|
The kind of software you guys are talking about isn't the kind of software that will be running future autonomous aircraft. That article I posted a page or two ago, (here it is again in case you missed it: http://www.economist.com/news/science-and-technology/21707187-artificially-intelligent-autopilot-learns-example-flight-response) and the fuzzy AI article Captain Postal posted on this page scratch at the surface of it. Computers will be writing their own operational instructions, developing their own fault tolerance, and do it better than humans. That's the future, but it's already happening now. You software guys aren't going to be checking any code for errors.
|
# ? Sep 28, 2016 02:26 |
|
|
# ? May 26, 2024 17:01 |
|
Murgos posted:Nah. It's trivially easy to spend a couple of hundred million dollars on fault tolerant software for systems with very little complexity at all. The hardware is getting cheaper but by the time you've established the pedigree of your parts and processes the number of zeros involved still gets to quite surprising levels. quote:Computers are good at evaluating problems that can be identified and processed as discrete inputs. Most cases where the word 'fault' comes into play don't fit those categories. Kind of by definition you are dealing with systems not working correctly and not telling you the truth. Computers have hard problems with liars, particularly when they may be the one lying. Just to be clear, I am in no way suggesting that it would be feasible to replace the pilot of an existing airliner with a computer and have it operate safely and reliably. I am saying that if you are designing a new aircraft and want it to be safe and reliable, it doesn't make sense to intentionally include a human pilot. A key point here is the cost savings from not building an aircraft that carries a human. If you already have to carry people that goes out the window. Linedance posted:The kind of software you guys are talking about isn't the kind of software that will be running future autonomous aircraft. That article I posted a page or two ago, (here it is again in case you missed it: http://www.economist.com/news/science-and-technology/21707187-artificially-intelligent-autopilot-learns-example-flight-response) and the fuzzy AI article Captain Postal posted on this page scratch at the surface of it. Computers will be writing their own operational instructions, developing their own fault tolerance, and do it better than humans. That's the future, but it's already happening now. You software guys aren't going to be checking any code for errors.
|
# ? Sep 28, 2016 02:34 |
|
Linedance posted:The kind of software you guys are talking about isn't the kind of software that will be running future autonomous aircraft. That article I posted a page or two ago, (here it is again in case you missed it: http://www.economist.com/news/science-and-technology/21707187-artificially-intelligent-autopilot-learns-example-flight-response) and the fuzzy AI article Captain Postal posted on this page scratch at the surface of it. Computers will be writing their own operational instructions, developing their own fault tolerance, and do it better than humans. That's the future, but it's already happening now. You software guys aren't going to be checking any code for errors. As it happens my specialty is actually machine learning. (I'm sure there are others here with that specialty as well, and likely plenty with more experience than me and if I am mistaken in anything I defer to their wisdom.) I'm familiar with what you're describing. "Writing their own operational instructions" has been happening for a very long time, as well as training machines on human operator input. That said there's a wide gulf between training an optimization algorithm within a constrained space (where ML is now) and having Lt. Data flying your airplane, or debugging programming code. You are going to run into an inherent issue in any learning system trying to deal with situations that are very infrequent. Anyone who works in the cybersecurity or fraud problem domain understands this issue painfully well. You can have millions of rows of data, and only a tiny handful that are "positives" on which the algorithm can learn. Ultimately what you are forced to do is bias your model in some way, such as by resampling (essentially copying or giving extra weight to positive rows). Anyway that sort of thing is generally tuned manually and as far as I know this is unlikely to change in the near future. The other half of this is automated program verification (take a look at NuSMV for an example) which has been around for a while and I know less about though I studied it a bit when I was in school. It's not really new. It also only really works for things that have a small number of discrete possible states, but complex interplay between those states.
|
# ? Sep 28, 2016 02:49 |
|
MrChips posted:Viking Air has done a pretty good job with the Twin Otter; supposedly their order books are full through 2020 at the moment. True, forgot about them. There's an HU-16D that's been derelict at OSU airport in Columbus since at least 2001 that I've been meaning to photograph ever since I found out it was there. Apparently it's tied up in legal wrangling related to some treasure hunter: http://www.dispatch.com/content/stories/local/2015/04/08/Thompson-pleads-guilty-to-contempt-charge.html
|
# ? Sep 28, 2016 03:17 |
|
https://www.youtube.com/watch?v=aYkJhmw3TIo Jaw dropping video of a 777 flying around patchy clouds/fog right around the dewpoint, and interacting with them (and the ones it makes itself) in amazing ways. Awesome visualization of aerodynamics. Wingtip vortices, downwash, low pressure fields...
