Jeremy Vine Show - Pilotless Airliners
Does your aircraft suggest the cruising levels?
Does your aircraft suggest the speed to fly?
Does your aircraft suggest the speed to fly?
Mine does, but it sure as heck can't make a reasoned decision as whether it is safe or sensible to fly those levels or speeds.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
No, because it is not designed to work beyond simple parameters.
It can easily be given wider parameters.
Can you suggest a parameter that a human can use that the machine could not?
I think that we can both agree that modern aircraft are well beyond a 2, even if I would not suggest we are quite as far down the list as others.
Certainly autopilot and autothrust alone are a well up the list, at least a 5 and they are ubiquitous since the 60s
It can easily be given wider parameters.
Can you suggest a parameter that a human can use that the machine could not?
I think that we can both agree that modern aircraft are well beyond a 2, even if I would not suggest we are quite as far down the list as others.
Certainly autopilot and autothrust alone are a well up the list, at least a 5 and they are ubiquitous since the 60s
I thought Jeremy Vine was quite good when he first started on Radio 2. However he has steadily got worse and now he behaves like a total idiot. Worse even than The Daily Mail for sensationalism and innacuracies. And the other day he ran a piece about why people use those ski pole walking stick things on hikes. Groundbreaking journalism at its best ???
When Ken Bruce finishes, I now switch over to Mark Radcliff and Stu Maconie on BBC 6 music until Steve Wright comes on R2.
Thank goodness for DAB radio !
When Ken Bruce finishes, I now switch over to Mark Radcliff and Stu Maconie on BBC 6 music until Steve Wright comes on R2.
Thank goodness for DAB radio !
"Mildly" Eccentric Stardriver
Can you suggest a parameter that a human can use that the machine could not?
1. fly through it - good for the bean-counter's economics, since minimum flight time.
2. Fly around it - good for not scaring passengers, spilling drinks etc. What is the cost to the company of "I wouldn't fly with x airline again. We had a terrible time; my wife (husband?) was terrified."
Program a computer for that one.
Nemo Me Impune Lacessit
Join Date: Jun 2004
Location: Derbyshire, England.
Posts: 4,091
Received 0 Likes
on
0 Posts
Can you suggest a parameter that a human can use that the machine could not?
As mentioned earlier, by John Farley and others, a computer is not capable of thought nor anything better than simple analysis. A computer can't think outside the box.
Tourist
You are looking at systems in isolation not as a whole. Sure TCAS would be up the scale in isolation but the pilot can still ignore it. And yes it has to be done sometimes.
Now a 737NG or a 767 has a TCAS unit but that doesn't mean it is anywhere close to being a automated aircraft. I think you're making the assumption that just because modern aircraft already have some sort of computer monitoring that we are only a small step to full automation. The problem is that to go from where we are to full blown autonomous is a massive massive technological, logistical, legal and operational leap.
Airbus operate on a different philosophy however they are discovering that this also has its drawbacks and could be argued is not necessarily the safest option.
For the record I'm not saying it can't be done I'm saying that the staus quo with some better techonology is probably the safest and most cost effective for now. I don't think full autonomy in pax aircraft will be happening anytime before 2090
You are looking at systems in isolation not as a whole. Sure TCAS would be up the scale in isolation but the pilot can still ignore it. And yes it has to be done sometimes.
Now a 737NG or a 767 has a TCAS unit but that doesn't mean it is anywhere close to being a automated aircraft. I think you're making the assumption that just because modern aircraft already have some sort of computer monitoring that we are only a small step to full automation. The problem is that to go from where we are to full blown autonomous is a massive massive technological, logistical, legal and operational leap.
Airbus operate on a different philosophy however they are discovering that this also has its drawbacks and could be argued is not necessarily the safest option.
For the record I'm not saying it can't be done I'm saying that the staus quo with some better techonology is probably the safest and most cost effective for now. I don't think full autonomy in pax aircraft will be happening anytime before 2090
. I don't think full autonomy in pax aircraft will be happening anytime before 2090
As an aside looking back at the thread I noticed a few quotes from sci-if visionaries thrown into the mix. Now the likes of Clarke, Asimov et. al. may at times have been remarkably prescient but they were often horribly wrong when it came to timescales.
We will go to Jupiter ( or Saturn if you prefer the book in question) one day....but it won't be in 2001. We will have fully automated airliners one day.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Nothing further was implied from his post or mine.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Not a parameter, but a scenario. "Bumbly" cloud ahead. Not a nasty cb, but one that will throw the aircraft around a bit. Two options:
1. fly through it - good for the bean-counter's economics, since minimum flight time.
