calibration problem

Latest News and Info about the AutoQuad development

Re: calibration problem

Postby LPR » Fri Jan 15, 2016 9:36 pm

I was able to fly today using the calibration parameters for the M4 board. I was very happy with the results. The large quad flew very well even though the quad was at room temps till I flew it outside in 5 degrees Fahrenheit weather.

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: calibration problem

Postby whizz101 » Sat Jan 16, 2016 12:47 am

i can't get past step 3. it just immediately outputs "nan" for everything. it does not even complete one loop.!

any ideas?

Tristan
whizz101
 
Posts: 82
Joined: Thu Oct 16, 2014 9:00 pm

Re: calibration problem

Postby whizz101 » Sat Jan 16, 2016 1:47 am

here are the graphs from my dynamic calibration

acc calibration.jpg


gyro.jpg


mags.jpg
whizz101
 
Posts: 82
Joined: Thu Oct 16, 2014 9:00 pm

Re: calibration problem

Postby LPR » Sat Jan 16, 2016 2:24 am

Tristan

We struggled with a number of issues and found it's possible to get to work. Just send me $20 and I will get the info you need. :D I'm thinking of putting a video together to show all the ways to not get it working. I know just about every way there is to not have it work.

The first thing to get right is to check the DIMU box in the Calibration GUI, if you are doing the AQ6 with the DIMU or the M4. The next thing is to be sure to use the Sim3_DIMU file if you have a DIMU board. The Sim3_DIMU has some parameters in it that need to stay in it when you add the parameters from all the steps to the All Generated parameters.

Now the steps in the wiki are good but you need to know a few things more.

Step one does not change if you run more than once.

Step two you can get the MAE lower by running it a few more times, sometimes. Just a little lower can make a big difference probably, I think, maybe, you tell me what you think.

Now step three, make sure you have the right sign on your inclination parameter and you have it in the All Generated Parameter file.

Now for the tricky part, if you have a good static and dynamic file it might work without seeing a "nan". Most of the time you will see more "nan"s than ever you'll ever want to see. The trick is to click on the Start sim button and then quickly place the cursor over the abort button. Now find as fast as you can the Dynamic MAE value which should be decreasing and if it increases hit the abort button or after you see too many (nan)s, click the abort button before you think that the MAE just might be increasing.

If you have deleted some parameters in the All Generated Parameter file then you may need to download a new Calibration IMU Gui. We did that plenty of times.

That's enough for now.

Good luck

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: calibration problem

Postby zhlearnaq » Mon Jan 18, 2016 1:03 pm

Larry
Congratulations, could you give me your dynamic data and static data, as well as your parameters? I will check your work with Matlab, and maybe I could find my problem by comparing with your job.
Here is my mail - 2568489626@qq.com.
Alice
zhlearnaq
 
Posts: 43
Joined: Wed May 14, 2014 10:47 am

Re: calibration problem

Postby LPR » Mon Jan 18, 2016 5:45 pm

Alice

I'm using the same static log as before.
The dynamic file is not the best because the 4 kg quad is not that easy to keep moving smoothly as your arms get tired.

I find that now the quad flies well when I move it from a warm house to 0 degrees outside but I have the altitude of the quad that keeps increasing so I'm doing another calibration now. The z accelerometer bias must not be correct. I have more to learn.
Here's links to the files you want.

https://www.dropbox.com/s/rzs7ihbt9kuwp ... 3.LOG?dl=0
https://www.dropbox.com/s/yb2n1j6ekjidp ... ..LOG?dl=0
https://www.dropbox.com/s/7zrlhw2cvhzc4 ... e.txt?dl=0
Larry
Attachments
4.jpg
3.jpg
2.jpg
1.jpg
Last edited by LPR on Tue Jan 19, 2016 5:44 am, edited 2 times in total.
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: calibration problem

Postby zhlearnaq » Tue Jan 19, 2016 4:19 am

Larry

Thank you, I will check the work of your calibration now, later i will upload the pictures.

Alice
zhlearnaq
 
Posts: 43
Joined: Wed May 14, 2014 10:47 am

Re: calibration problem

Postby zhlearnaq » Tue Jan 19, 2016 2:08 pm

hello Larry
I have checked your data by Matlab, The following maps are plotted by matlab, the blue one means calbrated data,your work looks much better than mine.
Alice
Attachments
acc.png
acc
gyro.png
gyro
zhlearnaq
 
Posts: 43
Joined: Wed May 14, 2014 10:47 am

Re: calibration problem

Postby LPR » Tue Jan 19, 2016 2:38 pm

Alice

I find that the static file are easy to get good data but I've done them often and I have a good place to keep the warming boards away from vibrations. I trim away the low temperature end of the file if there is two much movement in the data.

Learning how to dance the calibso is the hard part. I few dance lessons might help. I was very surprised that my first dynamic log was messed up by having very poor magnetometer data. Changing the location of where I did the calibso gave me a much better dynamic file.

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: calibration problem

Postby whizz101 » Wed Feb 17, 2016 7:07 pm

Hey Larry,

Thanks for your words of advice. I was able to get the calculations to finish, but when i checked the results they were not right.
I think i have identified a problem with the board and am going to retire it.
whizz101
 
Posts: 82
Joined: Thu Oct 16, 2014 9:00 pm

Previous

Return to News

Who is online

Users browsing this forum: No registered users and 2 guests