Automagic No-Fly-Zone Enforcement 536
An anonymous reader writes "SoftWalls is the name of an aviation project at UC-Berkeley that's developing a system for commercial airliners that establishes and enforces no-fly zones. Basically, through GPS, if a plane begins to enter a no-fly zone (eg, around a mountain, or over Lower Manhattan), an alarm goes off in the cockpit. If ignored, the system actively removes control of the plane away from the pilot and co-pilot to steer the plane out of the no-fly zone. The technology is intended as both an accident prevention technique and a deterrent to terrorists planning to ram a building. ABCNews recently profiled the project (with video) and also rode along with a working prototype built by Honeywell that successfully kept a Beechcraft from hitting a mountain."
My guess (Score:1, Insightful)
Please (Score:5, Insightful)
Why do people seem to think that terrorists are just dumb camel jockeys from the middle of the desert who are easily impressed by internal plumbing? If an al Qaeda operative wants to smash a plane into a building, he'll figure out a way to disable such a system.
The lower Manhattan nightmare scenerio (Score:5, Insightful)
So someone says "Oh, there will be an override for situations like that" -- well, why won't that override get used when someone is bound and detmined to fly a 757 into a tall building? At that point its just another warning system, which is fine, but the computer control part scares me. I like pilots in control when necessary.
Traditional Boeing vs. Airbus debate (Score:1, Insightful)
Where is aviation headed?
Re:Situation... (Score:2, Insightful)
The pilot has time to respond to the warning. During this time, he is fully in control of the plane. If he heads back out, he maintains control of the plane. If he does not head out, he is assumed to be incapable of operating the plane and is relieved of duty by the automation software.
Just like with any security issue, assuming that the end-user is in complete control of the machine at every time is a mistake. Grandma is not in control of her new Windows XP box. Joe P. Capitain is not in control of his plane when there is a gun to his head.
This is a Good Idea--let's hope the implementors make it a good reality.
PIlot discretion (Score:2, Insightful)
I think there are simply too many "what-if" situations that require a pilot have control over the aircraft to allow such critical remote control. What if the jet runs out of fuel? What if the no-flyover beacon directs the jet into other air traffic or really bad weather.
Moreover, what would stop a private citizen from enabling his or her own no-flyover beacon and causing havoc: From terrorists all the way to folks living next to an airport who deal with turbine noise.
A good idea at first, but with reflection seems to cause more problems than it solves.
-Alex
Re:The real question is ... (Score:5, Insightful)
Because it's a stupid idea.
A computer can't know the myriad of decision factors that the pilot does. There's a lot more to safe flight operations, especially during an emergency, than simply pointing the airplane in the right direction. There's considerations of how bad the weather may be in one direction, which engine may have failed causing difficulty in turning in one direction, which heading the aircraft needs to be on after completing a turn to line up with the desired runway so as to make a landing the first time (because there might not be a second time), how critical is it to get to a certain altitude rapidly, and hundreds of other factors that might (or might not) be important. The pilot is the one best to rapidly rank order what's important and what's not given the particular situation he's found himself and those several hundred passengers in.
Just what computer program could decide that it's a better decision to allow the plane to fly close to a mountain (how close? 1000 feet? 2000 feet? 50 feet?) in order to meet some criteria the pilot (you know, the person deemed competent to make such decisions) has decided is most critical.
Aids that assist in flying are wonderful - keep bureauacracy and "for the children" politics out of the cockpit though.
Overriding the pilot is the Wrong Thing (Score:5, Insightful)
The system being discussed here would take ultimate control of the plane away from the pilot. In the century of powered flight we have just completed, such ideas have have always turned out to be the Wrong Thing.
If we could always trust the flight computers and control systems, we wouldn't even need pilots: today's jetliners are smart enough to fly themselves. The problem is that the systems are just not reliable enough, and the system designers are not prescient enough, to handle every eventuality.
For ages, the question has been A modern corollary might be:
Re:Situation... (Score:3, Insightful)
Wrong. This is not a situation where a BSOD is acceptable. Grandma might be appreciative of Clippy helping her out, but the last thing a pilot in an airline full of passengers needs is some "oops" in the programming preventing him from turning when he wants because it might overfly Barbara Streisand's house. The pilot is the one to decide what is smart to do - if he needs help, then give him a partner. If terrorism is the concern, give him an armed guard.
Re:Lone Gunmen (Score:5, Insightful)
I'm all for a plane avoiding mountains, and taking over when the pilot is incapacitated so the plane doesn't crash -- but I hate seeing articles about new technology being promoted with "it'll stop those nasty terrorists! Woohoo!"
I'm sure the (surviving) terrorists are ROFL at us scrambling to prevent them repeating something they know they'll never repeat. We need to harden our other systems -- water/electric supplies, who's driving the oil/gas tankers/trucks, etc.
Technical Solution to a Social Problem (Score:5, Insightful)
Many others have posted great reasons why taking control away from a pilot is a bad thing, so you can read them- but if it's terrorists you're worried about, they now have much more to fear from the passengers than from a computer system. The stakes couldn't be higher now for airline hijackings, and knowing the stakes, no US group of passengers will allow any hijackers to carry out their mission. (Flight 93) This, incidentally, is a social solution to a social problem.
