Slashdot Mirror
Can Unmanned Aircraft Mix With Commercial Planes?
coondoggie writes "The Federal Aviation Administration this week signed a research and development agreement with GE Aviation to come up with a way to safely mix the burgeoning amounts of unmanned aircraft with commercial aviation.
With this research the FAA and GE hope to accomplish an aviation first by completing the research to facilitate flight of an Unmanned Aircraft System with an FAA certified, trajectory-based flight management system.
Integrating unmanned aircraft into the national airspace will be no easy task. The Government Accountability Office last year laid out the difficulties stating that routine unmanned aircraft access to national airspace poses technological, regulatory, workload, and coordination challenges."
But OTHER aircraft might not be so predictable. TFA mentions, for example, gliders. They don't file flight plans. They're too small to carry much in the way of radar or other collision avoidance devices. Both UAVs and gliders tend to fly at low altitudes. Traffic can get very complex, very fast.
Besides, there is no such thing as a "simple" collision avoidance system. They're hard to do (mentioned, oddly enough, in TFA).
Faster! Faster! Faster would be better!
Actually, that's not too far from the truth.
An unmanned aircraft can survive much higher stresses than manned aircraft, so you could essentially make the unmanned aircraft drop out of the sky rather than collide. Maybe it can pull a 300G turn to avoid the collision. It's sensor package and avionics would react much faster than those controlled by humans.
Not according to FAA officials, says the article:
FAA officials also point out that TCAS computes collision avoidance solutions based on characteristics of manned aircraft, and does not incorporate unmanned aircraft's slower turn and climb rates in developing conflict solutions.
Google "procerus kesterel" "piccolo cloudcap" "papparazi uav". these can all autoland. we've been flying a kestrel with the ability to consistently hit a 3" circle on the runway with a 100lb plane going 15mph at touchdown. yeah bigger planes go faster, but they have larger control surfaces to compensate...
You might start w/this little NTSB report about a UAV in the national airspace system.
http://www.ntsb.gov/ntsb/brief.asp?ev_id=20060509X00531&key=1
Auto landing has been around since the 70s. I remember a full page newspaper ad announcing that "Auto-landing is here." IIRC, in order to be able to prove their Auto-pilot is capable of auto-landing the airlines are required to have periodic auto-landings done. http://en.wikipedia.org/wiki/Autoland
A secondary system running the same code with the same flaws as the first doesn't cut it in this context.
That's why you build the computer systems with triple modular redundancy. Basically, you make three different systems which have the same job and they vote.
Of course, a human or two as another layer of redundancy is often a good idea.
Centralization breaks the internet.
Radio control suffers from the same issue as fully automated UAVs.
The issue with them being in the airspace system is that much of the system is based on VFR -- visual flight rules. See and avoid. And even the parts that are always ATC controlled (Class A and B airspace) rely on see and avoid when the weather is clear. (It is not uncommon at all for an airplane approaching a very busy airport to be told something like "traffic 2 oclock, two miles", and then when the pilot says he has it in sight he's told "follow that aircraft".)
A radio control pilot cannot see anything other than what his camera is looking at right now. He can't swivel his head and see the Cessna 182 bearing down on him from the left as easily as a real pilot can.
I don't doubt that fully automated aircraft are at the level of sophistication where they can operate in a fully controlled environment, but we don't have many of those, if any. And we don't have ANY method I know of for ATC to issue emergency instructions to an unpiloted UAV, so they become essentially uncontrolled controlled systems.
The FAA approach to aircraft software is loosely as follows (refer FAR 25.1309 as a starting point):
- When building the aircraft, as part of a wider safety program a System Safety Assessment is carried out on the system in question - This may be in accordance with SAE ARP 4754
- The SSA determines the required 'Design Assurance Level' of the system in question. - i.e a 'fly-by-wire' flight control is likely level A, and an inflight entertainment system might be level E.
- The system is then built, and the software developed using a suitable software lifecycle process (such as IEEE 12207).
- The Software is developed against an 'Assurance Standard' - most likely DO-178B. This requires various things to happen depending on the 'Level' of assurance required. If it is level A software (i.e for a flight control system), then there are lots of development and test requirements required (e.g full high-level requirement trace to low level requirements to hardware, with independence. - and full code coverage with testing of all inputs and outputs in every iteration). For something like an in flight entertainment (Level E), there are very little code / test requirements (to meet FAA regs - not passenger satisfaction!)
The FAA credit the (quite robust when followed) DO-178B process as the reason for so few software related accidents. Many examples of aircraft accidents the media attribute to 'software fault' is usually a hardware error providing incorrect input. - or a result of poor requirement definition up front.... (such as software had no requirement to disregard erroneous Angle of Attack data, causing severe pitch problems in an airbus.)
If you get into it, the FAA regulations around software are pretty safe. If you're in doubt, contact your local D.E.R.
Have the drone pull a 400G turn until there isn't a threat anymore. Worst case scenario, trigger an explosive to blow it out of the sky.
Who ever rated this nonsense "Insightful"?
The laws of physics don't change just because there is no pilot in the plane to observe them. Higher performance (= climb and turn rate) come at a price. You CAN design an UAV for higher performance, but then it needs big wings and a large engine.
Before it can pull 400g, it needs to create lift 400 times it's weight force. And withstand the stresses involved. With current materials and technology, 400g is a pipe dream.
You know it's time for the next revolution when your rulers' names end with roman numerals.