That's only too true
Just put a random driver used to automatic in a stick car, and watch the horror unfurl. And that's only a rather minor difference...
Tech and automotive firms are pushing driverless car technology on society, rather than there being a big demand for it, in the opinion of the Transport Research Laboratory's boffins. In a whitepaper published today TRL's Fellows opined that not only is there a "strong technology push for autonomous vehicles" but also that " …
Or the manual driver in an automatic and watch him make an emergency stop every time he tried to change gear!
As a manual driver who's driven an automatic (hire car) I resent this.
I only did that once before I discovered that the easiest way of driving an automatic (at least for a manual trained driver) is using only one foot for both controls. ;)
It's not that different otherwise. Apart from the fuel consumption.
What is really needed is a paradigm shift to 100% driverless technology.
When it is sharing the road with "real" drivers, it will never be good enough, because humans are poor drivers and don't drive "properly" (just look at the millions of hours of YouTube video of drivers wandering over the road, storming through red lights and generally being dicks.
If you have an automated car, they will work out, that if they cut in front of it, the car will avoid an accident and let the rodent in. The same for unsafe overtakes, the car in the wrong will be okay, because the AI car will avoid a crash and let them through (if possible).
Either you need to make the AI as bloody-minded as real drivers or you need to eliminate real drivers from the equation.
HGV drivers are already at a disadvantage with modern systems that automatically maintain the legally required distance to the vehicle in front, cars cut in front of the HGV and it automatically slows down to maintain the safe gap.
Autonomous vehicles will lead to autonomous complaints: if the lorry has to brake too hard when a vehicle cuts in then the video and sensor log of the incident can be automatically sent to the police.
If this were to take off then, within about 5 minutes, the police will be inundated and so will press for those prosecutions to be handled by another agency.
What will happen will be the equivalent of speed camera ticket processing: the lorry manufacturer's video and sensor log will be 'trusted' and tickets issued automatically. You can challenge them in court of course, but then you have to explain why you cut in and caused a vehicle to brake sharply.
..... and if its all done electronically by the time you have driven to work you will have clocked up a pile of automatic fines and penalty points.....
Reality is though that - like correctly placed speed cameras, if you obey the road rules then it shouldnt make much (if any) difference.
If I cut someone up in my driving then I deserve the slap/fine. Driving into work this morning I saw more than one car swerving across 3 lanes (without indicating) and back again multiple times (in moderate/heavy traffic) in a desperate move to get in front of other cars. Anything that helps reduce accidents caused by idiots like this is arguably a good move.
On my drive home tonight I can guarantee that I will meet:
Cars driving up the motorway (freeway) in the dark with no lights
Cars changing lane without indicating
Cars (and trucks) hogging the middle or "fast" lane
If I am lucky then today there will be no accident holding things up - but it is a regular occurrence and one that seems to spawn child accidents (from rapid braking/rubber necking/speeding).
I think I am just fed-up with how stupid we are when many of us get behind the wheel (and yes I include me in that at times!)....
I think I am tired - its been a long week, I've got no beer and its time to go fight the idiots on the motorway......
The driver who overtakes is supposed to only pull back into lane when they are clear of the stopping area + a safety margin of the vehicle(s) they have just overtaken. This is something I do. Overtaking a car, I will have to be at least 2.5 seconds in front of it before I pull back into lane. Overtaking a lorry, it's at least 50m before you can pull into road space which isn't within their braking distance, which can amount to an additional 2.5 seconds or more. When you see how big that distance is actually on the road, it looks ridiculous, but it isn't. For lead-foot Joe in the Audi or the Merc behind you, that 5-6 second gap is an eternity and they so often nip through on the inside, crossing into the crush zone to do so.
So that's why an AI system trying to maintain that safe space in front of a LGV/HGV will need to be pretty smart to avoid bringing the whole motorway to a halt trying to compensate for the nano-cephalic death wish idiots we have on our roads.