|
# ? Sep 28, 2016 06:24 |
|
vessbot posted:https://www.youtube.com/watch?v=aYkJhmw3TIo This is ridiculously awesome E: 1:30 onwards from this related video is stunning https://youtu.be/dfY5ZQDzC5s lilbeefer fucked around with this message at 14:10 on Sep 28, 2016 |
# ? Sep 28, 2016 14:03 |
|
Hauldren Collider posted:As it happens my specialty is actually machine learning. (I'm sure there are others here with that specialty as well, and likely plenty with more experience than me and if I am mistaken in anything I defer to their wisdom.) I'm familiar with what you're describing. "Writing their own operational instructions" has been happening for a very long time, as well as training machines on human operator input. That said there's a wide gulf between training an optimization algorithm within a constrained space (where ML is now) and having Lt. Data flying your airplane, or debugging programming code. It's a fascinating topic and I'll admit to layman status here. I read a lot of stuff about AlphaGo, which is pretty mind blowing, and really gets you thinking about the future of learning machines. Right now it's perhaps not something you can put in a mobile platform and have fly around like a conventionally crewed aircraft, moving cargo, performing and acting on surveillance, or combat operations, or what have you, but I can definitely see it being commonplace in the next 20 years, probably less.
|
# ? Sep 28, 2016 14:42 |
|
vessbot posted:https://www.youtube.com/watch?v=aYkJhmw3TIo That's incredible.
|
# ? Sep 28, 2016 14:47 |
|
vessbot posted:https://www.youtube.com/watch?v=aYkJhmw3TIo fickle poofterist posted:1:30 onwards from this related video is stunning e: drat you nickelsack
|
# ? Sep 28, 2016 18:15 |
|
StandardVC10 posted:Talking of which. I love the Albatross but aren't most (or all) on experimental exhibition registrations?
|
# ? Sep 28, 2016 19:14 |
|
ehnus posted:I love the Albatross but aren't most (or all) on experimental exhibition registrations? It varies. I don't know what the status of that one was.
|
# ? Sep 28, 2016 19:47 |
|
Alereon posted:Just to be clear, I am in no way suggesting that it would be feasible to replace the pilot of an existing airliner with a computer and have it operate safely and reliably. I am saying that if you are designing a new aircraft and want it to be safe and reliable, it doesn't make sense to intentionally include a human pilot. A key point here is the cost savings from not building an aircraft that carries a human. If you already have to carry people that goes out the window. There are lots of somewhat autonomous drones out there but the tasks they do are for the most part carefully constrained not to be a danger (or the aircraft is so small as to be not dangerous) and have a minimal cost for being unreliable. Real tasks, that have to be successful or operate in an way that could endanger people and/or incur large financial costs for failure are not currently suitable for drones because drones are not safe and reliable (unless you spend vast amounts of money to make them so).
|
# ? Sep 29, 2016 00:25 |
|
Space vortices best vortices. https://www.youtube.com/watch?v=hQ189a3Wnk4
|
# ? Sep 29, 2016 00:45 |
|
Enourmo posted:Space vortices best vortices. It’s cheating to go 18 times the speed of light.
|
# ? Sep 29, 2016 00:53 |
|
you're drat right it is
|
# ? Sep 29, 2016 01:09 |
|
ehnus posted:I love the Albatross but aren't most (or all) on experimental exhibition registrations? IIRC, the only ones not on Experimental or Restricted registrations are the ex-Chalks birds, all of which are still parked, I think. Going to a normal category requires some significant modifications, and I think the only ones ever done were done as a batch for Chalks.
|
# ? Sep 29, 2016 01:41 |
|
MrYenko posted:IIRC, the only ones not on Experimental or Restricted registrations are the ex-Chalks birds, all of which are still parked, I think. I think I saw a classified ad for a normal category Albatross someplace (with normal category as a big selling point, so it's certainly not common.) I guess they could have been selling one of the Chalks ships.
