Toward Autonomous Unmanned Aircraft Technology 137
coondoggie writes with a NetworkWorld piece that begins, "Researchers at Purdue will soon experiment with an unmanned aircraft that pretty much flies itself with little human intervention. The aircraft will use a combination of global-positioning system technology and a guidance system called AttoPilot ... to guide the aerial vehicle to predetermined points. Researchers can be stationed off-site to monitor the aircraft and control its movements remotely. AttoPilot was installed in the aircraft early this year, and testing will begin in the spring, researchers said."
Niche operation perhaps... (Score:4, Informative)
Migration to UAVs is an obligate journey. My last visit to Creech AFB [utah.edu] showed just how inevitable this is, yet I wonder if the move towards autonomous vehicles will really expand beyond a limited niche. Autonomous vehicles have a definite role, but one that is limited to very specialized circumstances akin to interplanetary probes. Platforms that gather data on say climate change or sea conditions are appropriate. However, in the absence of a complete revolution in the way data is gathered through sensors, large event surveillance, crowd and traffic control and hostage situations or crimes (or military applications) will almost always have to have at least a semi-autonomous component to them. I will say that efforts are already underway in certain combat situations to provide for single pilot control over multiple UAV platforms through semi-automated solutions, but those solutions still have an operator actively monitoring the platform.
Re:Niche operation perhaps... (Score:5, Insightful)
For example, UAV's could be used for fire suppression applications - Whenever there are forest fires (or even the threat of), UAV's could be prepositioned and in the air in a matter of minutes.
Crop Dusting - UAV's could perform this function with better precision, for longer hours.
Post/Parcel delivery
Search and Rescue
All these applications are prone to pilot fatigue and are dangerous commercial applications - I am sure there are many, many more applications where UAV's would make more sense and improve aviation safety.
Re: (Score:2)
I can see use beyond military applications.
But we need to do something besides crop dusting. Look at the deadzone in the Gulf of Mexico if you have any doubt.
Postal/parcel delivery of perhaps the most important things, like organ transplants or something. Otherwise the use is over the top.
Search and rescue, maybe. Would Steve Fossett have been found any sooner?
I know that from a defense standpoint, these things have to be developed. On the other hand, one less military weapon would do the world some good.
Re: (Score:1)
But we need to do something besides crop dusting. Look at the deadzone in the Gulf of Mexico if you have any doubt.
Because of high concentrations of pesticides - If you could spray more frequently, you would not need the high concentrations, therefore less run off... A UAV could be as cheap to operate as any other farm implement without needing specialized training to operate.
Postal/parcel delivery of perhaps the most important things, like organ transplants or something. Otherwise the use is over the top.
Have you ever tried to get next-day delivery to a rural area? There are many parts of the country still that are over 200-300 miles from any major airport. Private pilots are often contracted by the USPS to ferry mail back and forth to these areas.
Re: (Score:1)
Re:Nice operation perhaps... (Score:2)
Roger that.
We have to find someone who can not only fly this plane, but who didn't have fish for dinner.
Re: (Score:2, Informative)
I don't think the Air Force wants autonomous UAVs. They are just now starting a UAV pilot program, which has publicly announced last fall.
http://www.airforcetimes.com/news/2008/10/airforce_uav_volunteers_100708w/
Do you really think they are going to scrap their new UAV pilot program in its infancy to be replaced by robots? I don't think so.
This really will only have commercial applications, for non-living cargo. People are already scared enough of flying, let alone with Skynet in the cockpit.
Re: (Score:2)
The US Air Force has autonomous UAVs so your point kinda falls flat.
Global Hawk can fly to Afghanistan, perform its mission, fly home, and freakin' parallel park itself after the operator pushes the "Go take pictures of Afghanistan" button.
Re:Niche operation perhaps... (Score:4, Insightful)
If you listen really carefully, you can hear the sound of Jeff Bezos rubbing his hands together!
Re: (Score:2)
The USAF flies an assload of non-combat aircraft. Cargo and tankers.
Cargo: Take this load of stuff from here to there.
