Page 1 of 1

Using Lighthouse swarming with some question?

Posted: Mon Nov 30, 2020 8:34 am
by aqqz
Hello, I am a student, and now I want to use 12 crazyflies(one channel)to do a swarming flight show. I use two Lighthouse V1 base stations for the positioning system. I encountered many problems during this process. I would like to ask the official for such problems:

1. Is there a height limit for the base stations? Our planned maximum flying altitude is above 2.5m. We use a tripod to place the base station at a height of about 3.2m, but found that flying above 2m is very unstable (the horizontal distance between the two base stations is about 5m)

2. The reference code ishttps://github.com/aqqz/crazyflie-lib-p ... equence.py, but I found that this script cannot be run directly. The cluster will be turned over directly. The console print step x but the cf don`t fly (so big delay). My solution is to let each cf use hlcommander to take off once(i write a srcipt named prepare.py), because i see the demo using hlcommander, and then the cluster can fly. My firmware And cflib version are both 2020.09

3. the kalman fliter`s output variance often over threhold so that the cf can`t fly , if i restart the cf it will work sometime, i can`t explain this symbol is hardware problem.https://github.com/aqqz/crazyflie-lib-p ... #L105-L143

4. Even if swarming is successful, there will be a high probability of a few planes crashing midway, and the stability of the entire system is poor.

These problems have troubled me for nearly two months, but there is still no good solution.

Re: Using Lighthouse swarming with some question?

Posted: Mon Nov 30, 2020 10:42 am
by kimberly
Hi!

Let me try to answer your questions.

1- That is because the sensors are mounted flat on the LH deck. It is difficult to say but in general it is depended on how much the basestations are pointing downward. If the signals of the angels are being received at a too big of an angle, it will have difficulties. Our basestations are pointed downward with 45 degrees and we have difficulties at above 2.5 meters (about 0.5 meters from the height of the basestations), but we usually fly around 1.0-1.5 meters. Our basestations are about 4.2 meters apart in the diagonal.

2- It is a bit unclear what you mean with cluster. Are you talking about the threads?

It seems that the sequence script is using highlevel commander for take off as long as take off is part of the sequence. This first take off does not work for you?

There were some additions to the HL commander recently so maybe this broke some parts of the script, but since you are using the 2020.09 this should be the case.

3- It is an safety issue but you can turn it of or higher the threshold. It will just have an higher risk that it might have not converged, but with lighthouse the changes are lower for that anyway (as long as the crazyflies can receive the angles properly).

4- Have you tried to have 2 crazyradio PAs? We already are having trouble with more than 9 crazyflies. Also the crashing can also be due to the Mellinger controller, which is unforgiving if the tracking is just a bit off. Please uncomment this line to make the crazyflies fly on PID only. That should improve stability.

The crazyflie is a opensource development product and the Lighthouse positioning system is still in early access and is going through a lot of changes. So it is not strange that you are experiencing problems when working with these parts which are still very much in development. Please communicate anything that you are having trouble with with us on the forum. We monitor it everyday and always willing to help out!

Re: Using Lighthouse swarming with some question?

Posted: Mon Nov 30, 2020 1:17 pm
by aqqz
Thank you for your reply. Regarding the second point, I mean to let each cf use hlcommander.takeoff one by one, and then the swarm thread has

a high probability of success, provided that the kalman fliter output of each cf is below threshod. Sometime restart the cf can solve the kalman

fliter problems,so i need to restart the cf by software, for this, I am writing a script prepare.py which is used to autonomous test before swarm.

If there are new discoveries in the future, i will continue to contact you.

Re: Using Lighthouse swarming with some question?

Posted: Sat Dec 12, 2020 2:29 pm
by aqqz
We found that the cluster is easy to fall after adding the LED deck, and the heavier the deck is, the more unstable the aircraft will be. Is there any adjustment to the weight of the drone in the firmware regarding the flight control algorithm?

Re: Using Lighthouse swarming with some question?

Posted: Mon Dec 14, 2020 10:37 am
by kimberly
There is this weight define as a compile flag that you can put in your config.mk.

Have you tried also tried to not activate Mellinger as the controller and use PID only?