Australian Defence Force Builds $1.7m Linux-Based Flight Simulator 232
scrubl writes "The Australian Defence Force (ADF) has revealed its latest flight simulator runs on SUSE Linux-based clusters of Opteron servers and uses an open source graphics platform. The Defence Science and Technology Organisation's (DSTO) Air Operations Simulation Centre in Melbourne creates virtual worlds that allow pilots to experience real-world combat situations without leaving the ground. The visuals software was written in OpenGL, using commercial and open source scene graph engines and making 'heavy use of OpenGL Shader Language programs.'"
Re:I want one! (Score:5, Informative)
I want one! Where can I get myself a sweet flight sim like this?! :-o
In the Australian Defense Force?
Re:I want one! (Score:4, Informative)
Even their website is defence.gov.au [defence.gov.au]...
Re:Not really news. (Score:4, Informative)
Very cool, thanks! I was really impressed until you said "1GHz ethernet". That seems... unlikely =D
He was probably mixing up his terms when referring to gigabit ether [wikipedia.org]. It's not the fastest thing on the block, but it's still pretty nippy (and definitely beats what most people have deployed to desktop level) and the faster options (notably Infiniband) tend to only be used in specialist applications like tightly-coupled supercomputers.
Re:$1.7m is dirt cheap! but missing something (Score:4, Informative)
Re:Kangaroos (Score:3, Informative)
Prior art [imdb.com] I think. Though the Snopes article [snopes.com] is funnier.
Re:Simulating what, exactly? (Score:4, Informative)
Actually we'll be retiring the F-111 next year. We will have a mix of Hornets, Super Hornets and JSF for some time, though.
Re:It's pretty much a given that they saved money (Score:2, Informative)
Spend millions of dollars on a project, and do stupid things like cut corners that save you statistically irrelevant amounts of money on the project and result in a far more difficult to support product.
Agreed that in a project that size the direct cost of the operating system will be relatively small.
But there are many indirect costs resulting from the choice of operating system. There may be better or less expensive development tools available for Linux versus Windows. There may be more or better or less expensive graphics/rendering libraries and other software available for Linux as opposed to Windows. It may be that the software for turning a pile of Linux boxes into a rendering farm is free or less expensive or more efficient than the equivalent for Windows.
And if you are talking thousands or tens of thousands of licenses for a rendering farm, you are no longer talking about tens of dollars.
Re:It's pretty much a given that they saved money (Score:4, Informative)
It may be that the software for turning a pile of Linux boxes into a rendering farm is free or less expensive or more efficient than the equivalent for Windows.
Indeed. It's not a coincidence that only 5 of Top500's list [top500.org] are pure Windows environments.
Re:$1.7m is dirt cheap! but missing something (Score:3, Informative)
Military and commercial flight simulators do have gravity force feedback. They are mounted on a hydraulic platform so that when the pilot pitches or banks the simulator, the platform moves in accordance with the maneuver so that the pilot experiences a force on his/her body approximate to what they would feel in a real airplane. The computer takes all parameters into consideration so that a balanced turn, for example, would hold the platform perfectly vertical. But a slipping or sliding turn would bank the platform to one side or the other.
Well, this is a simulation after all. The hydraulic platforms can only approximate real G forces. To really experience flying, you have to take an actual plane up into the sky. And trust me, they do.
That said, simulators are an amazingly useful technology even if they don't recreate the entire experience with perfect precision. You can learn 90% or more of what you really need to know about flying without ever leaving the ground. That saves fuel, maintenance, the cost of a plane, and lives.
Re:It's pretty much a given that they saved money (Score:3, Informative)
Re:It's pretty much a given that they saved money (Score:3, Informative)
Re:It's pretty much a given that they saved money (Score:5, Informative)
Microsoft has twice before refused to give the source code for Windows for Warships to the Australian Navy, what makes you think the RAAF will have any more luck. The US govt will block this as they have fears that this will be leaked to the soviets (throwback to some 1960's paranoia when Australia was being blamed for intel leaks caused by a CIA double agent)
How, the chair is not just a copy of MS Flight simulator X on a big screen, it's a hydraulic control system that needs to make precise movements in real time to correspond with input, Windows cant even control a mechanical lathe with millimetre accuracy, that's why DOS is still popular in the assembly line. Besides the RAAF's biggest cost isn't in software or hardware, its in operational costs. To achieve similar results using Windows (.net and what not) you need to use more powerful HW, increasing the amount of power it needs, cooling requirements and above all else, maintenance. Windows breaks more often then Linux, so the RAAF would need to spend more time on maintenance with a windows based system.
