Confirmed: Driverless cars to hit actual British roads by end of year Driverless cars will hit the streets of Britain by the end of this year, the government has confirmed. In a briefing document titled Action for Roads, the Department for Transport confirmed that Oxford University boffins will start trials of autonomous cars later this year. The scheme to take the human element out of motoring …
Current models only work the steering, the driver still has to apply throttle & brake.
Volvo have recently shown off a car which will drop the humans off at the entrance to a multi-storey car park, then drive in and find itself an empty space.
How the driver finds or summons his car back to take him home was not revealed.
"Volvo have recently shown off a car which will drop the humans off at the entrance to a multi-storey car park, then drive in and find itself an empty space."
' Now everyone remember where we p....oh right. Ermmm...right you go with John and Lisa and I'll go with Claire, we'll meet back here in an hour if you can't find it, everyone got everyone else's number? '
Car parks of tomorrow full of lost souls endless searching for their cars, while their cars keep moving around!
> How the driver finds or summons his car back to take him home was not revealed.
Send it an SMS, or an email? A big "come here, car" button at the entrance that uses RFID/NFC to work out who you are and which car to summon?
The other thing it will be able to do is drop you off at the office door, then go home and park on your drive, thus avoiding parking charges. It will then monitor the traffic conditions to know when to set off to be at the office door to pick you up at the end of the day.
Imagine a big gaggle of people outside the office at 17:01 waiting for their car, all jumping on every silver Mondeo that appears...
No, but you will find it easier to park your car because those who would find this useful won't be filling up all the spaces at the end of their 20 minute drive (that could have been done on the excellent London Transport system).
It seems a bit odd you are using a car into London from Cambridge - the train service is very good, isn't it?
Ah so you've been to the short term parking at Manchester Airport as well. Let the car drive round looking for a parking space - as they don't believe in lighting the "car park full" sign - meet the wifey and stop the car on its next round past looking for a place.
"Volvo have recently shown off a car which will drop the humans off at the entrance to a multi-storey car park, then drive in and find itself an empty space."
So tell me @ WonkoTheSane ...if the car parks itself in an ILLEGAL parking space...just WHO is responsible for the parking ticket?
The car owner? The car's passenger(s)? The Programmer who wrote the software? The company who manufactured the car? The car itself? Because after all...the owner/passenger(s) didn't make the choice of parking spots...the CAR did.
And on top of all of this...just exactly WHY do we need "self-driving" cars anyway? If people don't want to drive a car...TAKE A BUS FFS!
"Volvo have recently shown off a car which will drop the humans off at the entrance to a multi-storey car park, then drive in and find itself an empty space."
Yup, same for Audi. On some markets they're also sold with an automatic drive in congestion mode. They basically add automatic steering to what adaptive cruise control already does. The automatic steering itself likely reuses technology from lane guidance and vehicle detection as used for automatic beam deflection.
One small step at a time seems the way to go.
I have tried VW's version, and it is useless... slow, and not very accurate... much faster to do it yourself...
Though to be honest they are still probably a thousand times quicker than the multiple people I have been behind while they try to get into a parking spot.
I watched someone attempting to parallel park the other day. It was truly amazing to watch. The space was, actually pretty big. She stuffs the nose in first (Noooo!), then spent at least 5 minutes nudging back and forth not getting any closer to the kerb. Then gave up, left it at an angle and walked off.
Truly stunning.
"You take the number of vehicles in the field (A) and multiply it by the
probable rate of failure (B), multiply the result by the average
out-of-court settlement (C). A times B times C equals X. If X is less
than the cost of a recall, we don't do one."
Seems apt!
First semi-autonimous
Then fully automatic
Then when somebody pisses off the government, their car will 'malfunction' straight into a tree.
Although it'll make it much easier to catch hit and run / GTA, a clunk on the bonnet of a certain force? Cut the engine, Guy steals your car? Cut the engine remotely.
There's just one problem with all of this, as long as there are people who will be able to drive manually then the automatic driving system just won't work.
People are opportunists, and they will zip and dash and squeeze and rush and crush and yell and scream and beep and always be about beating everyone else at the lights, and just getting past 'the light was only pink, officer!'.
Until the day you can force everyone to go to automatic driving, or perhaps change liability so the fault always lies with the manual driver, it just won't work.
It's been pretty much proven by Google already that it doesn't really matter whether other drivers are ducking and weaving all over the place. Your fully autonomous vehicle can see them coming via radar a few hundred yards off, detect the erratic driving patterns and act accordingly to maximise safety and efficiency. If anyone's going to get hit or delayed by someone driving like a nutter it's the poor guy with the non-autonomous car who has to try to react to what's happening all by himself, with only two eyes and his own wits, reactions and personal driving experience to help him. As soon as one of these things actually goes on sale, it'll have many millions of miles of driving experience per week and, as long as the programmers are up to it - which so far they seem to have been save for sitting in the centre of the lane on a Tokyo highway (nobody else does, they all make themselves an extra lane, so it caused a traffic jam, now apparently fixed) and the sheer idiocy of both the guy who rammed one in stationary traffic and the other guy who switched off the automation and drove it into a post - these vehicles will be safer and more efficient than anything we've ever seen on the road.