Tankers: Fly a predetermined racetrack in this area, and come home.
Yes, there are times when those missions need to deviate. But that might happen on any mission...fighter, cargo, tanker, or otherwise.
An AI that can take a Global Hawk from Edwards, CA to Oz can fly a typical tanker mission. (until things go weird, an
Re: (Score:3, Informative)
While it certainly isn't the easiest thing in the world to do, with processing speed and efficiency increasing, as well as things like modern GPS and other sensors, UAVs really are easy enough for your average electronics/computer geek to build, given around $1000 and some free time.
Re: (Score:3, Insightful)
An autonomous vehicle doesn't need to do absolutely everything autonomously. It simply CAN fly its mission autonomously.
That means, you can have it signal you when it thinks something interesting is happening, or when it's in an interesting area, and you can start paying attention to it.
I don't think that makes these vehicles any less "autonomous".
Re: (Score:2)
but one that is limited to very specialized circumstances akin to interplanetary probes.
You want to take a kilo from here to there? Perfect.
Re:Niche operation perhaps... (Score:5, Insightful)
I dunno. One of the things I quite like about playing video games is that nobody actually dies.
Re:Niche operation perhaps... (Score:4, Funny)
One of the things I quite like about playing video games is that nobody actually dies.
They do if you force-quit the application before they have a chance to respawn, like I do.
obligatory (Score:2, Funny)
Re: (Score:1)
Why is this obligatory? Besides, any AI-type system wouldn't want to deal with a language as ambiguous as English.
Re: (Score:1, Offtopic)
When I read the summary, all I could think of was the blow-up pilot from Airplane. I think his name was Otto.
Re: (Score:2)
I was thinking of that thing too... thought his name was just 'Autopilot'.
But "Attopilot"? They missed an opportunity for something much funnier...
Re: (Score:1)
But "Attopilot"? They missed an opportunity for something much funnier...
No, he's just a really, really, really tiny pilot!
Hasn't this been done before? (Score:3, Informative)
Ring Ring! (Score:5, Interesting)
All large commercial aircraft come equipped with automatic pilots which can land the plane in an emergency. Taking off again is largely just an exercise in FAA regulations and the proper engineering. (IOW, because there's little demand for the feature, and the FAA doesn't require it, Boeing, et al, have not implemented it.
Re:Ring Ring! (Score:5, Insightful)
Re:Ring Ring! (Score:5, Funny)
Even a flock of geese can put a plane in the Hudson, how hard would it be for a computer?
Re:Ring Ring! (Score:4, Insightful)
The crux wasn't landing in the river, it was deciding to land in the river. Even if remote pilots were on standby to "jump in" and take over in emergencies, there was no time to gain situational awareness.
Granted, in the long run, computers might have more general intelligence than people and be more trustworthy in making these multi-faceted decisions, but I think that will be a long time coming.
Re: (Score:3, Interesting)
Indeed, but whilst the such accidents are extremely rare, CFIT (the pilot flies a perfectly-working airplane into the ground) ones not. In fact, they are one of the most common causes of serious accidents & loss of life.
http://www.flightsafety.org/cfit1.html [flightsafety.org]
Presumably, UAVs would not have this problem...well, maybe not...
Re: (Score:2)
Re:Ring Ring! (Score:5, Informative)
What you're referring to is Category III autolanding, which in CAT III C has no decision height but instead the aircraft can land completely on its own (30 m in CAT III A and 15 m in B, IIRC). More landings are done that way than not - and all landings if the weather is bad since autopilots do a much better job than humans. Now it is obviously necessary that the airport is equipped with that capability so saying that it is for emergency use is a bit of a stretch since in an emergency you might have to land wherever you can (such as on a river...) - or maybe improvise to get it to the runway despite some techincal malfunction (who needs hydraulics when you can vary thrust?). However, Airbus have begun investigating the possibilities to create a "hijack button", which pilots could press in case of a hijacking and then the aircraft would automagically set its transpoder appropriately, notify ATC and land at the nearest CAT III C runway regardless of what is done with the flight controls since then ("sorry Mr. Terrorist, it's out of our hands now"). AFAIK no aircraft currently in service could, however, be equipped with that without some substantial changes (well, perhaps the A380 could, since it's not only FBW but also power-by-wire).
