Random Flyaways and Flips

Topics related to the Lighthouse positioning system, configuration and use
Post Reply
n_dronevent
Beginner
Posts: 3
Joined: Tue Nov 02, 2021 1:25 pm

Random Flyaways and Flips

Post by n_dronevent »

Hi,

we are using the Lighthouse positioning system and are experiencing random flyaways and/or flips on every other flight. This happens both when using the cf client and a modified version of the synchronized sequnce example script. When this happens the indicator led on the lighthouse deck turns orange.
When using the cf client and doing a prop test we get "Kalman perdiction rate low" warnings with values around 95.

The issue happens on the latest firmware aswell as older firmware.
We have plotted lighthouse.angle0x, lighthouse.angle0x_1, lighthouse.angle0x_2 and lighthouse.angle0x_3. The values seemed to be at a stable distance to each other, so we have ruled out that the problem is in connection to yaw. But when the drone starts to drift, the values freeze. We also have tried differend lighthouses, so the issue doesn't seem to be with the lighthouse hardware.

There is an example video where both a flip and a flyaway happen: https://www.youtube.com/watch?v=XbePpd3uS0c

Regards Nils
jonasdn
Expert
Posts: 116
Joined: Mon Mar 01, 2021 3:13 pm

Re: Random Flyaways and Flips

Post by jonasdn »

Hi Nils!

This is curious indeed! I would like one clarification, do you see the both the flipping and the flyaways using both the swarm examples and the cfclient?
Or is that only in regards to flyaways?

We have seen the flips here at Bitcraze as well, and are investigating them, we see them mainly in swarm control.

Jonas
n_dronevent
Beginner
Posts: 3
Joined: Tue Nov 02, 2021 1:25 pm

Re: Random Flyaways and Flips

Post by n_dronevent »

Hi Jonas,

Thanks for the reply. The flyaways happen both during operation with the cfclient and the swarm-script. The flips only happen when using the synchronized sequnce script. It seems to beirrelevant how many drones are controlled by the script, because the flips happen when the script controlls just one drone.

Nils

Edit:
When the flyaways occure and the indicator led on the lighthousedeck turns orange, it doesnt turn back green until the drone is restarted.
Edit2:
We have tried to move the lighthouses closer together 3m intsead of 4.5m and at first, when testing with only 3 drones everything seemed to work fine. When we added a fourth drone things went bad again(https://youtu.be/HM5QRZpkl7I). The flip on startup of the first drone was probably because a prop was not attached properly. What we can't explain is what happend to the second drone.
jonasdn
Expert
Posts: 116
Joined: Mon Mar 01, 2021 3:13 pm

Re: Random Flyaways and Flips

Post by jonasdn »

Thank you!

This matches what we have seen! I am in the process of trying to setup some tests to catch this out! All findings you can share would help!
Would you be able to test this with some other positioning system? To see if it is Lighthouse specific?

Jonas
v_dronevent
Beginner
Posts: 1
Joined: Tue Nov 02, 2021 1:24 pm

Re: Random Flyaways and Flips

Post by v_dronevent »

Hi there,

Different person from the same team here. We sadly do not have the resources to test other positioning systems.
This problem is most likely lighthouse specific. The main cause of flyaways seems to be the lighthouse deck. At a certain point (usually after takeoff) the indicator turns yellow and doesn't recover until the drone is restarted.
What should be mentioned is that we use new motors, which may be drawing more power.
Unfortunately in our testing this issue has been very inconsistent and frustrating to diagnose.
(My personal hypothesis is that some of our batteries might have gone bad, but even tests with new batteries are inconclusive.)
If you have any ideas for tests we can perform with the lighthouse system please let us know.

Cheers V
arnaud
Bitcraze
Posts: 2528
Joined: Tue Feb 06, 2007 12:36 pm

Re: Random Flyaways and Flips

Post by arnaud »

Hi,

We have seen that the FPGA on the lighthouse deck can be pretty sensitive to low power spikes, it is possible that you are putting the fpga in a bad state by using powerful motors and bad batteries. We have a ticket about the issue: https://github.com/bitcraze/crazyflie-f ... issues/695.

Maybe one test would be to look the voltage on VCOM with an oscilloscope while running the motors manually (from the client for example). This would allow you to see how close you are to the 2.6V that seems to be the limit for the deck.
n_dronevent
Beginner
Posts: 3
Joined: Tue Nov 02, 2021 1:25 pm

Re: Random Flyaways and Flips

Post by n_dronevent »

Hi,

thanks, this seems to be the root cause of our issues. Would you say using a higher capacity battery could fix this problem? could something like this work? We would like to use the stronger motors because we have to use a propguard for added safety. Or would you say we are better off switching to the bolt platform?
tobias
Bitcraze
Posts: 2272
Joined: Mon Jan 28, 2013 7:17 pm
Location: Sweden

Re: Random Flyaways and Flips

Post by tobias »

The battery you are linking to will most likely work. Possibly the 380mAh version is even better as it is a bit lighter. It is also thicker so possibly the long pin headers are needed.
Post Reply