really hard but as an assistive technology on graded minor roads upgrading to fully driverless on motorways and other grade-separated highways it would be a compelling feature permitting much faster journeys, longer commutes (especially important in Southern England, for instance) and better safety ratings. As the technology matures it will become applicable more widely.
unless anyone fancies spending a hundred billion of so bulldozing a motorway network through inner London.
There is a motorway network through inner London, it’s just that at the moment it’s used by trains. It would not be that expensive to pull up the tracks and turn the routes into roads. Add in the overground and underground network and you have a pretty nifty way of getting about in your driverless vehicle.
Of course, you might run into some opposition from the current users of the network, but if driverless cars take off, it’s going to happen.
Me, I’m still waiting for my driverless, (pilotless?) flying car that I was promised sixty years ago.
In Europe, particularly the UK there seems to be the political will to bring driverless cars in. Car rental schemes like Zipcar would likely takeoff big time in cities, if you can summon a car with one click from your Smartphone/laptop or tv. Cars currently spend something like 95% of their time sat around, and have fairly high ownership costs especially with parking in cities- think of central London, San Francisco or Tokyo. Not having to park a vehicle but getting the nearest one to you would be a major plus factor.
As the technology becomes everyday, I think driver controls will be phased out, perhaps over a 30 year period. I’d imagine private ownership would continue as well, especially for non city dwellers.
But some of the interesting problems have to be worked out. Could one use them as a suicide bomb delivery mechanism? How would you limit or allow access to sensitive areas? “I’m sorry David, I don’t know where Porton Down is?”
Not necessarily a “smart grid”. It would certainly require coordinated planning between any fleet operators and the local power company. A fleet operator could easily set up distributed power-stations around a city to spread out load across a grid, or use replaceable power-packs, like on your battery powered hand drill, that are recharged during off hours.
Those are just a couple option I can think of, there are probably many options I have not thought of
It is always interesting to see the enthusiasm and dis-enthusiasm that people have for wonderful hi-tech solutions – and how soon they are going to save (or otherwise vastly improve) the world.
As someone who has been working, off and on since 1976, on Automatic Speech Recognition (ASR) I’ve been there several times. ASR things are, of course, much better than they have ever been before, but still struggling to provide the general competence of an 8-year old. You can see how things have improved by noting my admission that I used to reference 6-year olds. So your mobile will now answer you spoken enquiry “Where is the nearest Italian restaurant?”. But mostly only if you currently have 4G connectivity.
Image analysis of the sophistication needed to drive a car is vastly more complicated than ASR, and with greater downside when things don’t work.
I’d be more convinced if a car could drive totally automatically from my house to Trafalgar Square (around 25 miles). Doing it on a Sunday morning would be a goodish start; I expect The Friday afternoon rush hour would be a tad more difficult.
Of course, having someone (something) sharing the driving on a long journey could be a help, but many drivers make bad passengers, especially if driving conditions are difficult. Just think how much worse it would be, than with the spouse driving.
Here are some incidents in my driving life.
Last week, an accident with 3 cars diverting to stationery on the M27 central reservation about 15 seconds ahead of me; half a ton of stones scattered across 2.5 lanes. Amazingly, all the people nearer than me covered the situation without further impact or stopping. Just what do those emergency flashers mean?
Later the same trip I passed the site of one of my most memorable emergency recoveries, from around 1980. Arriving at a bend on a single-lane A road, I find someone overtaking towards me, and nowhere to go. Fortunately I had enough room to strip my speed down to below 10mph and the overtaker recovered into the gap so created; then I saw the ancient narrow-tyred vehicle behind me was not doing so well on the emergency stop, was in a 4-wheel skid and and was likely to hit me at 40+mph. We did have seat belts in those days, but not airbags – so (at best) hospital for them, if not also for me and my passenger. Fortunately I saw the problem in my rear-view mirror, was already in second and floored the accelerator. Result: no visit to hospital, crematorium, or even repair shop.
