The robots are coming. Restructure the economy. Go. - The Bear and Badger

Funkstain

I am tickled that we feel the same way!

Off-piste: in the current affairs thread, the IsHam thing - it occurred to me after my last post on the subject that many people must feel similarly to the way I was feeling, when they argue with me. I can be obtuse, dogmatic, unresponsive to different inputs and highly responsive to confirmation bias. Reminded me of when I was defending the view that corporation tax is a bad idea, economically and does nothing to redress an unfair society; multiple people disagreeing, me sticking to guns no matter what. Humans are such flawed creatures.

monkey

In the future, robots will do all the forum debating for us.

g.man

*shrugs*

dynamiteReady

Funkstain wrote:

I am tickled that we feel the same way! Off-piste: in the current affairs thread, the IsHam thing - it occurred to me after my last post on the subject that many people must feel similarly to the way I was feeling, when they argue with me. I can be obtuse, dogmatic, unresponsive to different inputs and highly responsive to confirmation bias. Reminded me of when I was defending the view that corporation tax is a bad idea, economically and does nothing to redress an unfair society; multiple people disagreeing, me sticking to guns no matter what. Humans are such flawed creatures.

Man... Do what you do...

JUST ACCEPT THAT I'M RIGHT, INNIT!!?

Brooks

When robots can shrug convincingly, that's when we're proper fucked.

GooberTheHat

The car to plane analogy is so false. There is so little in common between the two systems of transportation that it's meaningless to compare them.

dynamiteReady

GooberTheHat wrote:

The car to plane analogy is so false. There is so little in common between the two systems of transportation that it's meaningless to compare them.

It's not a direct comparison we're (well... I;)m trying to make. It's more about trying to determine how much more work will go into one, or the other.

The articles I've read so far generally suggest that we'd be closer to automating flights, than buses...

For one, a lot of the work has already been done (that the phrase/word autopilot is in common parlance should be a big enough clue to this), and after that, (and this is me making this point) a lot of the work that goes into insuring a safe flight takes place on the ground... not in the cockpit.

That's very far from true for a bus route. Even something as simple as a zebra crossing is enough to fuck up a preplanned, automated bus journey.

Let alone a roundabout with a faulty set of lights, or a box junction.

Someone will automate a cargo flight...
I don't think that will be too far away either. Landing, et al.

And they'll have a good chance of doing so before they can do the same with a fleet of lorries... But then thinking laterally, why do it with a fleet of lorries?

You might want to do most of that work by train... And they'd be much better targets for automation than either the car or plane...

GooberTheHat

But the whole thing about automobile automation is that the safety net if any of the multitude of sensors stop working or provide contradictory information, is to pull over and stop.

If the sensors and instruments on an aircraft stop working it still has to try and land. The altitude accuracy of GPS can be out by as much as 10 meters at any one time. That's not uncommon. If the altimeter breaks landing is going to be interesting. I would imagine a miscalculation of 10 meters could be the difference between a clean landing and a collapsed landing gear.

Yossarian

dynamiteReady wrote:

a lot of the work that goes into insuring a safe flight takes place on the ground... not in the cockpit. That's very far from true for a bus route. Even something as simple as a zebra crossing is enough to fuck up a preplanned, automated bus journey. Let alone a roundabout with a faulty set of lights, or a box junction.

If all of this were true, how is it that driverless cars are currently undergoing testing for common use while pilotless planes are not?

dynamiteReady

http://uk.businessinsider.com/matternet-uav-delivery-drones-2014-11

Hulka T

Even if the tech for driverless cars is terrible, it will still be better than most drivers in london.

Yossarian

dynamiteReady wrote:

http://uk.businessinsider.com/matternet-uav-delivery-drones-2014-11

By launching in areas like Papua New Guinea and Bhutan — where its drones are sometimes flying over swaths of jungle or sparsely populated desert and where there isn't a strict existing aerial infrastructure — Matternet has fewer barriers to entry.