@TRT and that is the problem, drivers like you or me leave plenty of space and don't cause other drivers inconvenience, unless there is some sort of emergency.
But lead-foot Joe is unfortunately an all too common problem on our roads today. Or the idiots who overtake an HGV 50M before an exit, then cut rights across the front of the HGV, whilst braking hard, so that they still make their exit...
You need to get those idiots off the road, before such autonomous system can work effectively.
>@TRT and that is the problem, drivers like you or me leave plenty of space and don't cause other drivers inconvenience
The trouble is circa 5am on the M1 southbound, south of J15, typically at least one lane and sometimes two lanes are full of lorries, so joining the motorway can involve jumping into the 'crash zone' and either sitting among the lorries doing 50~60 mph or taking further risks and sliding into the lorry free centre lane where the other car drivers are doing 80mph. Naturally, you have to repeat this when you come to your exit...
The poster obviouly has never been tailgated by an HGV. That is a really scary thing.
As for them being speed limited... (Are you having a larf...). one overtook me on the M23 earlier today. I was doing 68mph according to my GPS. The HGV came up upon a middle lane hog. The HGV proceeded to tailgate the driver in front. there was less than 5ft between them. I was on a Motorcycle and being tailgating by any vehicle is just nasty.
@JimC if it was a fully automated system, the overtaking vehicle would drive far enough in front of the vehicle being overtaken in order to not have it brake sharply and thus causing unnecessary risk to the overtaken vehicle and those behind it or it would negotiate to cut back in and have the overtaken vehicle slowly reduce speed to make a gap for it to pull back in.
The point is, the AI would play an unthinking jerk and just cut up the HGV.
"What is really needed is a paradigm shift to 100% driverless technology."
Why? To make driverless technology work? Aren't you begging the question posed right at the start of the article?
"Tech and automotive firms are pushing driverless car technology on society, rather than there being a big demand for it"
humans are poor drivers and don't drive "properly"
Order of magnitude calculation. There are about 3,000 road fatalities per year in the UK. There are, I think, about 30,000,000 vehicles. That comes down to 1 fatality per 10,000 vehicle years. Multiply the number of vehicle years by the annual average mileage to get the rate in terms of miles. Apparently the average for cars is 7,900, vans over 12,000 and presumably HGVs higher still. Let's say 10,000 for an order of magnitude estimate, so one fatality per 100,000,000 miles.
At what date do the autonomous vehicle manufacturers reckon on being able to beat that?
one fatality per 100,000,000 miles. At what date do the autonomous vehicle manufacturers reckon on being able to beat that?
Immediately, assuming that:-
1) Two autonomous vehicles crash into each other, rather than one manned and one autonomous vehicle crashing.
2) nobody is in the autonomous vehicles. ;)
Fatalities aren't the biggest issue. What about people jumping read lights? Overtaking around blind bends? Driving the wrong way up a street? Changing langes at the last possible second, because they suddenly realise they need to take a different route? Staying in the overtaking lane until the last possible second, cutting across 2 or more lanes of traffice whilst braking hard to make an exit?
All of those are poor driving habits that you see almost every day out on the road. They don't lead to fatalities, but they are still poor drivers.
I used to drive around 140,000KM a year for business and pleasure and have probably a crash average of over 400,000KM, but I still wouldn't class myself as a good driver - considerate? Mostly. Safe? Generally. Don't make mistakes or get distracted? No, I am human and now and then I make mistakes or get distracted by things going on around me. Usually they are enough of a shock, that I am more attentive for the next few thousand kilometres.
Just watch an hour of crash videos on YouTube and tell me that human drivers are good drivers. Don't worry, you'll find an hours worth easy, there must be millions of hours of human stupidity on the road to chose from on YouTube.
Does that mean all drivers are stupid? No. Does it mean that you will have a crash the next time you step into a car? I hope not. But it shows where the problem is, we, as a society, seem to see a driving license as a right, not a privilege that has to be earned. The driving tests are often ridiculously easy and people who shouldn't be on the road get a license.
