Strange LPS issue with ROS

All discussions related to the Loco Positioning system
Post Reply
sali0
Beginner
Posts: 24
Joined: Wed Oct 19, 2016 7:53 pm

Strange LPS issue with ROS

Post by sali0 »

Hello Bitcraze,

My team and I have been flying up to 3 drones for a month now and have been testing this system to try to work out as many bugs as possible. However, there has been one issue we have been seeing that we cannot understand what is causing it. It has to do with the terminal when you launch ROS. You can see the issue in the picture below. It shows in the terminal that anchors 6 and 7 parameters are not found, although it still shows all 8 anchors in RVIZ. Do you know what may be causing this?

We also tried launching 4 drones today, and had huge issues getting them to hover correctly. They took a very long time to calibrate their actual initial position, and then when we launched, they flew very erratically all around the room. We are suspecting that it may be because of the issue above that because there are less anchors in the system than there actually is, the drones position is being diluted without the last two anchors.
Screenshot from 2017-03-26 16-36-03.png
arnaud
Bitcraze
Posts: 2538
Joined: Tue Feb 06, 2007 12:36 pm

Re: Strange LPS issue with ROS

Post by arnaud »

Hi,

This is most likely due to the fact that by default the Crazyflie is compiled with support for 6 anchors: https://github.com/bitcraze/crazyflie-f ... deck.h#L55. If you change it to 8 things should work properly.
sali0
Beginner
Posts: 24
Joined: Wed Oct 19, 2016 7:53 pm

Re: Strange LPS issue with ROS

Post by sali0 »

Thank you for the tip! I hope to try it out for our next test. Out of curiosity, but do you and any other users have to do this change as well? I have not seen this documented anywhere. Is this problem unique?
arnaud
Bitcraze
Posts: 2538
Joined: Tue Feb 06, 2007 12:36 pm

Re: Strange LPS issue with ROS

Post by arnaud »

This problem is not unique but recent, so the when we fly in a TDoA system we have done the change manually.

The context for us is that we are currently focusing on stabilizing and finishing the LPS for release (ie. out of early-access) and TWR with one copter is going to be the first 'stable' mode. This is certainly why this bug slipped through. We are not forgetting the other modes, we just want to stabilize one :-).

The number of anchor should certainly be added to config.mk and documented.
sali0
Beginner
Posts: 24
Joined: Wed Oct 19, 2016 7:53 pm

Re: Strange LPS issue with ROS

Post by sali0 »

Thank you for reply! I hope my reporting these issues on this forum is a good way of documenting these bugs.Your help has been incredibly invaluable to my team and myself, and I hope we are helping you guys hash out these bugs as well!
arnaud
Bitcraze
Posts: 2538
Joined: Tue Feb 06, 2007 12:36 pm

Re: Strange LPS issue with ROS

Post by arnaud »

Yes reporting issues is a good way to keep the system getting better :-). I created a ticket on github for it: https://github.com/bitcraze/crazyflie-f ... issues/212. If you ever find something that is most certainly a bug feel free to push a ticket in github directly ;-).
Post Reply