LibrePilot Forum

General Category => General Discussion => Topic started by: 12many on July 21, 2016, 03:18:24 am

Title: Tuning for stability
Post by: 12many on July 21, 2016, 03:18:24 am
I'm starting to test a pretty large new vehicle intended to be a test platform.   On the 'bench' (dining room table) fastened down with tape & rubber-bands (3" of play) it seems a bit squirrelly in Attitude mode.  Manageable, but pretty touchy.

FWIW this vehicle is about 4'5' across, has 9x 17" props, and weights ~14 lbs with a full battery & camera payload.  Hovers @ just under 50% throttle which is expected performance.

Considering that 'the bigger they are the harder they fall', etc I'd like to avoid crashing my new toy.  It's intended to generally hover in-place or move slowly while I test out various fault tolerance & redundancy ideas I've wanted to experiment with.

Looking at the LibrePilot defaults, the parameters seem more tuned towards acrobatics than to create a stable platform.

What parameter changes will 'de-tune' my vehicle without making it uncontrollably sluggish?

Can someone could point out the key attitude stabilization parameters I should look at?

Title: Re: Tuning for stability
Post by: 12many on July 22, 2016, 07:33:32 am
Title: Re: Tuning for stability
Post by: TheOtherCliff on July 22, 2016, 11:11:52 am
Just a few hours ago AutoTune was merged into next, so if you can build next, you could try that.  You can easily configure a transmitter knob to access the SmoothQuick setting to make is as smooth or quick as you want.  Of course the quick side is limited by the max roll rate parameters (GCS -> Configuration -> Stabilization -> Advanced -> Responsiveness -> RateMode&MaxRateLimit)

The problem with AutoTune form your perspective is that it needs to fly freely to tune.  It is a modified Attitude mode that shakes the multicopter in all 3 axes.  Both of these sound scary in your case.

I would suggest that you get it freely hovering in Attitude mode.  If the aircraft is correctly configured (e.g. doesn't immediately flip because motors are connected wrong) you should be able to test Attitude mode with zero damage.  If cutting the power at 0.5m and 1 m/s velocity would cause crash damage, then even this is not an option.

Maybe you should build a much smaller version for testing?  It could be as small as a 500mm class if you use high RPM / small props.
Title: Re: Tuning for stability
Post by: NightLord on July 22, 2016, 03:10:00 pm
Let me contribute with r941 for x86-64
Title: Re: Tuning for stability
Post by: 12many on July 22, 2016, 07:25:15 pm
Thanks for the very good advice.  I was thinking some of the same things.

Actually I have a great vehicle similar to what you suggest for testing AutoTune, it just has a different flight controller.  Time to place an order for a new FC.