|
# ? Sep 29, 2016 02:01 |
|
OK more forces of nature of aviation on video. https://www.youtube.com/watch?v=3hTdJzkHE1o Here he managed to capture the disturbances of rain behind airplanes! https://www.youtube.com/watch?v=WizVovGeHKg More wake vortices, this time right down on the runway at night, lit up by the runway lighting. Mesmerizing. https://www.youtube.com/watch?v=fSodzuCwRYI Ultra close up and slo-mo of landing gear touching down. We see the wheels gradually spinning up from the rubbing of the pavement and the smoke decreases, and then every bounce, jiggle and twist. https://www.youtube.com/watch?v=filq88Yvt5I Reverse thrust, up close and personal. https://www.youtube.com/watch?v=5avi26t8Xpo Another reverse thrust on wet runway, from onborad. (This one's from the same guy as all the previous ones.)
|
# ? Sep 29, 2016 02:16 |
|
Some of these were pretty sporty, especially a couple of those turboprops https://youtu.be/Vtyq-8fdCsY?t=1m23s
|
# ? Sep 29, 2016 05:52 |
|
slidebite posted:Some of these were pretty sporty, especially a couple of those turboprops I love how sometimes it looks like they aren't even moving at all, just sinking toward the runway.
|
# ? Sep 29, 2016 06:15 |
|
CharlesM posted:I love how sometimes it looks like they aren't even moving at all, just sinking toward the runway. Or, conversely, *hovering* upward away from it. Reminds me of a time way back when playing FS5 where I had a B-2A at O'Hare and dialed the winds up to like 200MPH and was able to 'aerofoil' it up and down from a standstill.
|
# ? Sep 29, 2016 06:26 |
|
Gotta love long focal length!
|
# ? Sep 29, 2016 06:35 |
|
Those Wizz Air approaches
|
# ? Sep 29, 2016 10:46 |
|
slidebite posted:Some of these were pretty sporty, especially a couple of those turboprops i've watched crosswind landing videos obsessively, but that one got me anxious. it's so zoomed in
|
# ? Sep 29, 2016 12:13 |
|
One of the cool things about living in South Auckland is you might hear a funny noise, look up and see a mosquito flying formation. I think the chase plane might have been doing a flying inspection because they did a couple of circuits of the airport and then landed. Unforunately I don't have quite the same level of camera as the previous guy, just a lovely work cellphone.
|
# ? Sep 29, 2016 20:38 |
|
Same, but living in Southern Ontario near Hamilton One of the two remaining airworthy Lancasters from earlier this summer. I heard the engines with enough time to run and get my camera, what a sight.
|
# ? Sep 29, 2016 20:51 |
|
StandardVC10 posted:I think I saw a classified ad for a normal category Albatross someplace (with normal category as a big selling point, so it's certainly not common.) I guess they could have been selling one of the Chalks ships. Sorry, that's what I meant, but words are hard. A few of the ex-Chalks birds are privately owned now, yes, but its just those aircraft that are normal category.
|
# ? Sep 29, 2016 21:23 |
|
Norwegian F-104 on test flight after restoration, with an F-16 flying chase. They're working on getting a permanent flying permit. Nice video with funny ending. https://www.nrk.no/nordland/filmet-den-unike-starfighter-flygningen-fra-f16-fly-1.13156073
|
# ? Sep 29, 2016 21:34 |
|
Sharing a good post from the OSHA thread of all places, where the topic of discussion was aerial refuelling.SelenicMartian posted:And one day the fuel spills out and all over the plane and catches on fire.
|
# ? Sep 30, 2016 00:32 |
|
PittTheElder posted:Sharing a good post from the OSHA thread of all places, where the topic of discussion was aerial refuelling. That link is worth clicking and a good read.
|
# ? Sep 30, 2016 01:15 |
|
While scrolling past the sports channels I had a moment of excitement, and then deep confusion and disappointment to discover that the Ryder Cup and the Schneider Cup are two very different sporting events.
