If it goes ahead, justifying train drivers will get hard.
Also will be a car company not Google. Arrogance of silicon valley getting ludicrous these days. Google might supply the data though.
The UK government is not hard at work amending its legislation to get Google's self-driving cars lawfully on British roads, despite reports to the contrary. The Department for Transport said officials are “keeping a watchful eye” on various driverless car trials, including Google’s, but it needed to see the results of these …
Locomotives and cars have vastly different risks associated with any failure state that results in an impromptu study of energy transference.
Automobiles have a cap on the amount of damage they can actually do. Yes, people might get killed, but the numbers get big from the sheer volume of car mishaps, not catastrophic single events.
Passenger trains obviously have a potential damage limit based on occupant capacity and whoever an angry Assyrian deity chose to put close to the accident site. But freight trains are an entirely different matter. Technology has made some huge improvements in the number of people required to operate a train, but Human operators aren't there to make the train go, they're their to make it stop and coordinate the activities of first responders in the event of an accident.
Stopping the train is a little bit of a bad joke, except at super low speeds. But the first responder thing is deadly serious. Freight train accidents always look like ground zero of an aerial bombing mission. Local police, fire and rescue personnel all have real trouble differentiating between the dangerous and the not dangerous stuff simply because it all looks fairly similar (fucked up and way out of place).
That's not a swipe at the first responders abilities, but they're just not trained in dealing with a mobile industrial accident. Water to extinguish a fire getting closer to a house, for example, may help with the flames, but kill everyone else in the area with chlorine gas released when the boxcar full of pool chemicals got wet from the fire hose. You can stop a lot of damage sometimes by opening the appropriate cars to prevent pressure buildup, but it helps a lot if the guys on the train are around to tell you how to do that. Examples, fringe and otherwise, abound.
At any rate, my point was that justifying train operators isn't difficult. Freight trains are as close as most of the general public will ever come to being in close proximity to truly dangerous 'stuff'. The train operators act as a, small, extra layer of protection in events that can be just stupidly precarious. It's one of those situations where value is quickly recognized when the investments haven't been made, but it's too late at that point. It's too late, plus when those jobs are restored the costs are always far greater than they were before you eliminated the jobs. The memory of a tragedy caused by budgeteers being cheapskates is about as large of a cost multiplier as you can get.
That should be "with her." And I think you just did. Given that Hell hath no fury like a woman scorned, you might want to start making apologies as quickly as possible. It might also be a good idea to alert your friends so they don't also get caught up in the mess you just created.
Actually, they're there to walk the train whenever a myriad of sensors goes off saying there might be a problem with the train.
In spite of my posting about Metro, I actually ride the MARC, which is a passenger service run on the same rail lines the freights in our area use. At least once a week we pass a freight train stopped on the opposite track for just such an inspection. I'm told usually it is something like a brake overheating. But whatever the cause the sensors can't confirm the problem and require the meatbags to confirm and/or patch the problem until the train can get to a station where a mechanic can work on it properly.
">The Docklands Light Railway?
Is it the Victoria line that has drivers only because computers can't strike?"
No. The DLR was built to allow evacuation by foot on footpaths. There's always a TFL employee on board, just not driving.
The model with the underground is that the driver is expected to be able to deal with most technical problems - usually by isolating faulty equipment. If they can't they're there to guide people off a train for a reasonably tricky walk down a tunnel with no footpath.
The models are different.
Although to be fair the DLR rarely reaches any speed worth worrying about since stations are barely 500m apart. There are 3 serving the Excel centre alone, and 4 in the square mile to spare the poor commuters having to walk all of 200metres in the morning.
First time I rode it I kept thinking "We're slowing down again? Back of the train's barely left the last station."
The western end is okay (at least 1km between stops), but if you're trying to get out East you're better on the Tube or going Overground by the time you've sat through the ExCel and London Airport stops and treat them as an express service compared to the "local DLR stopping service".
> First time I rode it I kept thinking "We're slowing down again? Back of the train's barely left the last station.
Here in Vancouver, the Skytrain is pretty reliable and there are no drivers.
The fact that the trains are fully automatic is hardly cause for conversation here since we have had them since 1986 when they were introduced as part of Expo '86.
We do have the occasional mishap due to equipment failure (mostly points/"switches") and some idiot failing on the tracks but by and large they're extremely reliably. People here whinge a bucketful when stuff goes wrong, but they've forgotten what it's like during the winter when drivers call in sick and trains are cancelled.
In all honesty, in this century, I cannot fathom any reason (other that cost) why any closed transport system is still driven by meatbags. Given Don's point above about first responders in the case of freight trains, people just cannot match the reliability, safety and consistency of computers. They don't get tired and they don't get distracted. In all of the years since 1986, we have never had what anyone could call a tragedy here on the Skytrain. It's bound to happen given a long enough time, but if you accept that a lot of the tragedies that have happened on passenger trains in recent times have been tired drivers banging through stop/warning lights or speeding through restricted zones, the case for automated transport is hard to ague against.
