Expected performance of altitude hold and GPS hold modes?

Post Reply
doozer
Posts: 1
Joined: Tue Dec 31, 2013 10:13 pm

Expected performance of altitude hold and GPS hold modes?

Post by doozer »

I have a new hex with the HK AIO board + NEO-6M GPS module, Multiwii 2.3.

I'm wondering what sort of performance to expect from the "hold" modes... I have tried them and generally am unimpressed at present; I'd like to know if I should try tweaking, or if my expectations are just a bit unrealistic.

Altitude hold - I've tried enabling this mode after sitting for several seconds in a stable hover. Its clear that the mode is operating, but I see swings of up to 2m or more in altitude thereafter. Its not tied to craft motion or control input - even just maintaining position there is lots of altitude variation. For reference the HK board has a MS5611-01BA01 barometer. When sitting dead on a table and watching Barometer output with Multiwiiconf the barometer is very stable, but there does tend to be a "blip" every 10 seconds or so... i'm not sure what units the GUI is displaying, but the value changes from ~1.5 steady to ~4 during the blip, before returning to a steady ~1.5.

GPS hold - I've had worse luck with this one. After enabling, again from several seconds of static hover, the craft will slowly migrate and then eventually flies determinedly in some direction (random? can't tell...). I usually lose my nerve and deactivate the mode when it reaches about 10m away (my flying area is not enormous). MultiWiiConf indicates that the GPS is connected properly and I am achieving a good fix; the LED blinking also indicates 5 or more sats acquired. I haven't tried the GPS home mode yet, that one scares me more :).

I'm hoping that some tweaking will improve both of these, yes?

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Expected performance of altitude hold and GPS hold modes

Post by Alexinparis »

doozer wrote:I have a new hex with the HK AIO board + NEO-6M GPS module, Multiwii 2.3.

I'm wondering what sort of performance to expect from the "hold" modes... I have tried them and generally am unimpressed at present; I'd like to know if I should try tweaking, or if my expectations are just a bit unrealistic.

Altitude hold - I've tried enabling this mode after sitting for several seconds in a stable hover. Its clear that the mode is operating, but I see swings of up to 2m or more in altitude thereafter. Its not tied to craft motion or control input - even just maintaining position there is lots of altitude variation. For reference the HK board has a MS5611-01BA01 barometer. When sitting dead on a table and watching Barometer output with Multiwiiconf the barometer is very stable, but there does tend to be a "blip" every 10 seconds or so... i'm not sure what units the GUI is displaying, but the value changes from ~1.5 steady to ~4 during the blip, before returning to a steady ~1.5.


the alt coming from the baro can very very steady when sitting dead on a table.
but it can be very unsteady in the air due to:
- air perturbation
- sun light
- non constant power of the board <- I think this point is too much neglected, a clean BEC is important for good results
you should expect +/- 20cm when well tuned


GPS hold - I've had worse luck with this one. After enabling, again from several seconds of static hover, the craft will slowly migrate and then eventually flies determinedly in some direction (random? can't tell...). I usually lose my nerve and deactivate the mode when it reaches about 10m away (my flying area is not enormous). MultiWiiConf indicates that the GPS is connected properly and I am achieving a good fix; the LED blinking also indicates 5 or more sats acquired. I haven't tried the GPS home mode yet, that one scares me more :).

I'm hoping that some tweaking will improve both of these, yes?

GPS needs a proper MAG calibration.
But even with it, you will never have the perf of a naza board (at least currently ;) )
expect a +/- 2m deviation

Adrianm1972
Posts: 7
Joined: Mon Sep 23, 2013 10:24 pm

Expected performance of altitude hold and GPS hold modes?

Post by Adrianm1972 »

I spent a lot of time reading about alt hold. I tested a lot of PID values and after reading a few thread settle on a common set of values that worked better than any thing I had tested.

P=3.5 I=0.010 D=7

With these settings and and a chunk of dark grey filter foam wedged between my Baro and GPS board I can hold alt +/- 30cm under just about any condition.