The RAAF would have evaluated all the options, Windows simply could not perform the job the RAAF asked of it. No corners were cut here. This didn't save "statistically irrelevant amounts of money", the project provided a machine that fits the specifications detailed by the RAAF.
Re:$1.7m is dirt cheap! but missing something (Score:1, Informative)
This is a common misconception by those who don't know anything about the *sensor* that records those.
The centrifuge is there for physiological conditioning/testing, it has nothing to do with your body using the G-forces for any sort of spatial orientation.
If you depend on your vestibular system and otoliths in an airplane, you will kill yourself. You can only depend on the instruments. Our inertial platform (vestibulum/otoliths) were not designed to be accurate over prolonged periods of time without visual frame of reference. Moreover, they get very much confused by G forces present during any maneouvering. People can't fly straight (wings/nose level) for a couple minutes when you take away the instrumentation and visual cues. During any maneouvering, it's mere seconds and you're lost.
Go fly with someone doing aerobatics, close your eyes, and try answering what attitude you are in, say every 10 seconds, while the pilot is maneouvering. Your second answer, and every subsequent one, will be as accurate as random guesses. BTDT.
Any sort of gravity force feedback is totally useless in a simulator. Heck, even moving platform is IMHO useless for anything but entertainment purposes. Your inertial platform can't integrate those inputs into your situational awareness (if it wins, you die), and whatever other inputs you have (seat-of-the-pants, visual cues, etc) are prone to be misinterpreted.
Heck, even reading the instruments is not all that cozy since G forces may induce nystagmus, which totally kills your visual acuity. During nystagmus, when eye position is essentially a sawtooth-shaped function of time, your visual system becomes a lowpass filter. This takes care of essentially removing any input from artificial horizon and HUD/EFIS. Your brain will play tricks on you and substitute expected values for everything that has high visual detail (all gages, all HUD/EFIS as currently designed, any glass cockpit displays with lettering/symbols, etc). This again will get you killed.
Glass cockpit design is, unfortunately, lagging some 20 years behind physiology, and this lag grows a year every year, with no end in sight. Those responsible should be summarily shot, IMHO.
Re:I want one! (Score:3, Informative)
The flight simulator pictured was built for the Royal New Zealand Air Force as a P-3 Orion Flight Training Device.
It runs x-plane. Austin Meyers (the author of x-plane) worked with Fidelity Flight Simulation to add unique features required.
I was the acceptance test engineer.
Re:Simulating what, exactly? (Score:3, Informative)
There's still a lot of work for human pilots, and there probably will be for at least another generation. The first UAVs that can handle manned-aircraft combat tasks are just now being deployed, and in many ways they're Not There Yet. Are you suggesting that air forces should stop training pilots now on the assumption that drones will take up the slack?
It's also worth mentioning that current-generation UAVs like the Predator are fully human-controlled by remote.
Related, interesting link: http://www.military.com/news/article/human-error-cited-in-most-uav-crashes.html [military.com]
Re:It's pretty much a given that they saved money (Score:4, Informative)
Linux has had an in here for some years now, due to earlier 64 bit support, and better/earlier support for large numbers of files in a directory. I work for a division of a fortune 500 that does datamining/text mining. Windows lost us in about 2004 for these two reasons, and there hasn't been any reason to go back. Cost wasn't the original reason, but cost keeps us from changing. $800 a machine adds up when you are looking at dozens or hundreds of rendering/compute servers. Linux has also proven to be easier for command and control of the jobs. The one thing that I long for is the full featured user identification/authentication support that Active Directory has.