And better still, if you root it, you can teach it to tease traffic wardens.
But the thing is, when Murphy intervenes, it's going to be from an angle that no one had even had the thought to cover.
For example, would an automated car be able to react well to a low-to-ground obstacle suddenly falling off the back of a truck? What about a child suddenly running out in front from between two cars (thus practically invisible beforehand)? Can the car detect small but significant patches of black ice? How will it react to an accident suddenly starting in front of them? And so on?
> would an automated car be able to react well to a low-to-ground obstacle suddenly falling off the back of a truck
No worse than a human. The sensors will react quicker than a human, and it might be more accurate in predicting which way it's going to bounce.
> What about a child suddenly running out in front from between two cars (thus practically invisible beforehand)?
Again, it should be able to react to the human-shaped infra-red/laser imaging signature appearing quicker than a human can. It would also be able to work out quickly whether it can go around them without causing a serious collision.
> Can the car detect small but significant patches of black ice?
Yes, the infra-red signature would be different. It also might be able to detect itself losing control quicker, with accurate accelerometers comparing actual movement to control inputs.
The other thing is that, with software updates, all the "drivers" learn from the mistakes made by other automated cars.
"For example, would an automated car be able to react well to a low-to-ground obstacle suddenly falling off the back of a truck? What about a child suddenly running out in front from between two cars (thus practically invisible beforehand)? Can the car detect small but significant patches of black ice? How will it react to an accident suddenly starting in front of them? And so on?"
Well, an autonomous car is not perfect - but so aren't we!
I'm not sure about the hypothetical box hitting the road, but would bet that the car would "see" it. Simple because it must knows where the sidewalk ends - and to that it must "see" things close to ground.
The child would be, at least, treated as another obstacle. The first priority (or one of them) is not hit things. So, the car would break/swerve in order to avoid the child. If it would be successful is another argument. But i'd risk saying the child odds would be better with a computer driving the car. Simple because it doesn't panic.
About the black ice... no idea.
"For example, would an automated car be able to react well to a low-to-ground obstacle suddenly falling off the back of a truck? What about a child suddenly running out in front from between two cars (thus practically invisible beforehand)? Can the car detect small but significant patches of black ice? How will it react to an accident suddenly starting in front of them? And so on?"
The answer is simple. Better than the manual driver will. Yes the automatic car might run over the little kid but then again so was the manual driver anyway. If the little kid could have been saved, it would be the automatic car. The car will have seen the child, calculated the path and speed, calculated that it cannot avoid the child and now aims the car at the child so the child hits the car in the middle of the best crumple zone maximising the chances of survival and reducing the chance of going under the wheels. All of this in less time than the driver can say "Holy Sh..."
"Your fully autonomous vehicle can see them coming via radar a few hundred yards off..." Yes, that'll work while only one or two cars are using radar in the same space. However, when the idea catches on and every one of hundreds of cars is pinging constantly, the system is going to break down. Even if it's not based on "radar", the sheer number of independent comms channels required will exceed practicability if channel separation is maintained, or alternatively there will be loss of signal integrity and consequent errors (aka accidents).
The very simple alternative solution would be to teach folks to drive properly - it's not that hard.
Hundreds of cars? Meh, guided missile cruisers and other military systems such as AWACS have been able to track and target hundreds of objects in real time and 3 dimensions for many years. That includes incoming threats moving at supersonic speeds.
Even on a busy round-about with 5 or more roads merging, you are probably only looking at a few dozens cars at any point in time being a threat or which the system needs to exchange signals with.
Compare that with meat bags currently driving vehicles in busy traffic; I occasionally ride through a complex round-about system with inner and outer lanes and 5 entrance/exit points. Even a ZX81 could navigate it better than some drivers I see failing to cope.
Should be pretty obvious from the road toll in most countries that extensive driver training and enforcement does not work that well.
> extensive driver training and enforcement does not work that well.
I'm not aware of any country that does either of those.
Most countries do *minimal* training and superficial enforcement. Except when it comes to speed cameras, of course...
Vic.
No, those are the challenges that make the autonomous vehicles difficult to develop.
If you were starting from a blank sheet, you'd build your towns/cities/villages/etc so that vehicles and pedestrians were completely separated, vehicle automation would then be relatively straightforward as you can significantly reduce the number of unexpected inputs they need to react to.
The problem is that we're not starting with a blank sheet, not by a long way.
