Moderator Comment by Gary Mortimer on November 11, 2011 at 9:25am

Is that using Bill's (yes I can't spell his surname) algorithm for wind estimation or your own?? Certainly an airframe you could use for T3 2 

Moderator Comment by Krzysztof Bosak on November 11, 2011 at 9:56am

No, its my own. Not Premerlani's. I use Kalman formulation + some ceative stuff, he used DCM.

Moderator Comment by Gary Mortimer on November 11, 2011 at 11:03am

Ah fairy dust, very cool. I wonder why every autopilot is not pitot free these days.

 

Does yours measure the vertical component as well like Bills?? Very handy for soaring perhaps.

 

That airframe is really efficient in the full sized version do you think it has any benefit in a smaller size?? I have wondered about any number of the recent ARTFs in Hobby King. 

Comment by brakar on November 11, 2011 at 1:34pm

A nice thing with those kind of planes is they have lots of room for autopilots and stuff. It does however not look like a stable airframe like the Pteryx, another airframe meant for testing?

Moderator Comment by Krzysztof Bosak on November 11, 2011 at 2:42pm

@Gary "Does yours measure the vertical component as well like Bills?"

Impossible to tell as if you develop your autopilot from scratch with the idea backed by scientific analysis and not belief, on one hand you know what are most important factors in comparing (and you cannot tell that because it's your hard learned klnowledge) and there are even more factors that are solution specific and eluded you (so you cannot tell how to compare since you don't know what to compare more).

Nominally FLEXIPILOT has all parameters well below most autopilots, including realistic installation weight, heigh, MIPS, raw ADC precision. So it is impossible to tell without comparison campaign, that being hard and annoying job, will never be done for any free autopilots.

@brakar

The little devil withotu undercarriage is neutral in roll, but the neutral point in all axes is also varying wildly with throttle. Effectiveness of some control surfaces is strongly varying with throttle (which has specific rpm vs airspeed vs throttle vs amp battery output dependency) in short it is impossible to identify without numerical analysis of the logs and serious knowledge about aviation.If you ignore it you get different turn radiuses and oscialltions all the way depending on speeds and throttle. It will fly nicely on some conditions, but get nasty all around those conditions.

Lancair has the major thing to avoid for amateur UAV hobbyist:

-large prop and power2weight ratio

-large span of airspeeds that are gained quickly

-smal size that makes identification by eye difficult

-easy to destroy airframe

-nasty stall characteristics

You cannot leave it for magic 'self-learning machine' because you got some 12 seriously 3D nonlinear variables with wide span and counter-acting side effects on most flight phases and acting together in some flight phases (in another platform some signs can reverse). If you don't know how to supply database even using automatic learning techniques you cannot guarantee the safety limits and would end with like 1000h testing time in real life.

This is why 'buy our autopilot it can fly everything' is something only True American would say, implicitly adding 'good luck dying trying unless you are aviation engineer of some sort'. In fact tuning anything except 'mainline' is a subject for PhD and not just anybody. Because the hobby is so full of over-ambitions guys, if you leave it for the public, you are sure you get the result like 'look it flew' boosted by a movie and spin-doctors not reflecting the fact it is less practical than SR-71 and requires like 20h of pure tuning time (this one did).

 

@gary

I have tried automatic soarign as Andrus Kangro did once, it worked somedays but I gave up the subject. The real point is everything can be done but then the autopilot design 'serves everything, but the pilot is always wondering why it did it'.

 

@brakar Pteryx was designed like it was mostly for easiest possible flying in manual mode at edge of visual sight (max backup range if everything fails)  also for safety as it never falls down in parabolic line even if any control surface fails. Besides strong stability, Pteryx has strong aerodynamic damping what makes it fly... completely differently than small plane with ailerons, and more difficult to tune in this regard (navigation lag to remove). You have to hold the rudder deflected in the air to keep it turning, otherwise slowly but inevitably it return to straight in 2s.

Comment by Mike on November 11, 2011 at 3:14pm

"Does yours measure the vertical component as well like Bills?" - I would have thought a simple "yes" or "no" would have been sufficient rather than a War & Peace length marketing pitch...but each to his own I suppose...I am still none the wiser!

Moderator Comment by Krzysztof Bosak on November 11, 2011 at 3:32pm

@Mike

the simple answer is, for the reasons mentioned, I don't care.

Does this sounds polite towards Bill and Gary to answer like this?

Is it short enough for you to understand?

 

Comment by Mike on November 12, 2011 at 1:14am

@Krzys - sounds like you are pleading the 5th Amendment to me!

As regards understanding - that is not a function of length of the response and I never said I didn't understand.

However, looking at your response again I should have realised that "Impossible to tell..." means you don't know rather than don't care?

 

Anyway good work and always interested to see what you are up to.

 

Mike

 

Moderator Comment by Krzysztof Bosak on November 12, 2011 at 3:20am

A longer-phrased response: I don't care abotu gathering the remaining 20% of knowledge about what to do just to  compare both systems and the publishing the results, while I am fully aware that Bill is even further from the answer since he sees no code of my autopilot. Therefore nobody knows fully, my guesses cannot be revealed and the whole question is a little pointless, because even if the answer is yes the solutions are even borrowed/stolen/identical it doesn't implies anybody will be ready to tune a platform to use this feature reliably. The last phrase is not because forum engineers are worse, but because they followed classic autopilot development path for the past two years.

Moderator Comment by Gary Mortimer on November 12, 2011 at 3:35am

I put my foot in it again, I was just interested.... Not having to use a pitot is a major feature in my mind however it is achieved.

Comment

• ‹ Previous
• 1
• 2
• Next ›
• Page