As long as these people are still on the road, autonomous cars are never going to be effective and the development will take much longer than "necessary". If autonomous vehicles only had to cope with other autonomous vehicles, they would be much simpler to develop, their software would be much less complex and there would probably be a large reduction in accidents. The vehicles would negotiate with each other at junctions to maintain an optimal traffic flow. There would be no egos getting in the way, no impatient drivers late for meetings.
If you mix autonomous and real drivers, the autonomous vehicles have to be several orders of magnitude more complex, because they don't just have to deal with other autonomous vehicles which cooperate with each other, they need to deal with normal drivers, who obey the rules, which isn't too bad, but they make mistakes now and then, and they have to deal with the pig headed, stupid, aggressive or downright crazy drivers who cause a lot of the accidents.
I am not saying that we have to go 100% autonomous or that we need to do it now or next year. I am just saying autonomous vehicles only make real sense when they are 100% of the traffic.
"But it shows where the problem is, we, as a society, seem to see a driving license as a right, not a privilege that has to be earned. The driving tests are often ridiculously easy and people who shouldn't be on the road get a license."
Because no one wants to be the one that condemns a person to starve because the ONLY job available to them is 50 miles away (and you're 75 miles away from any co-workers) along a route not serviced by public transportation for a radius of 10 miles, and cabs charge so much to go there you'd end up LOSING money. IOW, it's the car or bust. See it all the time when I go to court to see suspended license cases. Almost always, an exception has to be made for work driving or the person becomes unemployable.
As for leaving two seconds' worth of gap, the moment you leave a gap more than a car length, someone else WILL pull into it, guaranteed. THAT's why no one really does it anymore.
Tech and automotive firms are pushing driverless car technology on society, rather than there being a big demand for it, in the opinion of the Transport Research Laboratory's boffins.
I would agree with this.
In particular, it only makes commercial sense for companies if they can make a saving on having to employ qualified human drivers, and this is clearly not going to be the case for some time - if ever.
And of course, in the case of delivery vans, unless there is an able bodied person to receive the delivery, you still have to have a human on the vehicle to carry the groceries up to Granny in her third floor flat...
Tech and automotive firms are pushing driverless car technology on society, rather than there being a big demand for it, in the opinion of the Transport Research Laboratory's boffins.
I disagree with this, I would love to call up a driverless car whenever I need one, and so would a lot of other people.
So people not having to drive will reduce their driving skills? Shocker.
I'm sure tractor drivers have lost the key skills needed for riding a horse too.
The whole point of automated cars is that the driver doesn't need those skills, so what does it matter if they lose them?
I think the perceived problem is there is a valley to cross where cars are not fully automated and need a 'competent' driver aboard until we get to fully automated, autonomous vehicles (which could take 'some time'). But during this time many drivers will never actually drive and so, if called into action, will not be equipped to react correctly. This will be particularly acute for those that have just passed their tests but never actually accumulate the experience or driving 'for real'.
As I mentioned on another thread where such a 'level 3' vehicle had a camera to stop you looking away from the road I predict most journeys will last less than fifteen minutes as the minders repeatedly fall asleep staring at a road with no involvement in proceedings.
Surely that's just a regulatory issue? Don't let such a car get sold without a fully automated system in place.
No half and half. I'm pretty sure the various car manufacturers working on automation haven't indicated a desire to release such a half solution anyway.
Good point. I can't imagine any worse driving experience than having no control, but having some camera watching what I do and beeping annoyingly when I'm not "paying attention".
Even if I did obey the camera, and even if I was an excellent or perfect driver - how good would I (or anyone) be at suddenly assessing a situation and taking responsibility?
I wonder how many more epicycles will have to be added to the driverless-car project to make it look faintly desirable, viable or beneficial. As the first line in the article says, this is clearly a case of technology trying to drive society, rather than the other way round.
"driver doesn't need those skills, so what does it matter if they lose them?"
Because for all the AI hype, there are still worrying gaps in current capability where the car will shit on the driver by saying "Ooops, can't handle this - you take over NOW!!!" with possibly seconds till impact.