Two 25-ish ton lorries (fully loaded with gravel) racing each other on a single-lane B-road come towards me round a sharpish bend. For both of them, their speed made their backends break away and swing across the road towards me, first the one – and then round the bend comes the other with the same problem. Heaven knows how but I squeezed into the non-existent gap between each of them and the verge, having also stripped off much speed, but wanting to pass through the gaps before they narrowed. The 3 following cars did well too, each retreating totally onto the verge, as the lorries swung out further across to be totally on our side of the road – fortunately none upended in the ditch/hedge.
Driving late at night on a country road in fog that became so bad that the front of the bonnet was no longer invisible; dared not stop for risk from following cars; nowhere adequate to pull off the road (combined with a likely 6+ hour break). Wound down the windows and engaged acoustic sensors. Half a mile or so later, the fog cleared enough to get back up to around 30mph.
Now, with enough sensor technology, wonderful pattern matching and artificial intelligence programming, and many decades to make it all work, it is just possible that automation could do something along the lines of all the drivers that avoided crashing in the above, or just giving up (and creating a worse hazard in the case of fog). This is on both the cost of the technology (which will eventually come down) and especially the quality functionality to compete with humans. So I am definitely in the sceptical camp with Tim Newman and others.
And I think such incidents happen hundreds, even thousands, of times every day – just here in the UK. And I don’t think the technology is going to be up to it for decades, unlike many of the human drivers we currently have.
What would be better, as mentioned by Alex on 23 May at 11:12, is using existing to somewhat mature technology – in an assistive manner. For example, sensors that detect the distance to the vehicle (etc) in front (essential for automated vehicles) and displayed that on the dashboard as ‘seconds at current speed’, with suitable visual and audible alarms. This would discourage driving to close to the vehicle in front, given one’s speed. And I’m quite sure that human imagination is sufficient to set up such sensors, processing and display – to be useful rather than irritating.
This sort of technology could be provided now, for a cost manageable for upmarket vehicles. Economies from larger scale production would make it practical for most vehicles.
Best regards
Alisa:
I agree there will be a tendency for that to happen. But, once a certain threshold is passed, what I expect to happen is for voters to begin demanding that manual driving be banned. And it will be for exactly the reasons that Jerry (above) claims that it won’t happen: concern over safety.
I’ve been a motor racing fan since I was a young boy in the early sixties, and grew up to be an amateur racer myself (which I still am). But, even as a life-long fan and competitor, and even through the filter of nostalgia, I now look back at the racing of those days, and before, and can hardly believe it was real because the danger was so much greater than it is today. I watch video of drivers racing the enormous front-engined Grand Prix cars of the 50s, or of the pre-war era, on skinny little tree-lined back roads, without even seat belts, and without even rudimentary protection for the spectators, for heaven’s sake, and it literally makes the hair stand up on the back of my neck. One day, perhaps only a few decades from now, the average person is going to look back at the highways and city streets of today and feel much the same. They will find it hard to believe that we tolerated the carnage and that, as Jerry pointed out, we barely even tried to teach people how to do it properly before setting them loose to wreak havoc.
Hopping in an autonomous car will seem as normal to them as getting on a train does to us. Knowing how to drive will be as relevant to them as knowing how to ride a horse is to us. And, for the most part, the world will be a better place because of it. I think it’s fairly likely that the politicians will bugger it up, and there will be some lost decades as a result. (Consider the history of telephony.) But: Ars longa, vita brevis. Technology will prevail in the end.
Tedd:
Yep, that was my only point, really.
Technology will not win or lose, as there is no war on technology, appearances to the contrary notwithstanding. There is a war on choice – the choice of when, where and how to use any particular technology, and whether to use it at all. There are plenty of politicians who are enamored with technology, eager to “promote” it at any cost (although never to themselves).
As to safety, although a major factor, but not the only one and not for everyone equally – again, a matter of choice.
What a great comments thread! I particularly liked the contributions from cavehobbit, Myno, ams and MSimon.
Thanks.
ams, my point about Linux was not that it is some cure-all software merely that Windows is not a typical example of the reliability of software. Most desktop Linux distros are really unstable compared with *nix in commercial and industrial use yet considerably better than Windows.
