TDoA3 with sample code

All discussions related to the Loco Positioning system
Post Reply
albussimba
Beginner
Posts: 7
Joined: Sat Sep 01, 2018 3:31 pm

TDoA3 with sample code

Post by albussimba » Mon Oct 08, 2018 5:03 pm

Have anyone tried to run the sample python codes with TDoA3? It seems like it is taking ages to get out of

Code: Select all

wait_for_position_estimator
and when it does the position given by the estimate is stuck at pos: (-100.0, -100.0, -100.0). Have anyone had a successful flight with it yet?

arnaud
Bitcraze
Posts: 1826
Joined: Tue Feb 06, 2007 12:36 pm

Re: TDoA3 with sample code

Post by arnaud » Tue Oct 09, 2018 5:39 am

We have had plenty of successful flight at the office but always in the same environment so it is interesting to have feedback from different setup. In our setup we have 8 anchors placed at the corner of a space of ~3x4x3m.

How do you setup the anchors?
What firmware are you using? It looks like your Crazyflie is not ranging at all and its position is just drifting until it reaches hardcoded limits.

albussimba
Beginner
Posts: 7
Joined: Sat Sep 01, 2018 3:31 pm

Re: TDoA3 with sample code

Post by albussimba » Tue Oct 09, 2018 11:16 am

I'm using 6 anchors only. But I'm in a small room ranging abt 3x1.2x3m? Was wondering if it has anything with me setting the anchor positioning. But I don't really have anyway to check because the new cfclient update have the buttons greyed out.

arnaud
Bitcraze
Posts: 1826
Joined: Tue Feb 06, 2007 12:36 pm

Re: TDoA3 with sample code

Post by arnaud » Thu Oct 11, 2018 6:47 am

For the LPS tab in the client I am guessing you might have this problem: viewtopic.php?f=16&t=3168. See if the solution I posted works for your.

Though in your case the estimate seems to be drifting completely which normally indicate that you have no TDoA measurements pushed to the EFK. Are you running the latest git version of the firmware and are all your anchor running the latest git version of the Node setup in TDoA3 mode?

albussimba
Beginner
Posts: 7
Joined: Sat Sep 01, 2018 3:31 pm

Re: TDoA3 with sample code

Post by albussimba » Fri Oct 12, 2018 2:17 am

that solved my problem actually. I one of the anchors was addressed wrongly. But the error still seems a lot larger than in TWR mode. its abt +/- 20 cm in TDoA 3 vs abt +/- 5cm in TWR is there anyway to reduce this?

arnaud
Bitcraze
Posts: 1826
Joined: Tue Feb 06, 2007 12:36 pm

Re: TDoA3 with sample code

Post by arnaud » Wed Oct 17, 2018 8:28 am

We have just pushed a new outlier filter to git that should improve the positioning of TDoA3.

It is expected for TDoA3 to be more noisy. TDoA is geometrically going to be less precise and it will be affected by clock noise more than TWR. TDoA3 is adding to that collision between packets which adds noise and outlier. So TWR is theoretically the least noisy, then there is TDoA2 and then TDoA3. Though the new outlier filter seems to make a big improvement in our tests with TDoA2 and 3.

Post Reply