Re: (Score:2)
And since we are talking about UAVs I should note that autoland assumes that somebody can make sure the runway is clear, either the pilot or the tower controller. If you want your UAV to land on a footpath or road then you have a different problem.
Re: (Score:1, Informative)
Planes are also frequently diverted due to a cranky passenger. Any means of calling the landing off would make such a system useless since the hijackers would kill passengers until it was implemented.
Re: (Score:2)
The passengers are dead regardless (or at least would be put into a position of having to overpower the terrorists), I very much doubt that a bunch of Jihadists, when confronted with the news that the aircraft was on uninterpretable auto-pilot, would just resign to their fate, sit back, eat some peanuts and watch the latest infidel movie from Hollywood.
At this point (as mu
Re: (Score:2)
curious...
You think Category III autopilot can land a plan without power?
Re: (Score:1, Informative)
You think a human pilot can land a plane without power?
All modern aircraft have RATs - i.e. ram air turbines, which deploy automatically in case of a power loss and generate power from the airstream (in the brief period of time the deployment takes, batteries are used).
Re: (Score:2)
These statements are not quite correct, and reflect a lack of actual experience with such approaches. Generally, there are both visibility and ceiling requirements set for such approach
Re: (Score:3, Interesting)
One question that comes to my mind: could advanced autopilot tech lead to more ubiquitous personal aircraft?
I don't really know anything about it, but I've always assumed that one of the big hurdles preventing us from having "flying cars" (by which I don't necessarily mean an actual car, but something lots of individuals could buy and fly under casual circumstances) is the difficulty of learning to fly safely. If you could program a destination and have the entire trip flown by an autopilot, from takeoff
Re: (Score:2)
Count the number of disabled cars you see along the road during your typical commute. If they were flying when whatever it was went wrong, just pulling over to the side of the road wouldn't be sufficient. It is sufficient if existing cars just have a few nines of reliability. Once you start flying you need a lot more nines or "Oops, I forgot to buy gas." could be fatal.
One other big difference that further aggravates the above is the duty cycle required of engines and such once you start flying. A typic
Re: (Score:1, Informative)
Spot on. If you think about the size of the support crew for an aircraft, you'll find that it's much more than just the pilot and a mechanic. You have a very strict check regimen, etc etc, all things that are much more critical but harder to do than with a car. People already skimp on oil changes, burned out lightbulbs, etc -- they won't understand why a bent pitot tube is a big deal.
Re:Ring Ring! (Score:5, Interesting)
I don't really know anything about it, but I've always assumed that one of the big hurdles preventing us from having "flying cars" (by which I don't necessarily mean an actual car, but something lots of individuals could buy and fly under casual circumstances) is the difficulty of learning to fly safely. If you could program a destination and have the entire trip flown by an autopilot, from takeoff to landing, would that help the situation?
Possibly. Private aviation is, in many ways, a hopelessly archaic niche caused by the combination of limited applicability, high maintenance cost, hungry lawyers eager to make aircraft owners out to be "fat cats" (class warfare) and comprehensive government regulation.
Cessna tried to make aircraft ownership approachable to the "average Joe" by making the airplanes seem like cars, with yokes that look like steering wheels, and so on. And while Cessna has done (and still does) well as a company, they didn't quite get to the average Joe.
Personal aircraft are quite neat - they make medium-range trips (up to around 500 miles) into day trips. Just today, I flew 3.9 hours in a Cessna Skyhawk to replace some 10 hours of driving! No traffic, the flight is much more relaxing, and much more fun to boot! You go when you want to. You land at a small, local airport rather than get sheep-herded through endless checkpoints taking your shoes off. And you can take your ratchet screwdriver or coffee cup with you - no questions asked!
Heck, you can drive your car right out to the plane to throw your luggage into the back!
But there are some basic disadvantages to aircraft:
1) You'd never take one down to the local Starbux.