When you see all of the news/discussions saying automated cars need good connectivity, need accurate GPS/mapping, need road junctions/signs redesigned, and where insurers have got a cause proposed that implies you are only covered where the car is using automated driving for "appropriate conditions", it seems we still have some way to go.
"I'm sure tractor drivers have lost the key skills needed for riding a horse too."
The tractor driver doesn't find him or herself suddenly in control of a horse with about a second to react to avoid an accident. A human driver could find themselves in the equivalent situation unless the autonomous car is so good that this is never ever needed.
By the way I referred to the tractor driver as him or her because my neighbouring farming couple both drive tractors as required and she also rides a horse (no, not at the same time) so your certainty is a little over-rated.
I'm pretty sure the various car manufacturers working on automation haven't indicated a desire to release such a half solution anyway.
Despite the hype, I don't think any of the current AI car manufacturers are prepared to say that their systems are capable of being completely autonomous, and they all reserve the right to push the problem back to a meatbag when the AI panics.
"the guy down the hole who suddenly experiences the pleasure of an AI van landing on his head"
Back in the '50s my cousin was driving in a thick fog when she suffered a severe jolt. She stopped to look and found she'd crossed an unmarked narrow trench in the road. Then a head popped up...
It might be coincidence that a few days later a track rod failed.
In 30 years people won't own cars. There's no such need, how many hours per week do you really use it?
Unmanned vehicles are the shortest path to "car as a service". You need a ride, you make a call, it's there for you in 3 minutes. And after your ride is over you don't care for it.
I once sat down with friends and we calculated the total ownership cost of car. Everything included: oil changes, insurance, gas, spare tires, etc. And then how much one mile really costs you. Enjoy your math!
Have you ever ridden in a truly nasty cab? Now imagine that experience without being mitigated by the presence of another human being. The Internet has shown that people are less inhibited in their duchebaggery when there is not a person around to immediately judge them and they feel anonymous.
Much like people will spout vitriol in a forum or email (and even on social media, which has their name attached) that they would never consider saying in person, people will discard food/drinks/garbage in the floorboard of an autocab, and you will have people who take advantage of the "privacy" and leave behind traces of a more bodily type.
And after your ride is over you don't care for it. - This is actually the root of the problem; after your ride any mess in the autocab is "Somebody Else's Problem".
@KorndogDev & @Swarthy
I agree with both of you. I love the idea of getting the car I need (small economical for the commute, People Carrier for the family holiday) when I need it rather than having the drive full of 'best compromise I can afford' lying around idle most of the time.
But, in parallel with that, anonymous people are scum and that will need some social engineering and technical measures to resolve.
As always people are the real problem - nuke from orbit?
It seems rather quaint to imagine that anyone will be doing anything anonymously, much less hiring a car. There are plenty of examples today from rentals to car clubs of cars being used on-demand, and in my experience there is not a whole lot of abuse of these vehicles - probably because the users are anything but anonymous!
> Much like people will spout vitriol in a forum or email (and even on social media, which has their name attached) that they would never consider saying in person, people will discard food/drinks/garbage in the floorboard of an autocab, and you will have people who take advantage of the "privacy" and leave behind traces of a more bodily type.
After each journey the cab will take itself off to a automatic car wash before picking up the next passenger. And if the outside of the cab is washed as well, then so much the better.
> Have you ever ridden in a truly nasty cab? Now imagine that experience without being mitigated by the presence of another human being. <
Perhaps the self-cleaning public loo will provide the model; after each use, the auto-cab goes to the nearest cleansing station for a thorough hose-down and decontamination inside and out. Users will soon get used to the smell of disinfectant and the need to don a plastic mac before getting into one.
That's all well and good until I need a car to get to work. And I'm expected to be at work at the same time as everyone else. So either I have a car, or a car is available for me.
Sure, sure - you can use peak pricing to let my parents grab a car an hour later much cheaper, but that's still a shitload of automated cars that have to be provided by <someone>. So they all need paid for.
