Maintain a "pilot in control", which is to say that you must always be able to take manual control and fly the aircraft out of danger (in general, that means maintaining line-of-sight contact with the aircraft).
This is quoted from the Regulatory FAQ.
My question is - does this "manual control" need to be completely separate? I am using Arduinos and a data radio pair. The transmitting Arduino can send commands based on a joystick, which is my "traditional RC" mode. If my autonomous operation has a problem, there is a possibility that the running code has hit a bug and the controller may not respond to joystick commands. I am programming in safeguards (and this is not my first autonomous project; I have a working RC car and boat) but in some worst case scenarios the safeguards would just keep it from flying away and it would flutter to the ground (it is a light "foamie" disc design). Is this a violation of the regs?
The odd thing about this is that using the Arduinos as an alternative to traditional RC is fine and the first thing I am working on with respect to autonomy is actually an enhancement - heading back toward the takeoff location on loss of signal, based on GPS and compass readings. So it's only when it starts behaving better than a traditional RC that it would be considered not good enough...
Is there more detail somewhere that specifies redundancy or are you inferring it from what I quoted?
If you attempting to build in safety and redundancy to the level that the FAA (or whomever) would be satisfied with, let me assure you that that would be incredibly expensive!
The US Gov certification process has been explained before on this site, and involves an impossible amount of time and money.
P.S> ReturnToHome is already built in to the ArduPilot.
What regulation in the CFR covers UAV/UAS ?
No, I am not trying for certification. I was just trying to clarify the recreational guidelines. I do need to take a look at ArduPilot, but to start with I am using my own code that works on the car and boat. That code doesn't have elevation, but it has the GPS and compass input and bearing to seek calculation, compass monitoring in turns, communicatoin monitoring and hooks for other functions. Since the plane I am using to start with is a simple RET type, I pretty much have the R and the T. I am not trivializing what it will take to properly control the elevator, but that is a challenge I look forward to (and I have already found there is a sizable range where going fast climbs and going slower descends).
I am not sure my point was clear in the previous post. If I fly using my home built Tx/Rx to control the plane, I am certainly liable if I have a problem with it and cause some damage but I am not outside the law to use it as long as it works. But as soon as some autonomous feature is used, it sounds like I might be outside the regs even if it works perfectly and has adequate safeguards to cause no damage if it fails.
In the USA, at the moment, recreational users "are to look to AC 91-57 for guidance".
AC 91-57 has been ignored for the last 30 years because of its "voluntary" nature.
Recreational sUAS/UAS users can pretty much do anything they want.
That is in the process of radically being changed.
I did some searching and can't find the regulation about that frequency in particular, but I did find one about 72 and some info on 2.7GHz for model aircraft, 27 and 49 for toys and 900 for consumer devices. The 315 and 434 stuff you can get off shelf gets a pass because of the low power (though I wonder about the Seeed Studio long range 434 pair). Circling back to my original question, though, this could put a real damper on the idea of calling a plane driven by data com as being RC at all.
FCC regulations and information is available HERE
27.255 MHz @ 4 watts is more powerful than almost any other RC radio made today so I would hardly say it is used for toys.
I followed that link to the discussion of the current sub part 95 and its link to the actual document and found the reference to using the 26/27 frequencies. Unlike the 72 frequencies, they are not restricted to just model airplanes and there is a warning about potential interference.
One interesting tidbit is that there is a specific exclusion about sending data on RC frequencies if that data is device control information (otherwise, data transmission on those frequencies is forbidden).
However, I keep going back to that vague statement of "manual control" and it seems like that as long as you can demonstrate that you are able to over ride what the plane is doing autonomously and make it do what you want then the requirement should be satisfied. I have not seen anything that specifically says the link to control it has to be redundant or even an R/C link at all (no, I don't plan on trying IR or some other tech, just pointing out that it doesn't seem to specifically say). I mean no disrespect to those who have opinions that it can be inferred from the statement or perhaps have seen other more specific documentation. Please point me toward that other specific information or explain your conclusion. I am obviously biased because I don't want to spend the extra money and put the extra weight on my plane.
The 26 Mhz bands are not used much today. The 4 watt band is probably going to be free and clear. As for "sending data" that means you can't transmit video or audio over it or use it for comuter networking etc. just radio control sigmals (which is still data but the FCC doesn't seem to understand computer technologies).
Hoppe that helps.