Quote from: MikeCoulombe on September 01, 2013, 07:49:24 AM
So the tech exists, banning it won't make it go away.

Personally I can hardly wait for a fully programmable over the counter control system.

   Why? What is the point of doing that, aside from showing you can do it? If you buy it turn-key, you are showing only that someone else can do it. But no one cares what you sport-fly with, enjoy it however you wish to. Contest rules are not going to matter for sport-fliers.

   Banning it certainly will make it not show up in a contest environment, which is where the problem is. Virtually no one wants an autopilot to be legal, Kim included, because it fundamentally alters the event. Autopilots will be made illegal, period.

Given that, if it requires banning fly-by-wire systems unnecessarily (because we can't come up with a better way to ban autopilots than "mechanical connections")  that will be what happens.

    Brett

I didn't say "autopilot" I said "control system".
I don't want an autopilot any more than you do.
I think we need to define "where" exactly digital control ends and autopilot begins.
Some may say that accelerometer based throttle control is autopilot.
Some may say that a mechanical device to change you rudder in turns is autopilot.

Kim,

I just gave you a + to get you out of that -1 hole you were in. A hole I know all to well. 

I was at -375 once, remarkable but true! Tell ya, it's dark and lonely down there.  Really dark and really lonely. You can't dig your way out of a -375.

Model airplanes, will they ever be X-rayed? Scrapings from under the nails to test for BOM?

Anyway, I have interest in anything electric, and expect to enter that "cave" shortly. Shortly, well, hopefully before I die. Got some engines to sell first.

I like and have interest in what you're up to. Kudos! 

That servo thing operating the control surfaces. I know it quite well from R/C pattern and it works. Not sure for CL, but please, you will keep us informed.

I'm always late in Threads because I don't like to read. So, at first, I thought you eliminated one of the handle lines. Hey! That thought worked for me. 

In all honesty, when I saw your magnetic bellcrank, I was at first disappointed. I'm over that now and, two lines or not, I'm still interested in your endeavors, .

Can you get it down to one line, one servo and a "joy" stick?

Hey! I got your back with that +.

Keep up the good work!

Charles

There are some nice conveniences available for a fly-by-wire system, I think.

If you want a removable wing on your model, it is a lot simpler to do if there are no mechanical control linkages. Just an electrical connector and a few nylon screws to hold the wing on.

Plus if you want to play with control ratios flaps/elevator, you just modify program a bit. Also if you want a response curve on your elevator or flaps that is not linear (say exponential or logarithmic) it is easily accomplished in software. No fancy machining or mechanical design.

-Chris

Quote from: Chris Sarnowski on September 02, 2013, 06:06:38 AM
There are some nice conveniences available for a fly-by-wire system, I think.

If you want a removable wing on your model, it is a lot simpler to do if there are no mechanical control linkages. Just an electrical connector and a few nylon screws to hold the wing on.

Plus if you want to play with control ratios flaps/elevator, you just modify program a bit. Also if you want a response curve on your elevator or flaps that is not linear (say exponential or logarithmic) it is easily accomplished in software. No fancy machining or mechanical design.

-Chris

Chris, You Nailed It Exactly!

I have been playing around with the line spacing at my handle to get the response that I liked from a model that I just started flying. With the fly by wire system you probably can just change some system settings to get the control sensitivity that you like. Also if you move the position of the lines to get the same sensitivity inside and outside turns, you can do that electronically as well. That is pretty neat, too!

-Chris

Quote from: Tim Wescott on August 31, 2013, 05:44:37 PM
To do an autopilot that could beat one of the world-class guys would take a lot more than just the ability to string a few lines of code together.

Yes, it would.  It would take MEMS sensors, now readily available, and the specialized knowledge of control systems, with its kinda esoteric math.  Some of us have the latter, including Brett, me, Igor, and Tim.  I think most of us can see what's coming.

Quote from: MikeCoulombe on September 01, 2013, 01:26:47 PM
I didn't say "autopilot" I said "control system".
I don't want an autopilot any more than you do.

