Autonomous flight MarvelMind Pixhawk 3 beacons

Frequent questions asked online, offline, in forums are answered here
Hendrik
Posts: 31
Joined: Wed Dec 07, 2016 12:29 pm

Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Hendrik »

I would like to share current status of autonomous flight with pixhawk and MarvelMind with 3 beacons.

Mission:
20170320 Mission plan APM.jpg
20170320 Mission plan APM.jpg (163.74 KiB) Viewed 8899 times
Flight:
https://youtu.be/_oDFz2LL3SY
(The ticking is the ultrasonic sensor)

Result APM:
20170320 APM result.jpg
20170320 APM result.jpg (44.48 KiB) Viewed 8899 times
Result MarvelMind
20170320 MarvelMind plot.jpg
20170320 MarvelMind plot.jpg (86.25 KiB) Viewed 8899 times
Scale of 1 square is 25cm
3 Beacons


Still some tuning to do when hovering
Flight is very accurate

Hendrik

amolad
Posts: 34
Joined: Wed Jan 18, 2017 12:56 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by amolad »

Hi Hendrik,
Are you injecting with NMEA protocol??

Hendrik
Posts: 31
Joined: Wed Dec 07, 2016 12:29 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Hendrik »

We indeed use NMEA from the hedgehog (mobile beacon) to the GPS port of the pixhawk.

amolad
Posts: 34
Joined: Wed Jan 18, 2017 12:56 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by amolad »

Ah okay! Thanks.

You aren't using the beacon parameters, are you? The BCN_* parameters (http://ardupilot.org/copter/docs/parame ... parameters)
I know they are only for Pozyx system but i havent checked if they some how work for marvelmind too :D

Hendrik
Posts: 31
Joined: Wed Dec 07, 2016 12:29 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Hendrik »

We are not using the BCN variables.

We use Arducopter 3.5rc2 (betaversion) and use the EK3 parameters.

Arducopter thinks that barometer and accelerometers are more accurate than GPS.
This is usually the case, but with Marvelmind it is not.

So we modified a lot of EK3 variables to minimise influence of accels and baro and to have more influence from GPS.

That seems to work now. As you can see in de picture of Marvelmind result, the path of the drone when flying is within a range of approximately 10 cm.
For our application this is enough.

We have to solve accuracy when hovering.
The tests carried out were done with a quadcopter. For our application we need a hexacopter, so we don't bother tuning the quadcopter more than it is now. Tuning hovering will be done with hexacopter.

Vava
Posts: 1
Joined: Wed Feb 22, 2017 7:44 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Vava »

Hello, Hendrik!

Could you post the EK3 parameters you changed? Or just export all the parameters?

amolad
Posts: 34
Joined: Wed Jan 18, 2017 12:56 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by amolad »

Hendrik wrote:We are not using the BCN variables.

We use Arducopter 3.5rc2 (betaversion) and use the EK3 parameters.

Arducopter thinks that barometer and accelerometers are more accurate than GPS.
This is usually the case, but with Marvelmind it is not.

So we modified a lot of EK3 variables to minimise influence of accels and baro and to have more influence from GPS.

That seems to work now. As you can see in de picture of Marvelmind result, the path of the drone when flying is within a range of approximately 10 cm.
For our application this is enough.

We have to solve accuracy when hovering.
The tests carried out were done with a quadcopter. For our application we need a hexacopter, so we don't bother tuning the quadcopter more than it is now. Tuning hovering will be done with hexacopter.
Wow awesome. I am trying to control my rover on latest stable with marvelmind. But i am always facing the issue of jitter on the map!! It moves around a lot. Did you face the same issue? I am injecting the NMEA messages! I am not sure how to fix this. Also i have issues with my compass! The translation between systems for the compass isnt so seamless! :(

EDIT: Btw how did you get the precision of the coordinates upto 7 decimal places? I can get 6 atm! And this is giving me a precision of +/- 11 cms

freaad
Posts: 27
Joined: Fri Dec 09, 2016 1:59 am

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by freaad »

Can you give some parameter name to reduce the Acc value and others.
I am using the mission planner.
My parameter set is default. I only change the GPS type5.
Can you give some tip for me about your setting.
I have some errors about ACC variance.

Smithy
Posts: 2
Joined: Thu Mar 30, 2017 11:44 am

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Smithy »

I'm attempting to do the same, if you could post your params that would be awesome!

Hendrik
Posts: 31
Joined: Wed Dec 07, 2016 12:29 pm

Re: Autonomous flight MarvelMind Pixhawk 3 beacons

Post by Hendrik »

Hi,

We have made little progress due to a few crashes because of unreliable compass data. (Resulting in heavy toilet bowls and crashing)
We are now rebuilding hexacopter and try to start testing again today.

Pixhawk compass is very difficult to get right. Each calibration gives different results. The compass slowly moves.
We can't get Pixhawk to use GPS heading only.

I will show EK3 params when the test is succesfull.

Most important settings are _NSE and _GATE params. We are aiming at maximum GPS reliability and less EKF addition.


Hendrik

Post Reply