I agree. I was trying to be humorous, while pointing out that, while Linux is very nice, it isn’t *infallible*.
What would be better, as mentioned by Alex on 23 May at 11:12, is using existing to somewhat mature technology – in an assistive manner.
There is a war on choice – the choice of when, where and how to use any particular technology, and whether to use it at all.
This , and this. I’m not really against this technology, just as I don’t believe that computers *have* to be less aware of the world than we humans are in principle. (I believe they are right now though, and that every human has the equivalent of a high end supercomputer between their ears, with outrageously better software – even the ones that are texting while blowing through the red light are more aware of the world.) In a libertarian utopian near-future (as opposed to the one we’ll probably get), your driverless car function will be like your cruise control: The driver can turn it on when he’s on the highway and sit back and read a book or something. The autopilot will sound an alarm when something develops that it doesn’t know how to handle (blizzard, deer in the road, other driver doing something crazy, construction zone, etc. Then the driver will grab control to deal with the situation.
It’s a difference in philosophy: Whose tools are these? Whose purpose are they accomplishing? Are they tools owned by and controlled by the user to accomplish the user’s purposes, or are they tools only nominally granted to the user, not controlled by the user, and accomplishing someone else’s purposes to manipulate and constrain the user? Too many people see the promise of technological tools to be that of controlling other people and replacing their will, not in building better tools for those people to aid them in accomplishing what they will.
Indeed.
Yes, the difference in philosophy really is the core issue. I am worried about the potential for evil use of this and other technology, but I do reject anti-technology sentiments. However as the pace of this stuff quickens we (proponents of liberty) must also strive to win the philosophical debates or face a very disconcerting future.
cavehobbit
May 23, 2015 at 12:41 pm
It is not just the wires. It is the power stations. A Megawatt is significant to a power station. To get the effect you are looking for relative to the present system you would have to distribute the cars that need charging among the various power stations. i.e. in different cities. It all has to do with time constants and rotational inertia. So small changes over a 10 second period are manageable. Fast changes in under a second are not. Which is why solar and wind need hot backup.
As long as the number of electric cars is small (less than 1,000 per city) things are probably manageable. You get 100,000 electric cars in a city and you need a “Smart” (hackable) grid.
We will be using liquid fuels for some time to come. You can get a full charge in 5 minutes. And in places that get cold in the winter rejected heat from the engines can be used to keep the passangers warm.
Recharging during off hours. A battery pack is 40 KWh. That translates to 10 KW load for 5 hours (assuming moderate losses). 100 cars is 1 MW. 1,000 cars is 10 MW. Provided it is only off hours and they don’t switch to charge all at once fine. 100,000 cars is 1 GW. Even in off hours that is going to put a strain on the eqpt. Now the load might be 1/2 that assuming “short” travel distances in warm (not hot) weather. And that assumes passengers will do without the heating and cooling they are currently used to.
Lots of unknowables until we get a few thousand electrics in a given general area for a year or two.
But a “quick” switch to mass electrification is not in the cards without a new grid.
Yes, absolutely. The logistics of ownership and even minor roads may change, however.
No. Less so. The major benefits of a micro car (maneuverability, ease of parking) other than some slight increase in miles per BTU will almost disappear.
Yes. Likely moreso since other indicators of status (speed & power) become moot
Politicians. Otherwise there might be some efficiency considerations (which should be decided by the market but likely won’t). Don’t expect to see it in a hurry though (no pun intended)
Possibly to some degree. For example, I am disqualified from using the Nissan Leaf for commute due to the distance. However, there is an ultra-fast charger at the Nissan building just down the road. If my car could go and charge itself, that would make it viable.
There is probably less future for buses than trains. Busses are horribly inefficient (often being driven empty or with many seats), don’t go where everyone needs to go, take many people on routes that are inefficient for their destination and run on timetables that are often inconvenient and are not handy for carrying large amounts of luggage or groceries. The main advantage buses have over taxis are the one-to-one manpower for driving taxis vs the one-to-many for buses and oh look, driver-less cars.