2) They use a special fuel that's usually more expensive than normal car gas.
3) Perception: although they have a safety record that's roughly on par with automobiles for traveling, people tend to have strange pictures about what happens when a plane engine dies. "We're going down" is the usual picture, along with a plunging descent that's pretty much guaranteed to kill everybody on board - virtually nothing could be further from the truth. True, when your motor goes kaput, you are going to have to land pretty soon. But it's a controlled landing, as your plane is now a glider. And about 9/10 "forced landings" result in no fatalities or serious injuries whatsoever. Try explaining that to somebody, sometime.
4) They are generally too bulky to store conveniently, especially in your garage.
Could it help? Probably. It might make aviation appeal more to the "average Joe", or to at least more people. But doing so would detract from the immersive quality of flying - that feeling of freedom where you can go left, or right, or whatever, because you want to... if your GPS based solution were implemented, I'd sure want the ability to turn it OFF every now and then..
Re: (Score:2)
Re: (Score:2)
The pilot doesn't put enough fuel on for the flying they want to do....The pilot flies into thunderstorms, freezing rain, etc....The pilot doesn't maintain the aircraft or magic autopilot system...
I guess I was imagining that a sufficiently intelligent system might be able to account for some thing. For example, when you set the destination in your autopilot, it calculates a flight plan, calculates the needed fuel (plus some extra for safety), and checks weather reports for possible problems. Perhaps it also checks some self-diagnostics and has records of a maintenance schedule and won't fly unless the right conditions are met.
And, of course, engine failures requiring off-airport landings. The recent USAir landing in the Hudson illustrates where a skilled human is most valuable.
Perhaps in those cases you could still have the autopilot guide the plan
Re: (Score:2)
The harder problem is of the kind that bedevils game designers and ot
Re: (Score:2)
We still don't have a good handle on how to give computers 'good judgement',
I guess part of my point is that some of the problems you mention might be simpler than real AI problems. Weather, for example, might be a variable that could be handled by having a human look at a weather report and designate a risk level for a given span of airspace. The AI could connect to the service, download a map, read that the route it's about to calculate runs straight through a weather advisory, and tell you, "Nope. We can't go today. Maybe if you wait a few hours, we'll see if the weather imp
Re:Ring Ring! (Score:4, Insightful)
Here [salon.com] is a pilot complaining about the 'modern airplanes almost fly themselves' myth. (You'll need to wait for some ads before the page loads.) And here [salon.com] he talks about the sort of training which produces people able to land a crippled plane on a river instead of a skyscraper.
The military have autonomous planes, but this isn't really relevant to airliners. The military will (if it has to) accept a crash every thousand flights. The airline industry won't accept a crash every million.
Re: (Score:2, Interesting)
Yeah. Embry-Riddle Aeronautical University has had these projects going on for a few years. Pretty badass planes, too - all carbon-fiber.
Plug a few waypoints in, recognize a few targets and snap pictures. There have been competitions for these for a while. What's new with Purdue?
Re: (Score:2)
Re: (Score:2)
Just make sure you select the correct home! [strategypage.com]
Re: (Score:2)
Aren't current UAVs capable of flying from waypoint to waypoint with little human intervention. Call me back when they're capable of landing in a crowded urban area autonomously, then taking off again.
Um, that's not in the mission profile of any UAV that I've seen. 90% of the mission profiles for UAVs are take off from this airport/base point, fly to x location, and then cycle/wait for y event or until fuel runs low. When fuel runs low, it comes to base and parks exactly in its spot to be refueled. (Some of
Quick! (Score:1, Offtopic)
Everyone start making excessive Skynet jokes!
Re: (Score:1)
Re: (Score:2)
Everyone start making excessive Skynet jokes!
I like how this is 'off-topic' but 'skynet' is one of the tags.
I wouldn't mind but Terminator wasn't even a cautionary tale. Mentioning it when anything is automated IS NOT FUNNY.
Re: (Score:2)
the terminator seems like a cautionary tale to me. i know to steer clear of time travel cyborgs.