None of us can look at somebody else's electronic control system and tell whether it's a straight open-loop control, augmented control, or an autopilot.  That's the problem. The only way to draw the line is to forbid electronics in the pitch axis.

Quote from: MikeCoulombe on September 01, 2013, 01:26:47 PM
Some may say that accelerometer based throttle control is autopilot.

It's an autothrottle.  That's what it's known as in the industry, anyhow.  Now you have to have one to place in the world or European championships or to win the US Nats. The effect of automation in pitch would be much more than for propulsion. 

Quote from: MikeCoulombe on September 01, 2013, 01:26:47 PM
Some may say that a mechanical device to change you rudder in turns is autopilot.

Obviously not.  Writing the rule to ban electronics would allow such a linkage.  It's a simple distinction.  Using a yaw rate feedback to control rudder, as has been done since the early days of swept-wing airplanes, and as Igor has done with a stunt plane, is a different matter.  My people would call that a stability augmentation system.  Whether to allow a SAS in yaw and roll is a separate subject than allowing electronics in pitch.  I would recommend banning electronics from all aerodynamic control surfaces.

Quote from: Howard Rush on September 03, 2013, 05:24:09 AM
None of us can look at somebody else's electronic control system and tell whether it's a straight open-loop control, augmented control, or an autopilot.  That's the problem. The only way to draw the line is to forbid electronics in the pitch axis.

That would be one way to do it.

Another way would be to simply say the system must be open-loop, and the contestant says that it complies.
You know, take people at their word.
Works in F3A and F3P where stability augmentation is not allowed.

And it would be easy to see what devices are on a PCB. If there is only a microcontroller and no accelerometers or gyros then it is pretty clear that the system is open loop.

If one's goal was to cheat then even with normal bellcranks and solid pushrods systems could be devised that could augment stability.
Look no further than cars with stability control to see how it can be done.

I can't quite understand the desire by some (a few?) to kill this before it has even flown in a contest.
Plenty of time to formulate rules if things get to the point where electronic systems show some sort of clear and unfair advantage.
Till then it seems very reactionary.

Pat MacKenzie

P.S. Congratulations on making the US team.

Pat, that what you say is not enough. You probably did not read argumentation while we spoke about it in rules working group. We also do not want preprogrammed open loop clock work, not only feed back systems. We cannot say "it is compliant" because we even cannot define what is and what is not allowed, because there is not clear line between "controlled by pilot" and "autopilot". This what you wrote is clear example. We not need to see only rocket technics with gyros and accelerometers. I can say for myself, that on latest ECh I got marks between 8 and 9 for almost all figures. But I got twice (from 7 official flights) mark 6.8 for not so well done 60 degrees turn to triangle and hourglass (it was not so bad, but some judges simply want cut top guys for even small mistake to make clearer differences). I wrote it somewher in another thread and I posted also my score sheet, it is well visible. So for me it will be very very helpful a small function which will trigger preprogrammed 60 degrees turn when it sees that I do first corner of triangle or hourglass. This will give me may be 5 or 10 safe points what will make me good sleeping front of Richie and Alex. This is in my eyes clearly unfair, but I can still say that I send my pilot commands by lines to model. And even IF we can find clear wording to disable this then we cannot find way to inspect it. Because it is PROGRAMATICAL solution. You cannot see it on PCB, and you cannot check it on ground.

And why we want it stop as soon as possible? You know how much effort it costs you ... so that is reason, those who can do it also will must develop it if they want to stay competitive and those who cannot, will think if they want continue. We simply do not want to kill event with solution undoable by average pilot or builder.

Pat, again and again ... this is about stunt rules, nothing else. When I saw it first time, I did not wrote to Kim (or anyone else) that he is "bad guy". I wrote to working group and triggered speech about it. Result is that the best way is to allow only mechanical way from pilot to any control surface. And why? it is pretty simple and practical - because for passed 50 years no one invented useful autopilot. So we believe no one will do it also in next 50 years, while only existence of FBW opens in all heads how to use gyros, accelerometers and other tricks. I know you spent lot of time with FBW and you do not want to throw it to trash, but you knew long time that it will have strong opposition, that is why I opened it immediatelly I saw it first time. It is not plenty of time to formulate rules, as I see it is already too late.