"The risk needs to be low if the UAV falls," NASA's Kopardekar tells Business Insider, noting that low-density areas make the most sense for drone testing and operations. "You need to be able to demonstrate the risk is very low and the benefits are high: That's where you will see implementation happening."

Seems like it's a ways off being tested in populated areas to me.

GooberTheHat

dynamiteReady wrote:

GooberTheHat wrote:

The car to plane analogy is so false. There is so little in common between the two systems of transportation that it's meaningless to compare them.

It's not a direct comparison we're (well... I;)m trying to make. It's more about trying to determine how much more work will go into one, or the other.

The articles I've read so far generally suggest that we'd be closer to automating flights, than buses...

For one, a lot of the work has already been done (that the phrase/word autopilot is in common parlance should be a big enough clue to this), and after that, (and this is me making this point) a lot of the work that goes into insuring a safe flight takes place on the ground... not in the cockpit.
but if something does go wrong what is the reversionary, or safety procedure? If the device that measures air speed stops working how do you compensate for that?

That's very far from true for a bus route. Even something as simple as a zebra crossing is enough to fuck up a preplanned, automated bus journey.
That's a simple problem though, if clear, go, if not, stop, if in doubt, stop.

Let alone a roundabout with a faulty set of lights, or a box junction.

Someone will automate a cargo flight...
I don't think that will be too far away either. Landing, et al.

And they'll have a good chance of doing so before they can do the same with a fleet of lorries... But then thinking laterally, why do it with a fleet of lorries?

Trains are limited to train tracks, of which there are far less of than roads. Automated trains to auto medium goods vehicles depots for delivery to the local area seems like the best fit solution to me

You might want to do most of that work by train... And they'd be much better targets for automation than either the car or plane...

dynamiteReady

Yossarian wrote:

a lot of the work that goes into insuring a safe flight takes place on the ground... not in the cockpit. That's very far from true for a bus route. Even something as simple as a zebra crossing is enough to fuck up a preplanned, automated bus journey. Let alone a roundabout with a faulty set of lights, or a box junction.

If all of this were true, how is it that driverless cars are currently undergoing testing for common use while pilotless planes are not?

Don't know. And that's more of a political question.
That said, all the sources that I've found so far cite the lack of a business case.

Furthermore - http://www.gizmag.com/uav-future-of-unmanned-flight/27478/

So it's not like it can't be done, because it is being done.

GooberTheHat

Unmanned Aerial Vehicle (UAV) or Unmanned Aircraft System (UAS) – Background

The UAV/UAS has become a staple of the mission options available to the Department of Defense, with its roles and capabilities increasing along with utilization. However, this increase in utilization has been accompanied by significantly higher failure rates when compared to conventional airframes. The general aviation mishap rate is about 1 per 100,000 flight hours, while military UAV systems have experienced failure rates 2 orders of magnitude higher, nearly 1 per 1000 flight hours [1]. By another metric, some UAV systems have a failure rate of 25% [7]

From here:
http://www.dfrsolutions.com/white-papers/improving-unmanned-aerial-vehicle-uav-reliability/

Aircraft are so much more complicated to pilot than cars that, without a man in the aircraft (and not just "in the loop"), there is always going to be more accidents than cars, and any accident is likely to be fatal, unlike cars.

Cars have 4 basic control inputs: go, stop, left, right, and they operate on a 2D space. Aircraft have a multitude of inputs that need to be balanced against altitude, air speed, cross winds, weight, fuel load and many more probably. Any errors in the instruments that provide that data and you're looking at a potential crash. With a pilot with the necessary skill and experience, they would be able to judge and say "that's not right at all, let me make an adjustment".

Elmlea

dynamiteReady wrote:

Don't know. And that's more of a political question. That said, all the sources that I've found so far cite the lack of a business case. Furthermore - http://www.gizmag.com/uav-future-of-unmanned-flight/27478/ So it's not like it can't be done, because it is being done.