This isn't so awesome, I could do it. (Score:5, Insightful)
Hi because I go to a university so I can buy any off the shelf RC aircraft autopilot, throw it in a prebuilt airplane, throw it in the air and get school credits!
Here's another brand of autopilot.
http://www.u-nav.com/ [u-nav.com]
Here's a ton of videos of it being used in
http://www.u-nav.com/gallery.html [u-nav.com]
I'm a high school dropout who is perfectly capable of doing this. Yawwwn. Try doing something I can't do, like contributing code to an OSS autopilot package.
http://autopilot.sourceforge.net/ [sourceforge.net]
I'm sorry mods, slash... I just felt this story was too stupid for myself, therefore it must have been too stupid for the general /./pub Please do not mistake my cynical writing as flames. This story should be modded as
-1 unimpressive
Why is this new? (Score:2)
This is called a cruise-missile.
The fore runner of a cruise-missile was Kamikaze of WWII. 9/11 planes were the largest. Then again a suicide-boomer is a walking form.
Now we are making planes smart, to duplicate the ability of cruise-missile.
So when will it get a warhead?
Re: (Score:2)
Re: (Score:2)
Think "Terrorist".
Re: (Score:2)
I still don't see how the children are involved...
Re: (Score:2)
The fore runner of a cruise-missile was Kamikaze of WWII.
The V-1 buzz bomb [wikipedia.org] was a real WWII cruise missile.
Re: (Score:2)
"The fore runner of a cruise-missile was Kamikaze of WWII."
No, it was the German V-1.
The Japanese Oka version used people because they were cheap and available.
Re: (Score:2)
Re: (Score:2)
Now we are making planes smart, to duplicate the ability of cruise-missile.
So when will it get a warhead?
Who needs a warhead when the plane has nearly-full tanks of jet fuel and enough velocity and hot parts to light it off when it hits?
The 9/11 crews proved this. You can see the second twin-towers plane bank at the last moment so the fuel-filled wings puncture, flood, and light several consecutive stories - and the holes show the first one did the same. This lit enough stories of fire (and provided air h
Let me sum this up for all of you. (Score:1)
It will never land by itself.
Yeah, I said 'never'.
Re: (Score:1)
Re: (Score:2, Informative)
It will never land by itself. Yeah, I said 'never'.
Why not? The Global Hawk already does.
Re: (Score:2)
yawn (Score:2)
this has been done so many times its not funny - the problem with wide fielding outside of the military context is getting the FAA to let autonomous vehicles fly in national controlled airspace - its unlikely to happen anytime soon until such a system can exhibit all the robustness necessary to prevent a robot airplane from auguring into someones house or a school or a church or....
Re: (Score:1)
Re: (Score:1)
Re: (Score:2)
When the aircraft is no longer flying...i.e. not enough thrust to maintain lift, or completely broken flight controls, it matters little who in theoretical control, a human or a silicon chip. At that point, Sir Isaac is flying the aircraft, and it WILL land where he wants.
The pilot in San Diego was trying to nurse it enough to get to the canyon. Didn't make it. There was nothing more he could do. Could silicon have made a better
Humans like Human Pilots (Score:1)
The real question is, will people willingly get on an aeroplane that does not have a human pilot on board? I suspect that it will be hard to convince most people to do this.
Re: (Score:1)
Re: (Score:1)
Re: (Score:2)
I can think of at least one good reason [wikipedia.org] to like human pilots. Would you rather have a computer trying to pull that off or a human being?
Re: (Score:2, Informative)
I'm quite sure that a real autopilot would have enough emergency landing routines to pull it off quite good too.
Re: (Score:2)
Well, by that definition, there is not a 'real' autopilot in the world. Did you even bother to READ the wikipedia article you posted? Let me summarize the stats for you:
15-Jan-2009: USAir 1549 - 0% fatal
06
Re: (Score:1)
A suitable computer could eventually execute such maneuvers more reliably.
I'm more concerned about other [wikipedia.org] possible problems and computers not doing [wikipedia.org] what they ought to.