Now automatic cars are a different story. I'd like to see them but the problem of safety is markedly different.
Couldn't agree more. I suspect the reason we still have train drivers is because the transition period would be very difficult. Look at the tube drivers in London, as soon as they get a whiff of something that isn't completely in their favour they go out on strike and the city grinds to a halt. Can you imagine what would happen if you put in a driverless train? The only way to handle the transition would be to swap out every train in one go which would cost a fortune and probably go wrong so we just keep paying for drivers. Eventually the technology will become cheap enough and good enough that drivers will seem absurd and then it'll happen.
My vision of the future (TM) is that local commutes will be done by these autonomous vehicles, which will be electric motors, and should they need to step outside of their modest charge circle then they'd pick a route to the nearest train station, hop aboard a specially-designed carriage where they can charge, and then pop out at the nearest station to the destination.
Or they just kick the meatbags out at the station and make them travel with the hoi polloi.
Justifying train drivers (in the passenger context) is already difficult, especially in the US.
For example, I live in the DC metropolitan area which of course named their German engineered light rail Metro. In June of 2009 there was a collision because of a failure of in the automatic driving system (http://en.wikipedia.org/wiki/Incidents_on_the_Washington_Metro scroll for image). I say failure of the system in as much as sensors on the tracks weren't relaying information properly (Metro knew about potential issues, IIRC they thought they had just finished installing a fix, but the problem continued). Since that time they've had all the trains on the system running under manual control. This has resulted in jerkier stops as well as significantly increased wear on the breaking mechanisms. I'd rather the trains were back under the automatic controls and Metro had fixed their damn sensor system properly. No, I'm not discounting the 9 dead and 70 injured (I have a friend who was injured in the incident but wasn't part of that 70 person count). Just over 5months after they switched to manual control they had another collision incident. Only for that accident they got lucky. It was an out of service train at the rail lot and moving at slow speeds. I don't think a person at the wheel would have helped in the June incident. He would have been relying on the same signal that failed to be transmitted to automated system.
I've casually followed the fallout from this incident and it is revealing of how disasters like this take place. It isn't an unlucky combination of unfortunate coincidences. Metro knew they had sensor problems for several years. They claim they are underfunded and can't make the needed repairs. Meanwhile their overtime logs show workers consistently logging 60+ hours a week, with some key positions logging over 100 hours a week on a consistent basis. Even if I don't want a French work week, I recognize that consistently logging 60 hours a week means you're getting sub-optimal performance from a worker. Logging more than 70 is insane and except in cases of dire emergency no one should ever work a single 100 hour week (if they do it should be followed by a week of mandatory leave for recuperation). Next up we have the passenger cars. I think they have about 4 varieties of cars (1000, 5000, 7000, 8000). The NTSB issued safety warnings for the 1000 over 5 years before the accident. The problem, exactly the kind of car separation you'll see in the wiki picture. To this day, those cars are still in the trains, but Metro has updated their SOP to use them "only in the middle, not on the ends." What we have is a long list of known problems that ultimately resulted in the deaths of 9 people and injury of 70 more.
Whether the car is driven by a program or a person doesn't matter half as much as not ignoring known problems. In fact, I expect the car being driven by the program will work out to be a lot more like a server in the data center: you might get a person to work in 90F and 90% humidity, but the servers are kept in air conditioned room. Likewise if the safety measures aren't enabled, the program won't drive.
I recall seeing an article somewhere, possibly El Reg but wouldn't swear to it, and can't be arsed doing a search, that said there is a large body of law relating to horse drawn carriages and drays that can provide a good starting point for sorting out liability where a semi autonomous motive force is under the loose supervision but not necessarily direct control of a human.
so the civil liabilities may have already been sorted out over a hundred years ago. Think of the milkman's horse going along the street while the milko runs from the float to each house.
Maybe I'm showing my age even thinking of this scenario. (Still had horse and milko in my melbourne suburb in the early 1960s.)
Vaguely dickensian icon.
You can't sue a horse or its creator. I suspect you've never needed 'third party' insurance to have one pull your milk float either.
You can sue the company that designed the driverless car, the company that programmed it, their insurers, and anyone else vaguely in the line of fire.
If you were an insurer, would you accept that the software was bug free and would not cause a serious crash?
The biggest problem I see with horse drawn carriages is the fact that even the most elementary convenience accessories have simply untenable resource demands. The speed isn't too much of an issue if you're an urban dweller. Just getting to 20KPH is a rare occurrence in lots of congested areas, so that speed probably exceeds the average anyway.