|
# ? Sep 30, 2016 04:38 |
|
Ardeem posted:While scrolling past the sports channels I had a moment of excitement, and then deep confusion and disappointment to discover that the Ryder Cup and the Schneider Cup are two very different sporting events. I got really confused on why a building supervisor would have a cup named after him...
|
# ? Sep 30, 2016 05:46 |
|
I just found out that Mitsubishi Heavy Industries in cooperation with Fuji Heavy built the first Japanese jet in 50 years:quote:Mitsubishi Aircraft’s first test model of the MRJ90 regional jet is hopscotching its way from Japan to Moses Lake, where it’s expected to arrive Wednesday afternoon to begin a lengthy flight certification process. http://www.seattletimes.com/business/boeing-aerospace/mitsubishi-regional-jet-hopscotching-its-way-to-moses-lake/ And it's making it's way to Moses Lake.
|
# ? Sep 30, 2016 10:56 |
|
The Japanese have built jets recently*. The Kawasaki C-1. That Honda Bizjet. That new transport/ASW warfare aircraft. *Admittedly the C-1 might be 40-50 years old
|
# ? Sep 30, 2016 11:35 |
|
Humphreys posted:I just found out that Mitsubishi Heavy Industries in cooperation with Fuji Heavy built the first Japanese jet in 50 years: It's a pretty exciting time. It has a geared turbo fan engine from Pratt and Whitney. I've gotten a chance to work on the engine in depth and some of the tech is really cool. The reduction gear box for the main fan blades does not need to be serviced for the life of the engine. It's not as apparent as on the A320neo, but it's a surprisingly quiet engine as well.
|
# ? Sep 30, 2016 15:16 |
|
um excuse me posted:It's a pretty exciting time. It has a geared turbo fan engine from Pratt and Whitney. I've gotten a chance to work on the engine in depth and some of the tech is really cool. The reduction gear box for the main fan blades does not need to be serviced for the life of the engine. It's not as apparent as on the A320neo, but it's a surprisingly quiet engine as well. Sales dude: "gearbox doesn't need to be serviced for the life of the engine!" Gearbox: *consumes itself in a fit of pique, reducing it's reduction gears to a thick metallic slurry* Airline: "but you said..." P&W: "the engine must be removed for extensive overhaul now, so technically we're correct"
|
# ? Sep 30, 2016 15:29 |
|
That regional jet segment just gets more and more crowded. These players can't all survive in the market can they?
|
# ? Sep 30, 2016 17:19 |
|
It's not clear Bombardier will survive in the regional market without needing a massive bailout or being bought by someone. After the success of the CRJ, Bombardier essentially spent the next 20 years just stretching the CRJ fuselage to create new models while working on the C Series. Once technical problems began pushing the C Series delivery date back, Bombardier was stuck with the CRJ trying to compete with the more comfortable and efficient Embraer E-jet line, which was also getting orders that might have gone to CS100's if the program had been on schedule. The combination of the CRJ and Q400 both selling poorly (their combined order total is less than 1/3 the E-jet backlog) and the C Series being well behind schedule meant that Bombardier took substantial losses on the C Series (a $3.2 billion writedown in late 2015), and the company has already received a $1 billion bailout from Quebec, tried (and failed) to sell the C Series program to Airbus, and is presenty looking for another $1 billion bailout from the Canadian federal government. Even when Bombardier snagged an order for 75 CS100's (with 50 options) from Delta, they only did so by reportedly offering a 65-70% discount on the airplanes, which means Bombardier is basically breaking even on that order, if not taking a loss on it. That said, Bombardier does have other divisions that are profitable is probably "too big to fail" even in the worst case, so they're probably not going anywhere long term. azflyboy fucked around with this message at 19:54 on Sep 30, 2016 |
# ? Sep 30, 2016 19:47 |
|
slidebite posted:That regional jet segment just gets more and more crowded. These players can't all survive in the market can they? Not without significant propping up by national governments... *ahem ^^^^*
|
# ? Sep 30, 2016 19:52 |
|
|
# ? May 26, 2024 17:01 |
|
slidebite posted:That regional jet segment just gets more and more crowded. These players can't all survive in the market can they? I'm sort of expecting the Embraer E2 to just crush everyone, except where the national interest requires you to buy from the home team.
|
# ? Sep 30, 2016 20:09 |