See above. Many people will still need them though. Around these parts it’s not too unusual for the doorstep to be 1/2 mile from the driveway.
Could go either way depending on whether they adapt their business model.
Ask the passenger. Though rearward facing front seats might find a following
Definitely yes. The factor of not having to be concerned about distracting the driver is likely to have some as-yet unconsidered implications also.
10-15 years in my estimation. Ride those motorcycles while you got em.
That’s one of the problems I have with driverless cars. Halfway throught the development, as somebody counts all the parts, struggles with the complexity, and despairs at the costs they are going to say “Hey, why not get the passenger to do some of this?”
My issue isn’t really the complexity of the design or the inevitability of the politicians’ making the usual hash of it. It’s an idea that will last until the first time some Russian hacker syndicate takes over at rush hour and begins the world’s largest game of bumper cars. Have you ever watched their dashboard cams?
And you can have my 1986 Kawasaki Ninja 1000 as soon as you pry my cold, dead fingers off the handlebars.
Dang, I remember when those came out. I was more of a standard kinda guy but would have liked to have ridden one. I’ll hate when it happens too but the days of human-piloted vehicles on public (and general-use private) roads are numbered
And you can have my 1986 Kawasaki Ninja 1000 as soon as you pry my cold, dead fingers off the handlebars.
Likewise – although in my case, boxers.
Also, Americandigest.org today has a chart showing that truck driver is the number one occupation in the vast majority of states in the U.S.
I do find it amusing that near-30 years ago the first Ninja 1000 was 160 mph of screaming death wish – now it’s a classic!
Alisa:
You don’t need a “war” to be able to say that one thing wins out over another, you just need competing interests. In the case of technology, the competing interests are the new benefits and opportunities it creates versus the old values and interests it threatens.
Every technology that has had the potential to change people’s social behaviour has been the subject of demonizing and attempts at state control. Heck, even bicycles were the subject of much controversy and many attempts at regulation, owing to the deleterious effect they supposedly had on the morals of young people.
When I say that technology wins out in the end I mean that, as with the bicycle, the opposition eventually melts away, is forgotten, and comes to be seen as ridiculous by future generations (on the rare occasions they even become aware of it). The technology becomes accepted and, eventually, taken for granted, with most people “pleasantly sleeping and unaware,” as Kipling’s put it.
Tedd, you may be missing the fact that while new technology always faces opposition from certain people, it also enjoys robust support from others, with the more specific point being that one can find politicians in both groups.
To someone who generally tends to favor technology as such (myself included), the support for it seems natural, and so tends to go without notice – while anti-tech attitudes are seen as deplorable, and so tend to get the most attention. Thing is, I’m sure you would agree that so called Luddites should have the right to dislike technology and avoid it at all costs – after all, that’s what the free market is all about. When no coercion is involved – for or against the use of new technology – it’s all good.
But the coercion thing brings us back to politicians, and as much as I am in favor of new technology, I don’t want politicians to impose its use on me no matter how much I may like it, just as I don’t want them to prevent me from using it no matter how much I may dislike it on its own. And make no mistake, there is no shortage of politicians out there who are very much enamored with new technology – as I said above, tech lovers may not notice them because, well, they seem to be just like the rest of us. Problem is, they are not – they are politicians, which means they are there to impose things (“good” or “bad”) on others.
So we are told, it may well be true that a human driver was entirely at fault in each instance. But how does a driverless car behave when it is involved in a collision or suffers a failure? What can the cargo do about it? Can they direct it to as safe a place as possible after a collision? How? What if you get one that switches to Apple Maps?
Looks like there are still a few kinks to work out… http://www.independent.co.uk/life-style/gadgets-and-tech/news/selfparking-volvo-plows-into-journalists-after-owner-neglects-to-pay-for-extra-feature-that-stops-cars-crashing-into-people-10277203.html
The thing looks like a mouse, right down to the scroll wheel on the front. Only the bridge scaffolding on top to spoil the effect.