2. Fly around it - good for not scaring passengers, spilling drinks etc. What is the cost to the company of "I wouldn't fly with x airline again. We had a terrible time; my wife (husband?) was terrified."
Program a computer for that one.
1. fly through it - good for the bean-counter's economics, since minimum flight time.
2. Fly around it - good for not scaring passengers, spilling drinks etc. What is the cost to the company of "I wouldn't fly with x airline again. We had a terrible time; my wife (husband?) was terrified."
Program a computer for that one.
Human captains will all produce different results with this scenario. Everybody has their own triggers.
The computer could just be given a sliding scale, the equivalent of a commercial index for the day which decides exactly how much chop is acceptable vs the time delay.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
I think it would have to be a part of any fully automated system, and will have obvious huge benefits.
I don't think there is any technological reason why it couldn't be done immediately.
Any such fully integrated autonomous ATC system would obviously require huge backup independent systems
I think the biggest hurdle confronting computers ever having executive control will be their perception, by which I mean how the computer sees and analyses the world around it. The human vision system consists of the eyes and the brain. The eyes produce a rough upside-down image which the brain then processes by removing faults, flipping it over and combining views to give a stereo picture. The brain then uses short term memory to further enhance the scene by combining views of different focus and exposure to create an overall scene. It then uses long term memory to recognise what it is seeing.
An autonomous aircraft computer would need to be able to realise for example that the storm cell which wasn't painting very much on its radar was nevertheless there because it could see either the lightning or the large mass of the cloud by looking out of the window.
We are decades away from this ability in a computer - if indeed it could ever happen. For it to work safely we would have to build a complete human vision system.
There was an autonomous car that crashed recently, killing a person, because it couldn't tell the difference between a patch of bright sky and a white truck?
I saw Asimo, a humanoid robot a few years ago at the Science Museum. He was very impressive; he could walk up stairs by 'himself' and run unaided without a tether. However, to walk up 5 stairs he had to stand at the bottom for about 2 minutes, calculating what he had to do, and taking positional information from strategically placed marker dots on the stairs. It was nevertheless most impressive. However, at the end of the demonstration, as he walked out of the auditorium, he walked smack into a door that he was expecting to have been left open but had swung shut - such were his limitations of vision.
So this is the challenge. The human brain is the most complex thing in the known universe. It takes a human brain about 20 years to learn and program itself to fly an airliner: Given that when born it first has to make sense of the upside down sight from its eyes and the sounds from its ears; then learn language and to speak, to walk, to use its fingers etc etc, you get the point, I am sure.
There is no computer in the world that can take everything into account as we pilots do in order to fly an airliner, adapting to all the changes in conditions we meet on a daily basis. Yet there already are lots of fully programmed 'human brains' who are willing and able to fly airliners. There have indeed been some terrible crashes caused by humans, but all we need is to go back a step and restore proper selection standards for pilots; non fatiguing rosters; proper training; decent recurrent practice; etc, etc.
We don't need to reinvent the 'wheel' we simply need to make sure it is given a proper chance to work.
.
An autonomous aircraft computer would need to be able to realise for example that the storm cell which wasn't painting very much on its radar was nevertheless there because it could see either the lightning or the large mass of the cloud by looking out of the window.
We are decades away from this ability in a computer - if indeed it could ever happen. For it to work safely we would have to build a complete human vision system.
There was an autonomous car that crashed recently, killing a person, because it couldn't tell the difference between a patch of bright sky and a white truck?
I saw Asimo, a humanoid robot a few years ago at the Science Museum. He was very impressive; he could walk up stairs by 'himself' and run unaided without a tether. However, to walk up 5 stairs he had to stand at the bottom for about 2 minutes, calculating what he had to do, and taking positional information from strategically placed marker dots on the stairs. It was nevertheless most impressive. However, at the end of the demonstration, as he walked out of the auditorium, he walked smack into a door that he was expecting to have been left open but had swung shut - such were his limitations of vision.
So this is the challenge. The human brain is the most complex thing in the known universe. It takes a human brain about 20 years to learn and program itself to fly an airliner: Given that when born it first has to make sense of the upside down sight from its eyes and the sounds from its ears; then learn language and to speak, to walk, to use its fingers etc etc, you get the point, I am sure.