Sure, this kind of thing would be great for terrain avoidance. But I wouldn't bet my life on it. Between jamming, spoofing, misplaced confidence, programming errors and the like, it can be quite problematic.
Basically, you're swapping your trust in the pilot for your trust in the programmer. Not necessarily a good trade.
Ludite reaction or not ? (Score:4, Insightful)
In the end you can't defend against human decision making unless you remove the human from the process.... which means you used canned human decision making in the form of code which to my knowledge is not and cannot ( to date ) be made self-correcting. Thus if there is an unforseen circumstance for the code to encounter you don't know what will happen. The code can't think on the fly for itself. So choose your poison. A plane that will be consistently flown even if that consistency invovles a bug that flys into the ground given the proper circumstances or a pilot that can think for itself and do unthinkable things such as fly into a huge skyscraper, or come up with an inovative way to control a plane with differential thrust due to the failure of control surface hydraulics ( actual real world example ). In fact both of those examples are being subjected to CODE fixes for making such actions easier or more difficult, this being an example of 9/11 ( or mountain ) avoidence and the new implementation of a backup directional control system utilizing dissimilar engine thrust rates. But its impossible to account for all scenarios and untill code can be sufficiently capable to deal with unforseen circumstances you have an overide. You draw your own conclusions on what a pilot will decide to trust in an odd situation when presented with loss of control of the aircraft. If your response to that is not to allow that decision then why the hell do you have a pilot in the cockpit to begin with ?
Poor example, but yes, control w/out limited = bad (Score:3, Insightful)
Yes, but your example is a poor one. Pilots have a multitude of options at their disposal for avoiding collisions. Altitude changes(up OR down! Wow!), heading changes(left OR right!) and speed changes(faster OR slower!)
The real problem is that in almost every plane with an autopilot, there's a Big Red Switch the pilot can press. When I saw this in action, it was on a small(4 seater) single, and pressing the switch caused about 2-3 switches to solidly trip to the off position(think like a circuit breaker) and a loud warning tone. It completely cut the autopilot's control over the plane, and not by software- hardware. Furthermore, guess what's part of the checklist? Setting the autopilot while on the ground, making sure it can manipulate control surfaces cleanly in all directions, and then pressing the Big Red Switch and verifying the AP is dead.
The problem will not, I predict, come from legitimate restricted airspace; restricted airspace is often near legitimate popular routes, but not to the point of concern(and most restricted airspace has ceilings, rarely is airspace restricted to the ceiling airliners cruise at). The problem will come in the following forms:
What about GA? (Score:2, Insightful)
Re:Situation... (Score:3, Insightful)
On the early Airbus flight computers, you selected which mode the computer was to be in at any one time, ie cruise, takeoff, landing, manual operation, generic flight. If the computer was in the wrong mode to what you actually were doing, it would react differently to input from the flight controls. In this case, the flight mode was set as landing and the pilot is stated as saying that he increased the throttle and the engines did not respond. In the mode it was, the computer had no reason to increase the engine power, as it thought it was landing, and to increase the power while the emergancy breakers had not been tripped would be dangerous to the aircraft.
This was a common compliant with early Airbus pilots, that if given a go around signal while landing, they did not gain engine power until the computer was placed into a different mode of flight. So with this accident, the aircraft was in the mode for landing, and that is exactly what the computer was going to do, regardless of what the pilot wanted to do.
That the pilot and flight crew were sentanced for a criminal act is appalling, given that up until that crash, there were many complaints of Airbus flight computers misbehaving, when expected to do otherwise, being passed back to the company. It is heartening to know that Airbus redesigned the flight computers shortly after this crash to include less flight modes, and greater ability for the computer to disregard the current flight mode. This has led to less events where the pilot has failed to alter the flight mode and caused confusion.
Can You Say GPS jammer? (Score:5, Insightful)
Re:So, basically... (Score:2, Insightful)
They do this already... I think they call it a pilot.
Re:The real question is ... (Score:5, Insightful)
Yeah, I guess 3000 hours of flight time with 500 hours in combat including more emergencies from engine failures due to fire to hydraulic failures to electrical failures to lightning strikes to ... than I can count doesn't mean shit on slashdot.
How do you think the pilot knows of bad weather ... oh thats right he gets ground reports through his computer
Well, he could always look out the window ... perhaps a difficult concept to comprehend for those that believe that the pilot should only do what's been preallowed by some programmer who is completely unaware of the particulars of the situation that may occur.
Further more how do you know which engine is out without the computer
There's a significant change in thrust from one side of the aircraft when an engine has failed. Turning into a dead engine, especially with a heavy fuel load, is usually not the smart thing to do because of the greater difficulty of turning back - but if the pilot makes a concious decision to turn in a particular direction then a computer should not attempt to override it for reasons that are less critical than safety of flight (such as some BS no-fly zone).