Your major error here is mistaking remotely-piloted for autonomous. They're very different things.

On your average flight right now; be it an airliner, a small turboprop, a single-engine prop carrying 10 people, a military jet, a military RPAS or anything else, the aircraft will do a large chunk of the flying. Autopilot is like cruise control at its most basic, and at its most complicated it can follow a calibrated ILS down to the point of landing. However, there's still a pilot there.

For a remotely-piloted system, the pilot is still there, he's just not in it. That's the case with the BAe ASTRAEA aircraft, and all existing military UAVs. The pilot is there from start to finish and the aircraft is never left unmonitored.

That's the current best case in solving my main issue; the complexity of most aviation emergencies means you can't plop a pilot in the seat for the last 10 minutes and expect him to save the day. He needs to have been involved and invested in the sortie, understand the flight plan, the weather, the specific aircraft, the degraded systems, all the issues.

If you want remotely-piloted aircraft, they're obviously already here; I have over a thousand hours flying them. Autonomous aircraft, that get from point A to point B without any human involvement at all, are a much more complex problem.

Lord_Griff

Does my chauffeur count as a "driverless" car?

Elmlea

Lord_Griff wrote:

Does my chauffeur count as a "driverless" car?

Is your chauffeur a driver?

GooberTheHat

Or a car?

Lord_Griff

Elmlea wrote:

Lord_Griff wrote:

Does my chauffeur count as a "driverless" car?

Is your chauffeur a driver?

I have never had the inclination to enquire.

Elmlea

It sounds like you're just getting into some bloke's car and hoping for the best, which is a new form of automation we haven't explored yet.

Lord_Griff

Unknown destination creates a frisson of excitement...

Elmlea

Maximum lols because I know the pilot, but here's a good report:

“With his feet on the flight deck roof, the co-pilot reached down and attempted to disengage the autopilot by pulling back on his sidestick,” the report says.

A330 flight control laws saved Voyager, inquiry finds - 3/24/2015 - Flight Global

“Modern technology may be capable of reducing crew workload to historic lows and aircraft can now protect themselves as never before, but the requirement for crews to understand and interact with the aircraft and its systems when things deviate from the norm remains as challenging as ever"

regmcfly

Brooks wrote:

When robots can shrug convincingly, that's when we're proper fucked.

This will be shortly after they manage to legitimately shit the bed.

beano

Gonna post this again

IanHamlett

Some parts of automating flight seem easier than cars, because of the lack of things to bump into at 10,000ft, but the stakes are higher.

I see no insurmountable technical obstacles to automating flights though. Obviously there will need to be more sensors to cover for the lack of pilot eyeballs and such and the pre-flight procedure will probably quadruple in size.

Getting past public perceptions and pilot unions might be a bigger issue but, again, all obstacles will crumble in the face of our new overlords. Like in Beano's vid there.

Skerret

Lord_Griff wrote:

Does my chauffeur count as a "driverless" car?

Hugh Abbott: I want a new driver. Get me a new driver. I don't wanna see this guy ever again.

Glenn Cullen: On what grounds?

Hugh: Smiling! Inappropriate smiling! And smirking! Smiling and smirking! I don't wanna see that smile or smirk ever again, ok?

Hulka T

This is what robot driving may look like.

Elmlea

It's not a lack of pilot eyeballs, it's a lack of pilot brains you need to cover for. I've said it a couple of times, but it's the outrageous number of options and "what ifs" in an emergency that make it so different from driving.

Cruise flying is already effectively automated with modern autopilots because as you say, there's less stuff to hit.

dynamiteReady

http://www.baesystems.com/magazine/BAES_051920/look-no-hands?_adf.ctrl-state=ffwns6hy2_4&_afrLoop=1494514956185000&_afrWindowMode=0&_afrWindowId=null#!@@?_afrWindowId=null&_afrLoop=1494514956185000&_afrWindowMode=0&_adf.ctrl-state=1b45xgrce5_4

Howdy, Stranger!

Categories