2
$\begingroup$

Think something like ardupilot but for manned aircraft.

$\endgroup$
  • 1
    $\begingroup$ I doubt it, it could possibly be used in Experimentals, but there are manufacturing, quality control, and testing requirements that would be difficult or impossible to manage with public contributions. $\endgroup$ – Ron Beyer May 10 '18 at 20:56
  • $\begingroup$ Closely related. In particular, this answer might help you. I don't think we can really say if it will "ever" happen, because that's speculation and we don't know how/if the rules will change in future. $\endgroup$ – Pondlife May 11 '18 at 0:58
  • 1
    $\begingroup$ Since FAA does not force the source code to be confidential, so the owner of the code has full freedom to license it to whoever under whatever license. So while hobbyists developing such systems from scratch is hard, vendors developing it in house and open it to the public is highly possible. In fact I would say most of the contributions to FOSS are funded by vendors not volunteers. $\endgroup$ – user3528438 May 11 '18 at 3:41
3
$\begingroup$

Short answer: technically and legally yes, a free open source autopilot could be certified. But I think it is unlikely to happen anytime soon.

Long answer: Aircraft software certification is governed by DO-178C. There is nothing in there that says anything about whether the source is open vs closed, or free vs non-free. What it does lay down is a lot of very rigorous requirements for proving that the software does what it says it is going to do.

For example, for something like an autopilot, an independent testing/validation would typically be required. That means the person who writes/codes the software and the person who tests the software have to be different people. i.e. you must have a fresh set of eyes looking at it. Further that testing will have be very extensive, covering every possible scenario your autopilot could encounter. You might have to mock up an entire fake airplane, with simulated sensors and simulated actuators, and feed the autopilot simulated data and see if the actuators move like they are supposed to.

You also need to provide a massive amount of very careful documentation about all parts of your software process. E.g. the software requirements need to documented, and then you need to be able to show every step of the way on how each specific requirement gets translated into the final machine code.

An free open source project certainly could meet all of these requirements. However, they end up being very time consuming and expensive. You need an army of software engineers working all day every day for years to meet some of these requirements. FOSS project typically do not have the level of manpower that is required to meet these requirements.

$\endgroup$
  • 2
    $\begingroup$ As a software certification engineer, this is a very good answer. The trick with most FOSS is that there are not documented requirements that trace to design or test. It's impossible to show that a piece of software works correctly and has no untested code if the requirements aren't there to verify the code against. $\endgroup$ – selectstriker2 May 13 '18 at 3:56
1
$\begingroup$

There is no practical way for Free Open Source Software to be used in any FAA certified equipment. This approach to software development encourages users to modify and contribute code, so as to benefit the entire community. Much of the motivation is to see the code improved and put into use.

FAA certification requires such extensive testing and documentation that revisions do not occur except to correct a serious flaw, and this usually means removing the hardware and returning it to the manufacturer. New functionality might require a new generation product. It is not likely that a five or ten year wait for new features would be attractive to FOSS developers.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.