"If you were starting from a blank sheet, you'd build your towns/cities/villages/etc so that vehicles and pedestrians were completely separated"
But that is hardly practical. After all, if it was, it would have been done already with those unpredictable human drivers on the roads!
I have to agree with the general sentiment here.
And separated paths from pedestrians and other cars for autonomous vehicles to follow? We have that. It's called a train line.
I don't get where autonomous cars fit in as opposed to just tram/train-ifying everything properly. If you have to program the car to run the gauntlet of human drivers, but expect it to subvert human drivers, then just build another lane which the car CAN'T veer from (i.e. a tram line, in effect, but maybe with a "virtual" barrier of a white line). No problem with obstacle detection, reaction times, swerving, etc. and you can programmatically select your course as you go (we call them points).
Mandate that one lane on every motorway is this type of lane, and you can introduce them in a matter of years. Cars are much easier to design and use if they are separated off like this. Drivers can drive to the motorway, enter the lane, press a button, then go to sleep until they get near their destination. Drivers of old cars can still use the roads but have an incentive to upgrade quite cheaply (and maybe you could not tax the cars capable of using those lanes, or make them not subject to tolls, etc.). Over time, you can just add more lanes until it's viable to say "Okay, all roads are like this now" in 15-20 years time when all cars have the feature. THEN you can talk about car that have only autonomous functionality.
But, as it is, we've spent 30 years messing about trying to make cars brake together and follow lines and have only just got the legal framework in place to allow a prototype to touch a normal road for the first time.
I'm still interested to know what an autonomous car does if it sees a) a sheer cliff in front of it's wheels (i.e. road collapse, cliff collapse, unmarked road edge, etc.) and b) a large pothole / roadworks exposing the asphalt / blind summit that could be mistaken for the above. Because, to my mind, both scenarios are a potential "drive off the edge" / "slam brakes on and refuse to proceed" situation if you're an autonomous car, and neither are incredibly unlikely, and both reactions will cause serious problems.
Ignore autonomous cars. Bring out a "personal train". It's a car that follows a white line with sensors on the front that brake the car for any obstacle. No need for complex interpretation as the line-following/braking mode need only be activated when on a designated lane (make the lane somehow announce itself and it's automatically switched on when necessary). At all other times it's a normal car.
Road cost: A line of paint and a designated lane, paid for by tolls on other lanes or a tax on "automated cars" (recouped because you don't have to pay as much insurance if most of your journey is not you driving).
Car cost: Next to nothing compared to a completely autonomous system. Optional. New cars only, 15-years from now, every car could have it.
Driver cost: A button on the dash / indicator light to show when you're on an autonomous road.
Insurance cost: On the separated, illegal-to-stop-on, no-pedestrian, isolated, secure, monitored-for-hazards-by-a-bog-standard-CCTV, motorway-only, can't-cross-onto-without-a-compatible-car-going-through-an-automated-entrance-gate lane, the manufacturer / road maintainer cover accidents jointly. All other places, it's the driver's responsibility.
People would buy one now just to get through a congestion-charge-free or M6-toll-free lane to encourage their use. It doesn't have one millionth of the liability problems of driving next to idiots. It can be rolled out on motorways NOW and smaller roads follow if necessary once its successful. If it flops, we just paint the lane back a normal colour.
But for some reason we've skipped this step entirely and are trying to replicate humans directly in a computer, like the way early aviation pioneers use to flap huge wings and run up and down.
Lee D,
What happens if your special white line is broken? It could be blackened by skidmarks, for instance, or just plain worn out in places. The onus would be on the state to make sure all white lines were spick & span. Would you rather trust the state road-maintenance crew or your own car's systems with your life? And how much carnage could a saboteur cause by pouring some black paint on a white line on a motorway?
"If you were starting from a blank sheet, you'd build your towns/cities/villages/etc so that vehicles and pedestrians were completely separated"
Wasn't Milton Keynes built like this ? The downside being with no mixing of cars and pedestrians, it can be a very dangerous place to walk at night, as you're out of sight ?
MK was indeed built like this - at least for the grid roads and yes there are trees and shrubbery separating the two in many places.
You might be surprised how few incidents there are relative to the population though, my gut feeling is there are no more attacks in the entire city than would be expected if it was built to a traditional layout - maybe even less but I don't know why this is.
Probably a good thing to have the main roads separate to pedestrian areas though since the transit is so congestion free the roads are like drag strips nearly all of the time.
This post has been deleted by its author
I was travelling in the passenger seat, with my grandfather (99 last week) driving about 15 years ago in his British Leyland Mini Metro.
He'd previously mentioned his eyes weren't quite as good as they used to be, but didn't make a big thing of it, so didn't seem too bad...
We came around a corner, and there was a black and white traffic cone on the the left side of the road.
"Oh my god - there's a penguin!"
Suffice to say, we relieved him of the keys from that day... for 2 reasons!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