But I tell you what, 20KPH on a 12 lane highway in the summer is bullshit without A/C, and that's where everything falls apart. You've got to have a minimum of one high torque horse just to run the compressor, another to run the alternator, and a third horse with good general performance to handle peak demands like headlamps and TV's and safety systems like ABS and power steering. On top of all that, you've still got to have your propulsion horse.
By itself, having four horses per carriage isn't really a problem. The problem(s) are similar to those experienced in aircraft with multiple piston engines and super high performance tractors used in tractor pulls. All your engines have to be working in perfect harmony for optional performance.
But for the most efficient arrangement with a carriage that harmony is fairly impossible to reach. The breed specific attributes that make a given breed well suited for its role means you're going to have to use multiple breed types. Beyond the obvious gait length issue, you've got other challenges like cycle duration and total surface area of the drive unit contact surfaces and a stable full of other locomotion challenges.
Bringing all that together across multiple breeds just isn't going to be cost effective. But you've got to look for a solution or the carriage is going to travel like a car with steel belted radials on one end and bias-ply on the other (which is insanely unstable and extremely dangerous).
My off the cuff solution would be to have the propulsion horses from the same breed and as matched as possible. To solve the unbalanced energy distribution problem the key is to have multiple, accessory horses with narrowly defined roles (one for the radio, one for the TV, one for the radar on the SAM launcher, etc...).
The overall size of the accessory horses is key, the smallest possible that meet your energy needs with 20% overage. The size is so important because that's what will determine the overall length of the generator treadmill. The treadmill rollers are attached to alternators that provide the carriage with accessory energy distributed by a conventional power distribution system based on concepts proven by the automobile industry.
Because the accessory horses are on the treadmill and not in direct contract with the road surface they can function as an independent system without causing drivetrain imbalance. A giant treadmill protruding 25' - 30' from the front, or rear, of the carriage with several tons of horse on top is going to be challenges like tipping over on the treadmill end of the carriage as well as urban navigation, but those things are manageable.
The post mentioned the 1950's.
That was in the old days when people accepted that accidents did in fact 'just' happen, and it was just bad luck that they happened to get knocked down by a runaway horse, and anyway they should really have been paying more attention....
Then the lawyers got hold of the world......
60 seat bus with 58 lawyers on it goes over a cliff, whats the tragedy?
2 empty seats
The horse-carriage or dray is biologically governed to a maximum speed of about 20 km/hour. Would the auto-automobile be similary restricted?
I suppose the annual vehicle test could be extended to include a "driving test" on a rolling road with simulated traffic, pedestrians, weather, etc. Would these vehicles be rated and restricted to classifications of road conditions (snow, ice. fog, motorway, etc.), load and speed? Presumably instead of a driving licence, some sort of an operators licence would be required.
The road traffic conditions here are vastly different to those in the USA; our roads are generally more congested, the driving standards are variable at best, and frankly, I - and I suspect many others - don't trust the level of reliability on the highly complex electronic, mechanical, and computing solutions that would be required for the extremely active and reactive systems that would have to be 100% reliable fully automated travel on our roads.
In addition (and Suricou Raven, we were typing at the same time, looks like, you beat me to this just now!), there's the legislation and litigation angle to consider: There has to be a human to take responsibility for the movement of a vehicle in a court of law should something go irreversibly wrong; vehicle-related deaths, injuries, and such like, can generally be taken to have been caused by a Human, not a machine - indeed, the percentage of Road Traffic Collisions caused by purely mechanical failure are remarkably small these days; taking the human out of the loop means that should a computer or technological error creep in, human responsibility may not properly be apportionable; thus without a Human in the control loop, any prosecution resulting from a smart-system-related RTC may be immediately doomed to failure.
Don't get me wrong: I'm all for technological advances, but you have to consider that even unmanned aerial vehicles have humans in the control loop; if they require a human on the controls, then surely road-going vehicles require this as well?
What is needed for driverless cars is something akin to the Turing Test (except that it would not apply to the skill level of someone too young to drive) such as going round Hyde Park Corner in the rush hour from any direction and taking any exit without hitting another vehicle or forcing it into taking avoiding action.
You've clearly never driven in NYC!
Or, Goddess help us, New Jersey!
Oh, and UAVs are perfectly capable of operating entirely autonomously, and frequently do; you program a flight path, the drone executes it, traveling and loitering as programmed, and returns to the spot it took off from. I'm in the movie business, this is fairly common practice for our drones. They CAN also be flown under full manual control of course, but they don't HAVE to be.