Re: (Score:1)
I'm working a very fast cargo transport UAV for DoD that, after cargo delivery to the soldiers, has enough capacity to hold several people. It will have no pilot or joystick, rather flying blind like a cruise missile to its next waypoint.
Now tell me that if you are combat wounded or reacting to a poisonous snake bite, you'd rather wait for medical treatment via the next piloted lift dispatched once/or if it's safe enough for them? I predict MASH-style evacuations will happen sooner than later.
Global Hawk? (Score:1)
ahhh ... Global Hawk anybody? http://en.wikipedia.org/wiki/RQ-4_Global_Hawk [wikipedia.org]
"The Global Hawk is the first UAV to be certified by the FAA to file its own flight plans and use civilian air corridors in the United States with no advance notice.[21] This potentially paves the way for a revolution in unmanned flight, including that of remotely piloted cargo or passenger airliners."
and
"On April 24, 2001 a Global Hawk flew non-stop from Edwards in the US to RAAF Base Edinburgh in Australia, making history by being
Palindromic Acronyms (Score:2, Interesting)
I just have to point out that the acronym for the title of this post is a palindrome.
That is all.
Re: (Score:1)
Re: (Score:1)
Cruise missile tech (Score:2)
The 80's called and want their GPS-guided cruise missile technology back or at least acknowledged
Where have we seen this before? (Score:1)
Oh yeah, here [wikipedia.org]
And we all know how that turned out. Err, maybe not no one in their right mind watched that movie
Re: (Score:1)
What is in a name...? (Score:2)
If I were the decision maker to choose the name, I would have named it Otto [wired.com].
Global Hawk just uses a keyboard (Score:2, Informative)
As was mentioned above (I tried to reply to that one but web page errors wouldn't allow) the RQ-4 does this and more.
Unlike the well known Predator UAV, the Global Hawk control panel has no joystick or similar control. It's got a keyboard and mouse.
If you want it to turn left, you type/click commands to alter it's course etc.
What I think is particularly interesting is that it has a set of commands to follow if it loses communications with the humans. So if on the trip to Australia comms had been lost part
So how does it handle ... (Score:2)
... collision avoidance? And air traffic control interface?
Taking off, flying GPS waypoints, and landing, are a lot easier if you're the only thing in the sky and can do anything you want.
Re: (Score:2)
Oh, wait, sorry, that's Ethernet.
Great! (Score:2)
Give the terminators wings. Then we'll be all ready for SkyNet.
I for one welcome our new robotic (literally) overlords.
A more fitting name (Score:1)
Liability insurance? (Score:1)
Re: (Score:2)
Let the GPS fail, the inertial navigation system will take over and navigate back to base. No big deal. Heck I did this years ago in school.
But that would require knowing something about aircraft and UAVs and not screaming about how the sky is falling.
Re: (Score:1)
Re: (Score:2)
General failure on a UAV that is big enough to crash into a house and kill people would require one of the following:
1) Failure of redundant navigation systems (GPS/INS/photogrammetric navigation)
2) Failure of redundant power supplied to aforementioned navigation systems
3) Failure of redundant computers storing course data and mission planning
Generally, if any one of these has occurred, its because the aircraft has been shot at and is falling to the ground in several pieces already. "When a plane fails" is
Home made UAVs (Score:2)
What's perhaps more interesting is that nowadays even hobbyists can build UAVs as an extension of the radio controlled aircraft hobby.
Stuff like that in the article can be built by even hobbyists thanks to the few open source autopilot projects out there and the decreasing price and shrinking size and weight of digital cameras, wireless technology and other relevant components. I'm mostly interested in swarm robotics at the moment, but I must admit I'm somewhat tempted at having a go at building one myself,
Nothing new here... (Score:1)
We'll need a management system (Score:1)
Small unmanned UAVs have been around for years (Score:1)
A Canadian company MicroPilot [micropilot.com] has been sending small unmanned aircraft into the skies for years. Attopilot is nothing new.
1 warhead away (Score:2)
Skynet? (Score:1)
"Atto-Pilot"? (Score:1)
Am I the only one to read this as "Atta-Pilot"?
Kinda scary if you ask me (of course, nobody did...)