The height of new TDOA2 is inconsistent

All discussions related to the Loco Positioning system
Post Reply
Person
Beginner
Posts: 8
Joined: Wed Jan 10, 2018 5:40 am

The height of new TDOA2 is inconsistent

Post by Person »

Hi,I find the height of the latest version of TDOA2 is inconsistent on the space, I set up eight anchor point in the space, a rectangle of 3.6 X 3.6 X 2.65 m, I found that the greater X axis the height is lower;the greater the Y axis, the lower the height.The difference is lager to 30 cm. How can I improve this problem?
arnaud
Bitcraze
Posts: 2538
Joined: Tue Feb 06, 2007 12:36 pm

Re: The height of new TDOA2 is inconsistent

Post by arnaud »

Hi,
How did you find these inconsistency, do you have a motion capture setup to track your Crazyflie positions?

I would first check that the anchors position is setup correctly. Then, if the offset is consistent, it could maybe be accounted-for when sending set-points (ie. basically calibrating the system).
Person
Beginner
Posts: 8
Joined: Wed Jan 10, 2018 5:40 am

Re: The height of new TDOA2 is inconsistent

Post by Person »

Hi!
Thank you for your answer.I have traced the Quadcopter and get the position from the python script, the data is right which close to setposition. But I measured the height is inconsistent so much.I find this inconsistent when I using TDOA and TWR , and I find the inconsistent of height is larger when I using the TDOA. I have watched yours video and find it perform so(The height is inconsistent).I suspend this may related to the location of anchors or the order of anchors' data sending to the kalman filter ?
arnaud
Bitcraze
Posts: 2538
Joined: Tue Feb 06, 2007 12:36 pm

Re: The height of new TDOA2 is inconsistent

Post by arnaud »

One of the error source is the LPS is a time-stamping offset in the DW1000 chip that seems to be mainly correlated with receive power. This means that getting close to the anchors will change the measurement accuracy. Usually though the measurement from other anchors in the system will compensate for it.

As for the anchor order, we have observed that it can help in TDoA mode to swap a bit the anchors to have more TDoA measurements in Z. For instance, our current system had 0123 on the floor and 4567 on the ceiling, we swapped 2 and 6 to have more TDoA measurements in Z.

If the measurement error you are observing are static, you should be able to calibrate the system and compensate for it.
Post Reply