My GPS will hold my quad inside a 2M circle. I think that is pretty darned good.

enderffx
Posts: 9
Joined: Fri Feb 28, 2014 12:51 pm

Re: Expected performance of altitude hold and GPS hold modes

Post by enderffx »

Hi !
I am fighting with those two myself (who doesn't ?).
I can only say that the ***same*** Hardware with an APM (even using same GPS with the very same settings, UBX+38400 baud). Was able to hold ALT within 30cm and position within 50x50x50 cm with stock values on Arducopter 3.01...

After a whole lot of tinkering i am now at +- 50cm with ALT Hold and +- 3m with Pos Hold if all goes well with some possible strikes of fate when ACTIVATEING one of these modes.

So i know the Hardware can do better but my PID tuning skills together with the Software lacking can not do better right now.
Thats frustrating especially when comparing to STOCK values on APM.
I believe APM (using that board now in my Y6B) has surpassed Naza (at least the ones i have seen in my area) in terms of stability and Autotune blows away everything.

So: YES its hard but YES i do think we can get better with better PID's but i guess the core math is flawed in MultiWii...

I will try BradWii on a quad, thats not running on a Hex right now. I started to edit Brads code for Hex but i am no Arduino I/O wizard (to say the least) and i am stuck at the question how to control motor 5&6 from an I/O point. On MultiWii there is a flag to free two ports on my little HK MultiWiiSE board but i did not get there with BradWii until now.
Bradwii seems to have a much better PID aproach and Autotune.
I wonder why the core MultiWii ignores the better approaches of Brad and/or Arducopter...

Greetings,

Ender

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: Expected performance of altitude hold and GPS hold modes

Post by Alexinparis »

enderffx wrote:Hi !
I am fighting with those two myself (who doesn't ?).
I can only say that the ***same*** Hardware with an APM (even using same GPS with the very same settings, UBX+38400 baud). Was able to hold ALT within 30cm and position within 50x50x50 cm with stock values on Arducopter 3.01...

yes, that's not a secret, it's a fact and I have no problem to say it ;)
APM GPS PH and APM alt hold is better than what we have in multiwii.

So: YES its hard but YES i do think we can get better with better PID's but i guess the core math is flawed in MultiWii...

I don't think so.
I think it's a question of good algorithm and good tuning.

Bradwii seems to have a much better PID aproach and Autotune.

For Autotune, yes for sure, as there is nothing in multiwii about it.
PID approach "seems" : supposition ? fact ? based on what ?

I wonder why the core MultiWii ignores the better approaches of Brad and/or Arducopter...

Nothing is ignored and I would be glad to see a code suggestion to improve anything you think useful.

enderffx
Posts: 9
Joined: Fri Feb 28, 2014 12:51 pm

Re: Expected performance of altitude hold and GPS hold modes

Post by enderffx »

Hi Alex.
By flawed math i MEANT the quality of the algorithm, sorry to be misleading. So: possily no errors, just a not-so-good approach.

As i did not have the time to put a Quad together (just for testing BradWii) i only base my opinion on what others say in he main BradWii thread.

As for the code:
My only effort will be to make it work with my second Hexa if my Quad testing is okay. People who know the MultiWii code are 400% more efficient than i to port MultiWii autotune and / or PID workings.
I ony have very limited time for my hobby and if MultiWii proves to be too bad i will buy a second APM.

I know thats not nice and i wish i could contribute more but i have to focus and currently its my job and my children. I code for a living and only limited in my spare time. And i am a real NOOB on Arduino as well ;-)
I had an Arduino sitting here for 1 1/2 years unpacked with dozens of shields. I donT have a time to be of much help.
Sorry if i sound too harsh sometimes, thats probably my attitude of bringing home the point, its not in disrespect of the work that has been devoted to MultiWii.
And so i just wonder "IF i would invest so much of my spare time, why not look at BradWii and/or APM Autotune & PID handling and learn from it.
But probably the guys involved so deep can PID tune a new craft in 5 Minutes so they do not see a point in implementing autotune and different PID behaviour.
But i cannot do that and have no time or will to ever learn.

Greetings,

Ender

Post Reply