Bitcraze Forums

Hi all,

I consider using the Crazyflie together with the Lighthouse Positioning System for an experimental setup. But before buying it I just want to check if I understand the theory behind it correctly (just the general concept not in depth).

On the setting up page [https://www.bitcraze.io/documentation/r ... etting_up/] it is said that the drone can estimate its position using only one basestation and the extended kalman filter. This is important to me as in the experimental setup the drones might loose line of sight to one of the basestations.

So here is my question.
One basestation swipes two orthogonal planes across the space. When the planes hit the drone it will know that it is on the intersection line of the planes but it does not know on which point of the intersection line it is. So when the drone stands still or moves along this intersection line will I see a drift of the estimated position in the direction of the intersection line?
Second, if the drone moves otherwise will the drift disappear because intersection lines at different points are not in parallel?

Best regards,
devjalx

Hi!

One basestation swipes two orthogonal planes across the space. When the planes hit the drone it will know that it is on the intersection line of the planes but it does not know on which point of the intersection line it is.

Yes, this is correct for one sensor. The Lighthouse deck has 4 sensors though and this makes it possible to calculate the full position.

So when the drone stands still or moves along this intersection line will I see a drift of the estimated position in the direction of the intersection line?

No you should get a the full pose (position + orientation)

Note 1: If you use a Lighthouse V1 base station the light planes are orthogonal, in lighthouse V2 they are slanted +-30 degrees, but the result is similar.

Note 2: Even though it works with one base station, two is more robust and converges faster. The use case you are describing should work just fine.

Loss of both base station can usually also be handled for short periods of time (say <0.5s) as well, trusting data from the IMU only

Hi!

thank you for your answer. This helps me a lot.

Best regards,
devjalx