How many cars have only one occupant on the journey to work. With a bit of organisation car sharing works.
I was in a work based car pool, organised on a shared spread sheet. Imagine what could be done with AI. It was great not only do you save money, but the journey becomes a social occasion.
"There's no such need, how many hours per week do you really use it?"
Not many but if you use it to commute you need it about the same time as everyone else needs theirs. Now go back and do your maths calculating how many cars would be needed to meet your 3 minute criterion at peak times, how much of the day they'd spend idle and what the consequent per mile cost would be.
But even at peak times there are still plenty of cars parked on the road, in drives, in garages so, overall, you will still need fewer cars to meet the actual need than we have now - and they can all be sent off to park out of sight when not in use so streets become open, friendly places again.
Purchase price - automated cars still have this.
Fuel - automated cars still need to buy this.
Tax - automated cars enjoy a discount now, sure. But it's £10 a month or thereabouts to me.
Insurance - automated cars still need this, and will always need it (probably larger public liability, in fact)
Maintenance - automated cars still need this.
In fact, all cars, including automated, have pretty much these same costs.
What you're arguing against is "a car", not "an automated car".
And what you're doing is suggesting we outsource to a third party to manage our fleet for us, add 50% profit, and sell the same thing we could have bought ourselves back to us? With summoning apps and driving AI and fleet management logistics to always be ready for peak periods in the right places, etc. And that that will somehow be magically be less?
I've done 12,000 miles in the last two years alone, most of it for work. If you factor in that it's a brand-new car, that's about £2 a mile. Sounds expensive, I grant you. But it'll pay for itself next year and then that drops drastically - then it's just the fuel, tax, insurance and maintenance. Do I work 500 miles from home? No. Ten. 10 x 2 (back and forth) x 5 day x 48 weeks = 4800 miles a year, just for basic commuting.
If it's more than 50p a mile, I'll be amazed. Let's do the maths. 45mpg. £1.10 a litre. 4.5 litres in the gallon. That's 11p a mile in fuel. 3p a mile (at most) for my car tax. 18p for insurance, the biggest ongoing cost. That's 32p, plus maintenance (first three years, no MOT required, one tyre costing £100). It's due an oil change, sure, but that's lost in the noise like the tyre.
Even assuming I've vastly under-estimated, and call it 60p per mile to recoup costs over a reasonable liftime - it's still within what my work would pay for travel expenses if I needed to claim for it. They know how much it costs to run such things, because they run their own fleet too.
I defy you to find a taxi or car share or car rental service that averages to less than 50p a mile, or even less than £2 a mile if you're include paying off the car to own it outright at the end. Zip is £3 per hour plus 29p per mile plus £6 per month (they have a deal at the moment, though) for the cheapest car they have, if they have one in your area, at the time you need it, that's not being used. Weekends, evenings, larger cars, longer periods, etc. mean it costs a lot more. I couldn't afford it for my usual commute, I know that, let alone the hassle of trying to get home if they don't have a car available (or paying to guarantee the car is mine for the day).
Sorry, but the reason people use cars rather than public transport, hiring taxis, or automated vehicles, is because they are a cheaper option, and infinitely more flexible. So long as you don't work in a completely car-averse area (e.g. congestion zone), they are a no-brainer. Automating them and charging profit won't change that - the same costs apply, if not more. People will still use them as taxis, etc. same as taxis have a place now. But they aren't going to be the every-person, every-journey vehicle for a long time.
> ... the reason people use cars rather than public transport, hiring taxis, or automated vehicles, is because they are a cheaper option, and infinitely more flexible...
To put it another way, shared services only cover a small proportion of use cases, to be fair quite a few of those cases are heavily used, but that's not the point. This will all end up penalising everybody who doesn't fit some pre-defined customer profile, just like everything else as-a-service.