This is based on a few assumptions, one that drivers here are worse than in the US and also that traffic here is more congested which is flatly not true. Road here may sometimes be less straight and more complicated, but that's about the only difference. The other problem is that you assume a human has to be blamed in a crash, if the car fails at doing what it is supposed to do then its manufacturer at fault, much as it is today if a car fails. If manufacturers want to make automated cars then that's a risk they take. If however it was an accident out of the expectations of the car and the said car passes all lawful tests, than thats just life, accidents happen and lessons will be learnt, there doesn't always have to be someone to punish or someone to blame, its a sick world where its become so much that way.
You mistake localized highway congestion here in the US with overall congestion. We have an immense highway and road system, most of which is radically underutilized. Anywhere there's heavy traffic you'll find an invisible line that marks the farthest extent of urban traffic. For example, from our DC office it takes me roughly the same amount of time to cover the 15 miles or so to escape the beltway as it does to cover the 40 miles from that point to my house.
In the UK congestion is more of a medium(ish) density thing but it's just everywhere. Granted, I haven't been to every last place in the UK, but I've seen a lot of it, and not just don't see great, empty stretches of road that don't have cars on them. The 'country' isn't. Country just means the buildings are smaller.
It's not really a fair comparison, but there's no doubt that congestion (where congestion means being able to journey from A to B at the speed limit without being encumbered by other cars) is far worse in the UK. In the US road congestion is in, and around, a defined number of places, but once you get clear of those areas congestion means something really awful has happened up ahead and the best you can hope for is the bodies and body parts are covered up before you drive by with your kids.
I - and I suspect many others - don't trust the level of reliability on the highly complex electronic, mechanical, and computing solutions that would be required for the extremely active and reactive systems that would have to be 100% reliable fully automated travel on our roads.
Have you seen how stupid people are? We let them drive.
There has to be a human to take responsibility for the movement of a vehicle in a court of law should something go irreversibly wrong;
And yet today, when a driver kills another road user, the criminal justice system (if it is wheeled out at all) usually just shifts uncomfortably in its chair and fails to convict or issues a small fine for a low level strict liability offence.
Sure, liability needs to be worked out by enshrining a set of essentially arbitrary conventions which will be liked by the winners and disliked by the losers, but how will that differ from what he have surrounding driving (or almost any other hazardous activity) today?
The bar for improving KSIs and also the non-lethal impact of vehicle use (noise, related infrastructure etc) is set pretty low
> Have you seen how stupid people are? We let them drive.
We also let people design hardware and write software that will be used to control autonomous vehicles. Could you safely say that out of all the code required someone won't introduce a few bugs here and there?
Indeed I could not and there will be errors, maybe egregious and high impact ones with fail-dangerous outcomes but UK road KSIs stand at around 65 people per day with fleshy drivers and the entry bar for driverless car programmers is considerably higher than for a driving licence. I see the potential for enormous public health/economic net benefits.
It isn't the bugs that concern me, but the pathways into these systems that have been created to allow remote access.
Remote access, I hear you say.
Yes, the means, which our over zealous enforcers will undoubtedly demand are present, to, well, better enforce.
The means to upgrade firmware en mass, the means to allow connectivity to the outside world, the means to make the passenger experience, that touch more enjoyable.
All these things add up to a number of attack vectors, that some hackers will be drooling over as we speak.
It isn't right that a world, that has experienced such a huge problem with viruses, rootkits and all should be allowing the same tech companies that left our PC's and phones open to attack, the opportunity to kill motorists in their millions.
Not once have I read about the security of these systems and quite frankly until it becomes the number one issue in every facet of development, I hope they never get off the test track.
While some may point at the success of drones, I would be quick to remind you, malevolent teenage hackers have yet to take one apart, through a fortuitous combination of cost and availability, which the autonomous car in its ubiquitous take up will not enjoy for very long.
> Could you safely say that out of all the code required someone won't introduce a few bugs here and there?
Modern airlines can and do practically fly themselves, including take off and landing.
It is possible to write software that is as near to perfect as possible. It just takes a lot of effort and great care.
Thing is though, you only really have to crack the problem once and replicate it millions of times. With meatbags, you have to train them for ages and you still end up with a neurotic, emotional, tired, stressed, aggressive and error prone driver.
Modern airlines can and do practically fly themselves, including take off and landing.
And they do this by operating in largely uncongested space using scheduled flight plans and whole teams of meatjobs in centres around the world to ensure they don't crash when they do get in crowded spaces.
It is possible to write software that is as near to perfect as possible. It just takes a lot of effort and great care.
In flight's case, by putting constraints on the variables in a way that is not possible with current road traffic, and using mechanical turks in the form of air traffic controllers. If there is a human in the box, it's not automation folks.
"Have you seen how stupid people are? We let them drive.
We also let people design hardware and write software that will be used to control autonomous vehicles. Could you safely say that out of all the code required someone won't introduce a few bugs here and there?"
Probably not the same people in all likelihood. I don't think safety critical programming jobs are handed out randomly to people and the testing regime is likely strenuous.
