remove magnetometer and barometer, position ok?

[justinleeyang]

hi,

If we remove the magnetometer and barometer, only with acc, gyro and LPS's distance data, we can get the crazyflie2's position ?
I see that the mag and baro is optional, the following code:

Code: Select all

#define SENSORS_ENABLE_PRESSURE_LPS25H
#define SENSORS_ENABLE_MAG_AK8963

if I remove the two lines code, maybe affect the position's result ?

[arnaud]

Hi Justin,

The pressure sensor and magnetometer are not used by default by the Kalman filter. The pressure sensor could help but is a bit noisy and the magnetometer is more questionable indoor (it will be affected a lot by external magnetic fields).

So the answer is YES, you get a position without magnetometer and pressure sensor, it is the position you have been getting with the Kalman filter in its default configuration.

/Arnaud

[justinleeyang]

hi, arnaud:

Got it, thanks!

In addition, default need 6 anchors, if I use 4 anchors, do crazyflie2 get the 3D position ? I know that the convex hull will be small, but I want to
get the position with as few anchors as possible, the minimum anchors is 4 ?

[arnaud]

For TWR (ToA) the minimum is 3, for TDoA the minimum is indeed 4, however in TDoA case you will get very bad positioning outside the convex hull. Depending on how you place your anchors you might be able to optimize the system to get good 2D estimates (a bit like gps: with overhead satelites/anchors we get good X/Y position and worse Z position).

[justinleeyang]

Got it, thanks!

[justinleeyang]

hi, arnaud:

another question, do you test the positioning accuracy with lps, acc and gyro throught EKF ?
maybe <=5cm ?

[arnaud]

Hi Justin,

I think some test has been done with a Kalman filter that indicates about +/-10cm accuracy in X/Y and a bit more in Z. Though I do not have the data available so I am not completely sure of it, but it a reachable ballpark. With some calibration and more anchor better accuracy might be reachable.

/Arnaud

[justinleeyang]

hi, arnaud:

about +/-10cm accuracy in X/Y and a bit more in Z, is the LPS's position method TWR or TDoA ?

[arnaud]

With TWR uising 6 anchors with no further calibration, there has been no formal measurement of TDoA yet.