Most of your points are blunted by Economies of Scale. Buying a whole fleet of cars tends to get you a discount like any other commercial-level bulk purchase. Same with fuel, where fuel supplies can be obtained in creative ways (like fuel futures) to save costs. Insurance and maintenance, meanwhile, will likely be lower due to lower risks and less human control (something you CAN'T say about human-driven cabs). As for the fares, they're high BECAUSE so few use them, raising the per-passenger break-even point. Again, economies of scale can come into play once more people gain the option to eschew buying a car.
In 30 years people won't own cars. There's no such need, how many hours per week do you really use it?
Do you, by any chance happen to live in a city with busses going every possible direction that you could ask for, that appear roughly every 5 minutes?
Some of us in that wide open space sometimes called "the countryside" that has a one bus every two hours service with public transport, and the bus only goes on one route by the longest and most torturous route possible to get the greatest number of passengers. Personally, driving saves me just shy of 3 hours a day compared to public transport on my trip to work.
If I'm not driving in 30 years then I feel that it's likely to be because:-
1) I've retired
2) I'm working remotely.
Or, even better yet, a central server loaded to which every road diversion and sign change in the UK is reported. Just the sort of centralized and hugely expensive computer system Ministers, and senior civil servants, so love.
The question is not how to solve the problems around supporting driverless cars.
The question is should the UKG p**s away another shedload of cash to do so for the (relatively) few individuals who want (and can afford) them and the public companies who are dreaming (eventually) of firing all their delivery drivers.
If I were wanting to do this I would target the back end distribution centre --> big stores. Fixed sites, fairly fixed roots, and run the vehicles as AGV's with an RPV option. With proper scheduling a small team of drivers could be time shared between a larger number of vehicles for critical parts of the journey or emergency override.
Sorry, it's not exactly Thunderbirds, but it would trade the risk of failure for people who'd be prepared to test out the tech while (in principal) making them some savings while still having a fall back plan in case it goes TITSUP.
Do I need to state the obvious? British roads are not Californian freeways. Does Google even know what the word "roundabout" means? There's a shed load of evidence to be collected before any system like this is let loose in "full auto" mode.
What we need now, not in 10, 20, or 50 years time, is some AI in the bloody traffic lights, so that they work in the interests of the ordinary driver as opposed to the interests of the police, the bureaucrats, road safety campaigners, pedestrian campaigners, or cycling campaigners.
Can anyone explain why cyclists, the slowest traffic on the road, are allowed pole position at traffic lights rather than being relegated to the rear where they belong?
And why do the lights at some junctions have the green light cased in a box so it is almost impossible to see?
"And why do the lights at some junctions have the green light cased in a box so it is almost impossible to see?"
To stop people monitoring the cross-junction lights and jumping the gun - to meet the people who think the first three seconds of red are equivalent to green anyway.
Firstly it doesn't, I've not heard of a town that doesn't fall into chaos if the automation fails, and secondly these days road layout design and light sequencing is intended to minimise accidents, not to maximise traffic flow, and the stats suggest they are really doing rather well at that.
"Can anyone explain why cyclists, the slowest traffic on the road, are allowed pole position at traffic lights rather than being relegated to the rear where they belong?"
Because it is the easiest way to prevent an accident. They are slower, so the second gained by putting them on the front make it easier/safer to take a left (or right). Come to think of it, will probably be easier on the traffic too - as there will be no retention, while the cyclist negotiate the turn with the incoming traffic.
I'm curious how autonomous cars do in the snow. My guess is "what's snow and why can't I see any signs and why do all my control inputs seem sluggish?". This seems like a great idea, really; have a human driver that hasn't really driven in months suddenly have to drive in one of the more challenging situations. How hard can it be?
There are many reasons I would want an autonomous car.
- It could drop me off at the store door, then go find parking. When I'm done, I would call it to come pick me up. I wouldn't have to carry my packages half a mile to my parking space.
- I could relax and watch video or something on long trips.
- It could drive me home after an evening out.
- I could send it off to get gas, or be serviced.
Biting the hand that feeds IT © 1998–2019