Do you every fly in a commercial airliner, or is your disdain for programmers reserved only for those involved with cars?
I don't have disdain for programmers, just for bad programmers. Having seen what ass kissing and nepotism can do to a well oiled programming team the quality of code is important.
Using aircraft as a comparison is about as dumb as Microsoft's inept car comparison. As mentioned above, or below, commercial airliners have whole teams of people and systems to make sure their journey through relatively clear spaces is trouble free. If I thought for a minute that there was a chance that some oik in his Boeing Saxo GTR could be trying to impress his girlfriend by pulling stupid stunts in close proximity to other craft I would never get on one.
quote: "In addition (and Suricou Raven, we were typing at the same time, looks like, you beat me to this just now!), there's the legislation and litigation angle to consider: There has to be a human to take responsibility for the movement of a vehicle in a court of law should something go irreversibly wrong; vehicle-related deaths, injuries, and such like, can generally be taken to have been caused by a Human, not a machine - indeed, the percentage of Road Traffic Collisions caused by purely mechanical failure are remarkably small these days; taking the human out of the loop means that should a computer or technological error creep in, human responsibility may not properly be apportionable; thus without a Human in the control loop, any prosecution resulting from a smart-system-related RTC may be immediately doomed to failure."
In the UK and the US, a Corporate entity is recognised as a legal person. Thus you can just as easily take the Corporate entity which created the autonomous driving software to court regarding the traffic incident, as you could the driver of a manual vehicle, should the cause of the incident be attributed to the decisions made by the autonomous driving software rather than, say a human driver's bad decisions.
Of course the last thing Google (or BMW) want is to be held responsible for the behaviour of their own product. Thus all this noise about requiring changes to existing legislation, which will IMO either be dropping liability onto the lap of the vehicle owner, or asking for public funds to be set aside to cover payouts for incidents. Meaning that either we pay for it out of personal insurance policies, or we pay for it out of tax.
Out of interest, if a gearbox failure (or other mechanical issue unrelated to driver decisions) happened on a vehicle causing loss of control and subsequent road traffic incident, who is currently considered liable? Both for a vehicle in-warranty with a full service history, and for a vehicle out of warranty being self-serviced?
should the cause of the incident be attributed to the decisions made by the autonomous driving software rather than, say a human driver's bad decisions.
Interesting. Currently, it is almost always a human driver's bad decisions that cause an accident, however we do not assign blame like that - you can make all the wrong choices, cause a serious accident and not have any blame assigned to you - quite rare though, Id have thought.
Say there was a death. The CPS have to determine if they can get a prosecution for manslaughter, for dangerous driving or for driving without due care and attention (not full list, some names probably wrong). Sometimes, it is just bad luck.
Now, this is hard enough to do even when it is a human driving and can tell you what they did. How do you determine if a computer program was driving dangerously? If the radar detector is dirty and your autocar runs in to the back of someone, are you liable, as you didn't clean the sensor, or is the software liable, because it didn't detect the sensor was faulty? If it's you, are you "without due care and attention", or is the software "driving dangerously".
"Out of interest, if a gearbox failure (or other mechanical issue unrelated to driver decisions) happened on a vehicle causing loss of control and subsequent road traffic incident, who is currently considered liable? Both for a vehicle in-warranty with a full service history, and for a vehicle out of warranty being self-serviced?"
It's always the insurer of the vehicle that caused the incident.
That insurer is then free to attempt to reclaim its loss if it feels that someone else is culpable. That could be the manufacturer but it's unlikely to be the owner if the car has a valid MOT and they didn't knowingly drive the vehicle on the road with a fault. If the car was road legal at the time of the accident, the owner isn't liable for mechanical failure, even if they have never had the car serviced.
This post has been deleted by its author
DoT's complete inaction to amend the out dated regulations to allow this vehicle (with a driver no less) to be used in public has killed it off for the UK.
I expect driverless cars will get the same amount of inaction. They may dust off the law requiring somebody to walk in front of it with a flag as that would require the least amount of effort while still giving the impression that the department actualy do's something.
If you're referring to the Segway Gyro Stabilized Mobile Target Carriage then the cost killed the large scale adoption of those things, not legislation, or lack thereof.
There are lots of places here in the US where it's legal to ride them on the road, but they're still rare. They are used at some government facilities, airports and by the worst sort of students at some universities, but for actual 'normal' people transportation, not very much.
A sightseeing place here in DC, 'Segs In The City' operates, I believe, the largest Segway fleet in the US. They do decent business, but if you talk to riders afterward any ideas of buying one die when they find out you can buy a decent used car for the same money. Governments fuck up lots of things, but the Segway killed the Segway, not the government.