But tomake it clear and I wrote it already, I do not see problem with FBW itself, if FBW will allow newcomers to start safely as Kim wrote, it will be very good, so simply do it, I will welcome it in area where it helps, I simply do not want to push stunt to programmers event for 5 people of planet.

Something like a preprogrammed 60 degree turn will require massive amount of feedback.
I doubt it is even possible without  RF connection between a ground based computer that employs multiple cameras to track the model combined with a precise mathematical model of the models flight.

The latter is how the many videos you see of quadcopters dancing, bouncing balls back and forth or playing music have been done. None of them are autonomous.
Until you see someone setting up a half dozen video cameras, some of them 60 feet in the air, and parking a mini van full of equipment next to the circle you don't have anything to fear about being beaten due to prep-rogrammed 60 degree turns.

Without all that "tech" the only thing you can do is simply preset the amount of deflection and hold it for a certain period of time.
The end result will be about as good as a blindfolded pilot would be capable of accomplishing.

In any case, again looking to other events pre-programmed sequences are also not allowed by rules in F3A.
So we are once again back at taking people at their word.

I did see some of the working group discussion, and in the end there was no change to the rules that precluded a FBW system.
What they ended up doing was allowing RF control of non primary systems, so rather than restricting technology the actually expanded it and defined where it was permissible.

Quote
1.3.2 Category F2 - Control Line Circular Flight
a) Control Line Circular Flight is flight during which all control is accomplished via physical connection
to the pilot through one or more inextensible wires or cables directly connected to the model aircraft.
The control wires or cables must be attached to a hand held device (control handle). Automatic flight
path control and/or automatic manoeuvring are not permitted.
b) Powertrain control may be accomplished by the pilot via the wires or cables or by an onboard selfcontained,
automatic process.
c.) For permanent shutdown of the engine(s), any device or system is permitted including the use of 2.4
GHz Spread Spectrum technology legal for use in the country of competition. The competitor will
determine the suitability for use of the chosen system.
Any such device or system:
1) must be operated only by the pilot, and
2) must not affect any other model.

Pat MacKenzie

Quote from: pmackenzie on September 04, 2013, 03:22:19 AM
Without all that "tech" the only thing you can do is simply preset the amount of deflection and hold it for a certain period of time.
The end result will be about as good as a blindfolded pilot would be capable of accomplishing.

   No. If I take a reasonably accurate (over the period of 200 msec) gyro and integrate the output, I can make it do a nearly perfect corner. You couldn't do a whole flight that way, but the effect of drift would be negligible and the scale factor error could be calibrated out. We do it all the time. It would be open-loop, but dead-nuts repeatable. You could get it right in a few flights.

   With enough radio links, you can do what you can do with multiple cameras. A completely passive tracking array could be generated using 3 transmitters and a properly-designed antenna and tracking system in the airplane. It could be as accurate as half a wavelength, which if I did my guzzintas right, is about 2 3/8". Easy - no, but certainly possible with sufficient trickery. Since there will be nothing preventing a 2-way link, all the processing could be on the ground.

   Of course, I also know how to jam such a system, so if you see me with a 5Kw spark gap transmitter, that's why.

    Brett

 

Well I have to say Kim's various threads about Fly by Wire has been a very interesting read and sorry for going "Off Topic" but.....

I'm still not convinced that it's even possible to put together a package that would respond to prevailing conditions even close to that of the handle waving capabilities of <Insert your favourite Stunt God here>. But I'm not saying never, and I guess future proofing is what establishing the rules are all about.