99% of the time you are in the air the FMC is guiding the plane including *gasp* your headings out of turns
Sigh. An autopilot system is an aid for the pilot - it's not a substitute. A system that calculates the headings and lead points is fine for rolling out on a certain course, but is irrelevant to the situation of a pilot flying the aircraft where he wants it to be, vice some erroneous decision made by software.
FMC is going to be able to calculate out how fast you can and can't climb a hell of alot better than any human
No it can't. It can calulate the optimum climb rate to get to some altitude using the least fuel or least time or least distance - but that is not definitely better - the system doesn't know what may be most important at any given moment. That's the job of the pilot. People that advocate being able to hinder the pilot's options are doing no favors to either the pilots or the passengers they are responsible for.
Responsibility for the lives of hundreds of others is a big deal. Maybe you've never been there.
Re:The real question is ... (Score:2, Insightful)
Once someone (or some thing in this case) takes control over the plane, they also take responsibilty for it. The first time a plane goes down and one of these things doesn't stop it (or worse, it causes it), the company that was dumb enough to make it is screwed.
Of course the fun starts now because we get to sue the airlines for _not_ having these "life saving devices" the next time a plane flies into a mountain.
Re:The real question is ... (Score:5, Insightful)
As for the weather I full understand looking out the window is very important. but can you really tell that much about which way the wind is going when your 200 miles off? If your in a bad weather cell that different. (but at no point in my post did I say the pilot should not be able to fly his plane)
As for the signifacant changes due to an engine lose
See I understand in emeregence situations autopilot is not a substitute for a pilot. At the bottom of my post I said I hate the idea of taking the control out of the pilots hands. I also freely admit computers are not perfect and that why pilots routinely change setting in the FMC to make it do what the pilot wants
I will give you the point that the computer may not make the right descion and there should be a way to take control away from the computer. You also have to admit that there are times where the pilot wont make the best descion (see the crash in swiss airspace where the pilot went against his TCAS and people died because of it).
I ask you to find one thing in my post where I said the pilot should not have the ability to break away. Hell I know computers aren't perfect and I agree a 100% that taking control away is a bad idea. I just disagreed with your reasons because they weren't applicable to current commercial jetliners.
As for responsibility of hundreds of lifes, your right I have not been up there with the lives in my hands but before my current job I used to write ATC systems which are mission critical (in production in germany) so I do full understand its not kids play and that when things malfunction you put lives in danger.
I think you had the right idea your reason just where off. And I'm sorry if I offended I was probably a little harsh but I didn't agree ande I need to post on it.
Re:The real question is ... (Score:2, Insightful)
Re:The real question is ... (Score:3, Insightful)
A piolet has to take in a lot of data all at once and put out a lot of data, something a computer isn't complicated enough to do yet. Piolets can feel the plane, the controls have 60+ years of customization just for them. We're talking about engineers who are designing this thing, not piolets.
Not to mention what will happen if the computer breaks during midflight and goes haywire (although a certain amount of redundancy can be built into a device, they still will fail). Or if the goverment decides they want to add in remote flying so they can crash or redirect any plane they feel like. I don't know about you people, MKULTRA + self-pioleting planes + media monopoly = people being abducted to be guinea pigs.
The new cars that are coming out that are control-by-wire running windows CE aren't all that safe to drive. People are getting trapped in their cars because the locks freeze shut and the moter isn't powerful enough to open it or getting locked in when the battery is low. No reason to trust that fly-by-wire planes will be all that much different. I like redundancy built into the system thankyouverymuch. I like my power steering and breaking only to be assisting the steering and breaking, not doing all the stearing or breaking. If the alternator belt breaks wtf do I do then? Crash?
I'd sooner give the piolets 9mm handguns and really good door locks and spending more money on good mechanics and actual metal parts than introducing this kind of non sense to the cockpit.
And finally, don't insult the parent poster. It does little to correct them and all it does is work to break down what little sense of community we have left.
Introducing Reality (Score:2, Insightful)
1. Letting autonomous systems take control has some very very important repercussions.
2. Irrespective of those repercussions it's going to happen more and more throughout our society. The longterm advantages are just too useful.
Some interesting scenario's....
Perhaps a software update that enforces no-fly zones in such a way as to force the automatic systems to crash the plane where it's wanted.
Perhaps a device that transmits to the flight controls information that results in the previous example.
Perhaps an external device that can disrupt or worse, control the onboard systems.
Of course, some such dangers are inherient in any fly by wire system. So a balance between the degree of dificulty in compromising such and the increased accuracy, redundancy, and control that fly by wire allows must be made.
Of course having systems that can't be overridden by the "pilot" (legit or otherwise) on site has it's own dangers. A massive software glitch in the systems that may autonomous control... BAD. A purposeful version of such a glitch. And any other outside interferience.
Re:One word: Bugs (Score:1, Insightful)
The coach I drive has a GPS navigation system in it. It is supposed to aid the driver. However, it often gets lost as to where I am, and which way I'm pointing. [I've realised that it assumes that you're going to drive down roads and not along access roads it doesn't know about, and so assumes the GPS must be in error and puts you on the road it thinks you are travelling along. Which will the plane obey?]