Not sure what the EU will say to that, but Germany as one of the more influential members will probably have no interest for now. Cars are much more there than a means of travelling from point A to point B. Germans like to DRIVE cars, not be a passive passenger. And the automobile lobby is strong there, very strong.
It's not that clear-cut.
When EU legislation is enacted, it does not immediately become law in the member states. Each member state is supposed to enact local legislation that covers the EU law, but there are reasons why this is not done. A member state may, under certain pre-negotiated circumstances within EU Treaties veto a law (and thus not be bound by it), or can state a derogation, or delay local legislation almost indefinitely, or can just ignore it.
Of course, if a country just ignores an EU directive, then the country (in reality, the incumbent government) can be taken to one of the various EU courts, but that is a long and expensive process (the costs of which will normally be borne by the complainant), and even at the end of it, all that is likely to happen is a slap on the wrist and a fine (which can it self be ignored with relative impunity). The ultimate sanction of expelling a country from the EU is extremely unlikely.
This is, of course, a very simplistic view of a very complex process, but one example of where this has hit the news moderately recently is the controversy over prisoner voting rights in the UK last year.
Is there actually anything in current UK law that says that a car must have a human driver? Has anyone bothered to *check*? Much of the law in this area is so old that the possibility wouldn't ever have occurred to anyone, and there may well be nothing in law either allowing or forbidding this. And if the law is silent... qui tacet consentire videtur...
I'm sure it's an offence to operate a vehicle without a license, but try taking a machine to court for that offence. They might try a case against the occupant... but which occupant? The one who happened to say "OK Car, take me to Glasgow"? How to prove who said it, if more than one occupant?
And if the occupant has a license anyway, what charge could be offered? Can't be driving without a license. Failing to maintain control? "I maintained perfect control! I had a computer do it for me! If you think it didn't do a good job of controlling the car and an offence occurred, prove it!"
This could get very interesting :-)
If it's going to be sold widely, yes. Are you aware of any provision in the construction and use regs that would specifically prohibit a car controlled by a computer?
If it's not - if it's a small test fleet - there are special provisions for this. SVA etc.
And if it's imported from overseas, as is very likely, then the approval there is good for temporary use on UK roads. I've moved cars between countries on several occasions. Currently driving a UK-registered car on UK plates & MOT in New Zealand as a temporary import. The regs that apply are those in the country of registration, not the country where it's being temporarily driven.
Microsoft branded vehicles will actually be comfortable and look fine, but they will only take you to three places, one of which is Slough, no matter where you ask them to go. Also they will occasionally download an update and just stop wherever they happen to be for ten minutes. When they crash, they will turn blue.
erm, you do realise one of the main driving forces for automated cars is to increase safety?
Most crashes are caused by bad driving, either causing the crash in the first place (driving too fast in bad weather etc.), being unable to react fast enough when something happens in front of them (driving too close to someone while not paying attention etc), not being skilled enough to regain control after hitting oil on the road etc. etc. Most of these types of accidents and crashes can be avoided with automated cars (and busses, and HGVs etc etc.).
There would also likely be a reduction in overcrowding, as more vehicles are automated, they can start driving closer together, while still remaining safe, even driving in road trains on major roads, thus reducing the space the vehicles need to take up on the roads themselves.
Making the manufacturer liable for system errors, which seems reasonable, would imply:
- black box plus cameras to record all aspects of a journey, incident, etc
- obligatory servicing at manufacturer's approved facility
- sealed system, no user access to any part of vehicle except passenger compartment
- only allowed to connect approved devices to vehicle
plus anything else the manufacturer or their insurers can think of to limit their risk and/or increase the number of reasons for them to refuse liability.
I like technology as much as the next person - but I really don’t like the idea of self driving cars. I have no doubt that they’ll make the roads safer and so forth - but what happens in twenty years time or so when the majority of cars on the road have the required intelligence?
What happens is an orgy of scrapping as the old, dumb, cars that have served us faithfully until now are banned and forced from the roads. And what of those of us who have classic cars? Now as viable as a horse and cart for long distance travel, will exemption be made to keep a little character on our roads?
Yes, yes. I know. I’m being self interested again. But I really would prefer it if my fun wasn’t spoiled in the name of ‘elfen safety and the preferences of a bunch of politicos who don’t enjoy driving anyway.
They'll porbably go the same way as other outdated forms of transport - provide a mild annoyance for normal road users stuck behind one on the very rare occasion that they're on their way to a vintage rally, or be used in sporting events. Like horses, traction engines and classic cars are today.
You're driving up a country lane. 50m ahead you see a car coming the other way. One of you has to pull into the side to allow the other to pass. Usually it's obvious who has to do it and it can be resolved easily by humans.
How is a self drive car going to do that? How does it know that it must pull over? How does it signal its intention to the other driver? How does understand when other driver signals back? What happens if the car has to reverse to pull in?