I was wondering given the assertions of various people that this could be done now if a better use of the perceived auto pilot technology would be to develop it's use it as a scoring device. If it really is possible to get the perfect pattern with an auto pilot then presumably it is even easier to remove the complication of actually moving the control surfaces and just log the flightpath? Then you could just download the flight and let software compare its path with the manoeuvres as they should be flown and score accordingly.

I'm not trying to do the judges out of a job as it's a thankless task but Igor commented that sometimes he feels that as a top pilot he gets hit harder than some if he flies a bad manoeuvre and others have suggested the feeling of a "Halo Factor" or discrepancies between different judges scores for the same manoeuvre.

One of the issues with F2B as opposed to other Cl disciplines has always been that a flight could be considered open to a certain amount of interpretation rather than an judged by a measurable baseline such as a stopwatch or streamer cuts etc.

I'm not having a dig at judges, far from it but if an application which just scores based in a known baseline was possible it could be a great "leveller" at Nats and World Champs for example.

I would be all for it.

TTFN
John.

Quote from: Brett Buck on September 04, 2013, 01:03:04 PM
   No. If I take a reasonably accurate (over the period of 200 msec) gyro and integrate the output, I can make it do a nearly perfect corner. You couldn't do a whole flight that way, but the effect of drift would be negligible and the scale factor error could be calibrated out. We do it all the time. It would be open-loop, but dead-nuts repeatable. You could get it right in a few flights.

   With enough radio links, you can do what you can do with multiple cameras. A completely passive tracking array could be generated using 3 transmitters and a properly-designed antenna and tracking system in the airplane. It could be as accurate as half a wavelength, which if I did my guzzintas right, is about 2 3/8". Easy - no, but certainly possible with sufficient trickery. Since there will be nothing preventing a 2-way link, all the processing could be on the ground.

   Of course, I also know how to jam such a system, so if you see me with a 5Kw spark gap transmitter, that's why.

    Brett

 

You are making my point, thank you.
Gyros or radio links = tech.
Already prohibited under the rules and not a part of our system.

My point was that a FBW system as allowed under the current rules incapable of doing a pre-programmed 60 degree corner.
Or maintaining level flight, or any of the other things some seem to fear.

Pat MacKenzie

Quote from: John Hammonds on September 04, 2013, 05:06:24 PM
I'm not trying to do the judges out of a job as it's a thankless task but Igor commented that sometimes he feels that as a top pilot he gets hit harder than some if he flies a bad manoeuvre and others have suggested the feeling of a "Halo Factor" or discrepancies between different judges scores for the same manoeuvre.

John, I did not want to comment judges work, pilot never see his mistakes well, in reality I saw it and others (well Tania :-P ) also told me that I did not do it well. I wanted to tell that such mistake can cost lot and judges are not forgiving and I did it twice on the same place. The point is that such a system not needs to make pattern instead of me, it is plenty to give little advantage which can be enought to securely win or move forward. Differences on top places were small this time. And if you look to results of your NATs you will see it also. Therefore such a simple trick like preprogramed one small invisibke point in whole program can easily change result. We not need rocket technic.

And speaking about computer judge is completaly something else. It need RECOGNIZING ... that is relatively difficult, but theoretically still doable, but knowing the path of the model, I cannot imagine how to convert observed difference between real path to theoretically proper in terms defined by rules ... and then even worse, how to deduct mark. Making autopilot from this point of view is trivial job :- ))))

Quote from: pmackenzie on September 04, 2013, 05:45:58 PM
My point was that a FBW system as allowed under the current rules incapable of doing a pre-programmed 60 degree corner.
Or maintaining level flight, or any of the other things some seem to fear.

Seems to me that you are really trying to keep open doors for such things. I am surprised you are trying to say something is not doable without trying I wold never say such thing, because I saw "undoable" things many times in my life.

Such preprogrammed maneuvers are successfully done in R/C world may be 20 or 30 years ago. By simple timers. I saw perfect and repetitive rolls always finishing in perfect level position with F3A airplane and timer executed by simple button. So it is obviously doable and replacing muscle memory be electronic device does not have space here in my eyes.

