Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

sfolgar
Hello,

we are trying to fly our AscTec Pelican with the asctec_mav_framework, by using a combination of amcl and px4flow as position input for the HighLevelSSDK.
We successfully tested our setup with the provided waypoint_server a few times, being able to fly and land through waypoint commands.

The problem we are facing now is that, sometimes, when we switch the serial interface to the "on" position the motors start spinning at maximum speed, sending our unit straight to the ceiling.
At first be blamed our Z input, but after a few test we think there has to be another cause.

Did anyone experienced the same behaviour? Any ideas on what sanity checks we should do before switching the serial interface "on" or how we could trace this issue?

Best regards,
Sergio
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

MarkusAchtelik
Hi, 

Did you go through the sanity checks here:

Is it possible that there is a setpoint somewhere in "space" ;) ?

You can see the actual output to the low level controller here (serial port switch has to be on and flight mode switch to "GPS", just do not start the motors):
This needs adaptation to rqt_plot, however, the channels are the same - Happy to accept pull requests, and there is actually one here, which never got ready to merge:

Hope that helps!
Markus 

From: "sfolgar [via asctec-users]"
Date: Tuesday 19 May 2015 20:39
To: Markus Achtelik
Subject: Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

Hello,

we are trying to fly our AscTec Pelican with the asctec_mav_framework, by using a combination of amcl and px4flow as position input for the HighLevelSSDK.
We successfully tested our setup with the provided waypoint_server a few times, being able to fly and land through waypoint commands.

The problem we are facing now is that, sometimes, when we switch the serial interface to the "on" position the motors start spinning at maximum speed, sending our unit straight to the ceiling.
At first be blamed our Z input, but after a few test we think there has to be another cause.

Did anyone experienced the same behaviour? Any ideas on what sanity checks we should do before switching the serial interface "on" or how we could trace this issue?

Best regards,
Sergio


To start a new topic under asctec-users, email [hidden email]
To unsubscribe from asctec-users, click here.
NAML
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

sfolgar
Hi Markus,

thank you very much for the hints.

Today we went through all the sanity checks. The only problem we found was bias convergence, as it doesn't behave as described. This is the plot from one of our flights:



We also reviewed bag recordings from yesterday's flights and detected an odd behaviour on some debug channels related to Z measure.

 

This plot shows filtered z (blue), reference model z (red) and input z (cyan). Reference model oscillates heavily. The unit was tied to the floor, with motors "on" and height mode "on", then we turned the serial switch "on" (~10") and experienced the full-thrust issue. Note that our height sensor (px4flow) starts at 0.30m.

Now, this is a plot from one of today's waypoint flights:



Same values: filtered z (blue), reference model z (red) and input z (green). Reference model z shows no oscillation and everything went smothly.

This last image also shows another of our problems: during the first seconds of flight we experience some "bouncing" behaviour. Could it be due to our luenberger parameter configuration? We use the default values.

Besides of the reference model Z oscillation, we observed similar oscillation on the debug channels 37 (reference v_z)  and 38 (¿?):



To sum up, these are our current "suspects":

- The odd reference model Z oscillation.
- IMU aligment. Our Pelican is in an X-configuration, instead of +-configuration. Do we have to tweak something in the asctec_mav_framework in order to allow this?
- Waypoint flights at 0.3 m/s work fine, but if we use a higher speed like 0.6 m/s the unit is not able to hover smoothly and the vibrations increase. Next image shows the output to LL controller. Could it be due to luenberger parameters? X-configuration?



Again, thank you very much for your advice.

Best regards,
Sergio




Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

Adrián
Do you have any update on this?

We are trying asctec_mav_framework and have the same full-thrust issue.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Serial interface switch "on" sends full-thrust command (asctec_mav_framework)

sfolgar
We think that it was due to human error in our initialization procedure which produced the Z reference model oscillation. We are now checking that the Z reference model has a positive value everytime we launch the unit and it has never happened again.

The X-configuration could not be the cause, you just have to handle it when you use the imu readings, and the hover vibrations were caused by the delay of our height sensor.

Regards,
Sergio
Loading...