Or is it going to be completely brain dead and drive right up to the other car and then halt because it is an obstacle? How is it going to extricate from that?
And that's just one very simple example of a problem that happens every single day that is almost intractible for a computer. There are thousands more like it - faulty traffic lights, police directing traffic (corollary: telling a police officer from a loony directing traffic), traffic calming islands, diversions, oil spills, lorries unloading goods, floods, potholes, buses stopping to change drivers, broken down cars, carparks / long tunnels with no gps, a lorry on fire, cones & roadworks, emergency vehicles w sirens etc.
I think the tech would be very useful for motorways and as an advanced safety system but completely self drive vehicles on public roads is so far off I don't see happening any time soon. There will have to be a conscious, sober driver behind the wheel ready to take over and extricate the car from a situation it can't solve itself.
Hear hear! But drivers are increasingly reliant on technology and rather dopey about solving these kinds of problems themselves. I know someone who insists that she's a good driver because she can apply her make up, talk on the phone and do a million other things whilst driving - and her car remains straight and true and stable the whole time. Of course, it all went to a can of worms when I turned off the dynamic stability system (and, with it, traction control, lane departure and so forth). A salutary lesson (and yes, the road was empty in both directions when I pulled that stunt).
The more electronics get thrown at cars, the more incompetent the average driver will get.
All good points.
with my crystal ball, I see these things coming in two flavours.
For the car without any controls would be of use in a 'closed' environment, say a city centre, where there is no interaction with any other type of traffic.
The more open type/mixed traffic conditions of auotmated self-driving car will still have all the controls to allow manual driving which can be used for when the computer just can't work what to do.
There is still a lot of work to be done before any of this is even close to a commerical product that would be allowed on the roads in any meaningful volume.
Didnt they have a "manned" test in the states for a year or two. Where it was an ordinary car with ordinary controls, just driven by a computer. There was a driver that could step in at any time.
Would it not be prudent to conduct the same tests in what is a very different driving environment in the UK before sending in the drones?
Like many others i love the idea of being able to head out to a suitable countryside watering hole and then being able to summon my robot car with my smartphone, collapsing into the back for a snooze whole it chauffeurs me home.
The tech seems robust enough to follow a simple route on wide (American!), clearly marked and signed roads (i.e. basically not much different to an automated railway)....but what about the infinite number of unforeseen things that can happen on the road? How do you program the car to make judgement calls in marginal situations?
Can the car's computer:
- Detect an ambulance or police car approaching from behind at speed with sirens going, pull over onto a kerb, bus lane, verge etc to allow it to pass and rejoin the flow of traffic, even when there is not an obvious space to pull into?
- Anticipate that a child is going to run out into its path because it sees a football roll across the road in the distance?
- Tell the difference between traffic lights that are out of order/switched off/have a bulb out/are working but hard to see due to those grilles they fit or low sun?
- Push out onto a gridlocked city roundabout or junction where, if the strict letter of the highway code was followed, progress would be halted for hours at a time?
- Manoeuvre through cones and a contra-flow where a carriageway is being dug-up, even though the GPS will tell the car that its travelling on the wrong side of the road? (and the guys digging forgot to notify Google or whatever)
- Tell the difference between a pedestrian in the road and a car-jacker deliberately stopping the vehicle by blocking its path, and take emergency action to get away?
And so on....
Sure, human drivers get these things wrong as well, but I don't see how these scenarios can be programmed for at all without AI to do the thinking and reasoning.....in reality it would mean that human attention and intervention always needs to possible, negating much of the reason to do 'driverless' cars in the first place.....
The very fact you have thought of these things means that engineers will most likely have already thought of them and are working on these things. Things like ambulances/fire engines/police cars are all cases where it will be likely that cars will be communicating via a mesh and advertising their immediate intentions to each other. These things are not intractable.
Useful stuff that we probably wont get to do
- Getting drunk and getting the car to take you home from the pub
- Getting a few minutes extra sleep on the way to work, possibly after the previous night out
- Using you mobile while google drives you
- Police trying to pull the car over while google drives
- Taking your car on holiday from the UK to Europe
Changing Left turn and Right turn commands ? Switching accelerator and brake commands ? Turning a Goggle car into the ultimate remote control car ? I expect it will take hackers about 10 minutes to find the first flaw in the Google car's OS. (Black hat icon if El Register had one).
All I want is a windshield display that turns night into broad daylight, eliminating the need for headlights altogether. I may want a self driving car, but I'm not "dying" for one.
Jeez half you guys sound like the to do that went up in Victorian times about trains traveling over ten miles an hour.
Automated vehicles will save lives by driving more safely than humans, why because they will be programmed by the manufacturers to take no risks at all unlike humans and if there is an issue they will simply shut down and refuse to move.Any warranty as suggested will lean heavily on the owner to have services and part replacements done exactly at specified periods unlike most modern vehicles which miss servicing or don't replace parts i.e. tyres and brake pads as required.