Quote from: Chris Sarnowski on August 31, 2013, 05:04:18 AM
Berkeley had a model back in the 1940's or so that had mechanical flight stabilization. If you are going to have a rule to eliminate electronic stabilization then you also need to include mechanical stabilization as well. It had a pendulum to augment the user control inputs. It was the Berkeley "Bug" and the control system was called "Autotrol."

-Chris

And it was bogus.  

Quote from: pmackenzie on September 04, 2013, 05:45:58 PM
You are making my point, thank you.
Gyros or radio links = tech.
Already prohibited under the rules and not a part of our system.

My point was that a FBW system as allowed under the current rules incapable of doing a pre-programmed 60 degree corner.
Or maintaining level flight, or any of the other things some seem to fear.

  Agreed. The problem is that those very same rules DO permit a pre-programmed corner, and that is very likely to get fixed in the very near future. If you don't want FBW to get caught up in that, then you need to come up with a reasonable rule that permits one and not the other. Personally, saying it isn't an autopilot is probably OK with me but that's not going to be good enough for everyone else, you are going to have to prove it somehow.

   Brett

Quote from: Howard Rush on September 05, 2013, 10:20:31 AM
And it was bogus.  

Do you mean "bogus" as in it didn't work worth a damn or "bogus" as in it against the rules?

I have to dig out the plans for it again...

-Chris

Quote from: Howard Rush on September 03, 2013, 05:05:53 AM
Yes, it would.  It would take MEMS sensors, now readily available, and the specialized knowledge of control systems, with its kinda esoteric math.  Some of us have the latter, including Brett, me, Igor, and Tim.  I think most of us can see what's coming.

I'm not sure that the current crop of consumer-grade, lightweight MEMS sensors is up to the task of doing an entire pattern autonomously -- but that's mostly because there's very little information available to the Kalman filter about the airplane altitude until it pops into the first corner of the wingover.  After that, I'm pretty sure you could just attach the handle to a speed pole, then sit down and enjoy the show.

With GPS augmentation it would be a no-brainer, in a corporate-funded half a man-year of effort sort of way.  I've got about 50% of the code for that sitting on my computer, left over from a vehicle tracking project that included some generous terms about what technology I could keep for my own use.  Attach handle to speed pole, push "start" button, sit back...

And Howard -- I don't think the math is kinda esoteric.

Quote from: Tim Wescott on September 05, 2013, 11:54:18 AM
I'm not sure that the current crop of consumer-grade, lightweight MEMS sensors is up to the task of doing an entire pattern autonomously -- but that's mostly because there's very little information available to the Kalman filter about the airplane altitude until it pops into the first corner of the wingover.  After that, I'm pretty sure you could just attach the handle to a speed pole, then sit down and enjoy the show.

With GPS augmentation it would be a no-brainer, in a corporate-funded half a man-year of effort sort of way.  I've got about 50% of the code for that sitting on my computer, left over from a vehicle tracking project that included some generous terms about what technology I could keep for my own use.  Attach handle to speed pole, push "start" button, sit back...

And Howard -- I don't think the math is kinda esoteric.

   You aren't going to do a full pattern on inertial guidance with the current MEMs types that would fit in an airplane without an external observer of some sort. the drift and scale factor with 360-degree-per-second maneuver over 6 minutes. And I might suggest you don't want a Kalman filter to incorporate the observers, either, or at least I have seen plenty of cases where the Kalman filters were not optimal in these DOF-constrained cases.

  But overall, the notion that it couldn't be done and therefore there's no reason for concern is clearly not valid, even assisting in a few spots could make all the difference in the world, per Igor's example.

    Brett

Quote from: Brett Buck on September 05, 2013, 12:51:33 PM
   You aren't going to do a full pattern on inertial guidance with the current MEMs types that would fit in an airplane without an external observer of some sort. the drift and scale factor with 360-degree-per-second maneuver over 6 minutes.

I've started up a new thread for this:

http://stunthanger.com/smf/index.php?topic=32691.0