There is no computer in the world that can take everything into account as we pilots do in order to fly an airliner, adapting to all the changes in conditions we meet on a daily basis. Yet there already are lots of fully programmed 'human brains' who are willing and able to fly airliners. There have indeed been some terrible crashes caused by humans, but all we need is to go back a step and restore proper selection standards for pilots; non fatiguing rosters; proper training; decent recurrent practice; etc, etc.
We don't need to reinvent the 'wheel' we simply need to make sure it is given a proper chance to work.
.
Last edited by Uplinker; 23rd Aug 2016 at 13:36.
I listened to the Jeremy Vine show and JV was just doing what he is paid to do - provoke debate. The chap defending the role of pilots (ex BALPA, sorry can't remember his name) didn't put forward a very convincing argument. All I can recall him saying was that humans were better at 'detecting thunderstorms'...err, I'm not sure about that.
It is a pity that someone like John Farley wasn't asked to contribute.
It is a pity that someone like John Farley wasn't asked to contribute.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
A specifically not autonomous car with a basic autopilot which has never ever claimed to be autonomous and is sold as "beta testing" hit a truck because despite warnings that it definitely was not autonomous and required constant supervision the driver did not pay attention.
Elon Musk: Tesla's Autopilot is twice as safe as humans
It is worth noting that with millions of miles on real roads already driven by these various prototypes and halfway-house cars, that despite the over-excitement by such as Uplinker, they have a better safety record than manned vehicles.
1 million per year die on the roads in cars driven by our perfect little brains. They don't have to be perfect, they just have to be better or equal.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
I think the biggest hurdle confronting computers ever having executive control will be their perception, by which I mean how the computer sees and analyses the world around it. The human vision system consists of the eyes and the brain. The eyes produce a rough upside-down image which the brain then processes by removing faults, flipping it over and combining views to give a stereo picture. The brain then uses short term memory to further enhance the scene by combining views of different focus and exposure to create an overall scene. It then uses long term memory to recognise what it is seeing.
.
.
Our vision system is notoriously quirky and easy to confuse. This is why there are millions of optical illusions. Computers don't suffer from them.
http://www.simplypsychology.org/perc...-theories.html
https://www.newscientist.com/article...-what-you-get/
https://www.newscientist.com/article...ict-the-world/
Last edited by Tourist; 23rd Aug 2016 at 14:55.
Parson
Maybe, but OTOH I most certainly wouldn't want to rely on the autopilot blindly following the guidance provided by some of the more advanced digital weather radars for CB avoidance.
Tourist
Computers learn to fall for optical illusions - Futurity
"“The field of image recognition has been revolutionized in the last few years,” Yosinski says. “[Machine learning researchers] now have a lot of stuff that works, but what we don’t have, what we still need, is a better understanding of what’s really going on inside these neural networks.”
All I can recall him saying was that humans were better at 'detecting thunderstorms'...err, I'm not sure about that.
Tourist
.........there are millions of optical illusions. Computers don't suffer from them.
"“The field of image recognition has been revolutionized in the last few years,” Yosinski says. “[Machine learning researchers] now have a lot of stuff that works, but what we don’t have, what we still need, is a better understanding of what’s really going on inside these neural networks.”
Last edited by wiggy; 23rd Aug 2016 at 15:44.
Join Date: Jan 2001
Location: Home
Posts: 3,399
Likes: 0
Received 0 Likes
on
0 Posts
Parson
Maybe, but OTOH I most certainly wouldn't want to rely on the autopilot blindly following the guidance provided by some of the more advanced digital weather radars for CB avoidance.
Tourist
Computers learn to fall for optical illusions - Futurity
"“The field of image recognition has been revolutionized in the last few years,” Yosinski says. “[Machine learning researchers] now have a lot of stuff that works, but what we don’t have, what we still need, is a better understanding of what’s really going on inside these neural networks.”
Maybe, but OTOH I most certainly wouldn't want to rely on the autopilot blindly following the guidance provided by some of the more advanced digital weather radars for CB avoidance.
Tourist
Computers learn to fall for optical illusions - Futurity
"“The field of image recognition has been revolutionized in the last few years,” Yosinski says. “[Machine learning researchers] now have a lot of stuff that works, but what we don’t have, what we still need, is a better understanding of what’s really going on inside these neural networks.”
That article you link to is very interesting, but I don't think it has relevance to this discussion. They are deliberately trying to find ways to confuse cameras to aid understanding of neural nets. Not really the same thing as humans having inbuilt flaws.
Of course they are not really flaws, we are merely optimised to be on land at slow speed so all our senses are adapted to that environment.
Some of these many optimisations in vision, balance etc are a hinderance in the air.