All travel data will be recorded I suspect specifically so the manufacturer can cover his ass but this will be available in the event of an incident involving any vehicles manned or not to apportion the failure.
The right to own and drive a vehicle is just clogging up the roads to congestion and the "communal vehicle" may relieve at least some of this.
We all know how good satellite navigation data and software is - what happens when the vehicle has incorrect data for the speed limit of the current road and gets clocked over the limit?
I presume that the vehicle cameras would see, and read the speed limit signs (as they should all be the same size, shape, font, etc) and adjust accordingly. For the flashing temporary signs, well, one more circuit and it'll broadcast 'r=M4, d=w, s=40' and the car's computer will know what to do: either slow to 40 because its currently westbound on the M4; or ignore it because its not.
"...without requiring someone in the so-called driver’s seat"
So what happens if a so-called hacker manages to gain control of the so-called wheel? Current commercial airliners have the capability to self-navigate between destinations, including take off & landing, but I still want a pilot behind the controls, thanks. Will there be some sort of manual override facility to account for unforeseen technical glitches or emergencies?
Also (for me, at least) where's the fun? Christ, even in city traffic I have some autonomy & I do enjoy driving.
The article implies that development won't happen in the UK at Oxford or elsewhere if the law isn't changed like California's is. That's a load of bull. The cars are still in an early stage and won't be for sale for many years. What's wrong with having someone in the driver's seat during development? Presumably they are nowhere near a stage where the cars will go off for test drives with no passengers at all, so there's no reason that passenger can't be in the driver's seat and have a steering wheel and brakes available to him.
Google developing the car without the driver's seat was a publicity stunt, it has no purpose in the world right now when they are so far away from being able to offer this as a product. It isn't like Google Glass where they can release something for what is essentially a public beta. It will take law changes dealing with liability and such. Until that time, there's zero benefit to having a law that allows cars with no one in the driver's seat, or even to build such cars.
Just require a man to walk in front of the passanger transport device ringing a bell to warn other road users that they are comming, simaler laws are still on the books in many commonwealth countrys,
Just not enforced, the whole meat bag thing will just not be enforced.
As for who is responcable, it will always be APPLES FULT just as it is now.
A bit of research by Google and there's an immediate reaction to accommodate them. Bicycles have been around a lot longer and given their numbers on London and other roads, the Highway Code has been in need of revision to accommodate them properly rather than tolerate them for a long time.Some examples (with due regard for pedestrians:
1. Left turn on red lights
2. Go on pedestrian lights if there are no pedestrians
3. Right of way over powered vehicles at all junctions
The driverless DS was tested in the 1960s on a track at Crowthorne. They've had an awfully long time to think about legislation
I have been thinking about self-driving cars differently. I was thinking we'd never actually get to own one. I pictured that car manufacturers would make the vechicles, and licence them to fleets (e.g. taxi firms), who would provide the insurance, cleaning, servicing and energy usage etc. (assuming electric here). Whenever we need to travel somewhere, we get out our smartphone, and request one. It'll then trundle up to us (using the GPS in our phone) and we get in, tell it where we want to go, we then pay for the transport using some electronic payment on the phone.
All this means that no-one owns the vechicles except them manufacturer, who is the ultimate receiver of our money. this then pays for updates, improvements and provides a stable income for them. We'd never need to own one, but the'd always be one ready for us. Basically, everyone rents. Old cars can be recycled better.
The other side is we move from oil gradually to electric.
To me that sounds like a good solution. Only paying for what you need.
There may be a few existing companies that disagree though (oil, taxis, garages etc..)
What do they do when someone is standing in the middle of the road waving their arms?
Does it treat them like a hazard and drive round them. They could be there to warn that the way ahead is dangerous (bridge about to collapse).
Does it treat them as someone to stop at, which case it could be a member of a gang, trying to stop your car so the rest can pounce on it, force it open and rob you.
And I spotted a nice little article about networked cars:
"Human operators aren't there to make the train go, they're their to make it stop and coordinate the activities of first responders in the event of an accident."
In a major accident, they're probably impersonating a large deposit of strawberry jam already. That's why they're all clearly labelled with Hazchem markings or similar. For example, petrol's '3YE' tells any firefighters responding that they must use foam or dry extinguisher, it's an explosive/violently combustible chemical, use breathing apparatus and evacuate the area. No need to interrogate squished meatbags.
(I remember being quite disturbed after a Virgin train derailed in the UK, then Richard Branson popped up on TV talking about how the driver "tried to steer the train to safety". If that's his level of railway knowledge, no wonder they lost the franchise...)
Biting the hand that feeds IT © 1998–2022