CF as anchor

All discussions related to the Loco Positioning system
Post Reply
Posts: 14
Joined: Mon Aug 13, 2018 3:20 am

CF as anchor

Post by estrategico » Wed Nov 21, 2018 9:45 pm

I recently read in your blog a short description of the roadrunner device and is described as a CF2.1 and it acts as an anchor.
Can i use a CF2.0 as an anchor in a LPS using TDOA3 .
I like to evaluate how the use of more anchors improves the precision of the location especially in z axis, currently i have 8 anchors, one in each corner of a cube 4 x 4 mts.
We like to test before of make any purchase.
We also test the Flowdeck V1.0 with the flow disabled, only using it as Z ranger with good results, but the distance range is limited to 1,5 mts aprox and can't be used in combination with the led ring. (we broke a LED ring trying to use the hole trick) and this applies to V2.0
We suggest that in future versions of the LED ring take that into consideration, it is better a ring that a disc and is also better below than above.

We have been testing the new outiler filter in the last firmware, but it seems to perform better in TDOA3 than in TDOA2. can it be possible?

Posts: 2021
Joined: Tue Feb 06, 2007 12:36 pm

Re: CF as anchor

Post by arnaud » Mon Nov 26, 2018 12:57 pm


The roadrunner is acting as a tag not an anchor, it runs the same firmware as Crazyflie 2.0/2.1. That said, it is technically possible for Crazyflie to run the anchor code to behave like an anchor but it requires to port the anchor algorithm from the LPS node to the Crazyflie firmware. The way ranging algorithm are implemented should make it quite easy to port the algorithm, you might just have to mock the LPS node configuration to store anchor position and other stored info of the node.

Flow V2 will solve the range problem and yes we have been thinking about making a new revision of the LED ring with holes for the flow sensors but this is not planned yet.

As for the outlier filter: TDoA3 ranges in a more random order than TDoA2, which can be beneficial. Though TDoA3 has more outliers due to packet collision so I guess in your case the outlier detector does a great job removing the outliers due to collision. In our tests though we have always observed TDoA2 to be on par or better than TDoA3 in a 8 anchors setup.

Post Reply