Flight controller's ability to handle loss of thrust

[chipwich]

I was test flying my completely rebuilt F550 w/NAZA V2 yesterday. I lost M2 about 3 mins into the flight, but didn't notice it until I began doing some aggressive forward/aft movements and it dipped hard. I was really impressed with the NAZA's ability to compensate.

The only other FC that I have is a CC3D on a Honey Badger, and haven't lost an engine yet. I'm wondering how well the other flight controllers handle this. I'm assuming that a quad would be SOL.

 

[DucktileMedia]

It seems DJI and a few others have found a way to make a hex controllable in a motor out situation. Yes a quad is SOL no matter what FC. It has a lot more to do with how much power reserve you have though. So don't expect a hex that is overloaded with gear to act the same if you are flying at 75% throttle.

 

[chipwich]

a few others

Who though? APM, Pixhawk, Revo, Wii, KKD? Would be good to know and it begs the question, why build an X8 or Y6, unless you have a specific need for those designs? Perhaps in a heavy lift scenario the X8 is easier to transport. Or perhaps the reduction is frame materials outweighs the reduced efficiency in a coaxial motor arrangement. 3DR seems pretty convinced that the X8 / Y6 designs are good.

 

[jes1111]

None of the controllers (that I know of) have any mathematics or algorithms or adaptive code of any sort to cope with a motor-out situation. In fact, the controller remains completely unaware that anything has changed - it just keeps issuing throttle commands to the motors as if they were all present and correct. The fact that a 6- or 8-prop machine can keep flying is dependent, as has been stated above, on the motors having sufficient headroom not to "top out". Thus, if a motor stops whilst you were in a stable hover then there's a good chance you might remain in control. On the other hand, if you're in a "banshee" full throttle tight left turn when it happens then you may well find yourself "SOL"

 

[chipwich]

Thanks for the reply jes1111. OK, still trying to sort that one out. I guess that since more than one motor would be responsible for a particular attitude change, that the controller would keep increasing/decreasing thrust on those motors to compensate for what it is sensing.

 

[R_Lefebvre]

I flew an octo on APM with 3 motors out, just to see if it could be done (of course, I got to choose the motors... and chose wisely)

Jes, I'm not entirely sure that DJI is doing it through blind PID luck. If you can recognize the problem, you can change the controller strategy to one that works better. Hex's have a specific problem that Octos don't, and could benefit from extra help.

I'm not sure how APM would work on a hex with a motor out. Nobody has given me one to find out yet.

 

[DucktileMedia]

I suppose that could be true but it does seem like "they" have made some improvements over the years. Maybe it's just coincidence that the videos we see are better setup to deal with a motor out. So far I've seen every config except a quad land safely with a motor out. We have all over thought this subject, certainly I have. I would just do your best to use good components and test your gear. If you are really flying expensive gear you will most likely be on an octo/x8 anyways.

 

[chipwich]

but it does seem like "they" have made some improvements over the years.

I'll say. I've owned my F550 since they were released and can remember it throwing a prop that I had over-tightened causing the hub to split under load. The hexi went in to a spin and then pitched over into the grass. Now I can basically hold hover, perform gentle maneuvers, and fly back to home without drama. Pretty impressive IMO.

 

[jes1111]

Well, it sounds like the more recent DJI code is NOT doing something that it used to do, e.g. they've properly limited an error term in the mathematics. In other words, they've corrected a bug. This is quite different from "adaptive coding", the ability to change the control strategy to cope with a motor out. We can't see the DJI code but DJI does have the benefit of being able to peruse all the Open Source code out there to work out what they were doing wrong.

 

[PhotoShip One]

It's possible for a quad to fly with two opposing motors inoperative. Impressive. Watch here at the 6:40 mark

http://www.ted.com/talks/raffaello_d_andrea_the_astounding_athletic_power_of_quadcopters.html

In the full scale fixed wing world there's been considerable research done recently to flight control systems to be able to recognize and adapt control algorithms on aircraft that have major control systems inoperative or even missing flight control surfaces altogether.

Adaptive algorithms is where multirotor flight controllers are soon headed.

 

[DucktileMedia]

Man that would be nerve racking demonstrating that in front of all those people. Here, let me show you how my 800 SRH still lifts a RED when I whack it in the gonads with this baseball bat!

 

[R_Lefebvre]

Must always be understood when watching those videos, that it's sort of an apples to oranges comparison. They are running a Vicon vision system which determines the attitude of the copters using a desktop computer. That info is then relayed to the copters. Half the battle of getting a controller to fly while spinning like that, is just getting it to understand it's attitude. There's a lot of forces going on and it can easily get "dizzy". The Vicon system solves that.

The same is true for all these awesome videos they put out. They rely on the Vicon, which means this stuff can only be done indoors.

 

[DucktileMedia]

Vicon, is that Motion capture?

 

[R_Lefebvre]

Yes, that's why they all have silver balls on them.