Issue with LPS Deck

Post here to get support
Post Reply
thevoltmaster
Beginner
Posts: 3
Joined: Wed Jun 05, 2019 11:59 am

Issue with LPS Deck

Post by thevoltmaster » Wed Jun 05, 2019 12:34 pm

Hi I have 2 crazyflie copters both with LPS Decks. I'm able to successfully fly using the python 'autonomous sequence.py' file with one of the copters, however the other one always flips over and crashes while taking off. While I was initially worried it was my CF, I was able to rule that out because I was able to fly both copters with a bluetooth gamepad just fine.

I narrowed the problem down to the LPS deck because when swapping the decks between copters, the problem always follows the 'bad' deck. While looking at the Loco Position Tab in the CF GUI, I have a totally stable and valid looking position from both decks as far as i can tell, so Im not sure whats different about the deck that works and the one that does not.

I recently switched to TDOA3 and I'm wondering if there is any configuration settings, or firmware in the LPS Decks that I'm unaware of. Is there anything I can try to understand whats up with my decks?

Thanks!



kristoffer
Bitcraze
Posts: 109
Joined: Tue Jun 30, 2015 7:47 am

Re: Issue with LPS Deck

Post by kristoffer » Wed Jun 12, 2019 12:03 pm

Hi!

That is interesting! I have not seen this type of behaviour before. In theory you could for instance imagine a scenario where the antenna in one of the decks is not as good as it should be, or some other defect.

I suggest that you take a look at the buckets in the outlier filter that is used in tdoa2 and tdoa3. The buckets are basically counters that keep track of the number of TDoA samples that are deemed to be too far away from the current estimated position (and thus discarded). As a side effect, the buckets give a pretty good indication of the health of the system. Log the outlierf.bucket0, outlierf.bucket1 and so on. In a good system, all buckets should be 0, except bucket0 that might go to 1 or 2 from time to time.

If there is a difference in performance between the two decks, I would expect more errors in the outlier filter with one of them. Let me know what you see!

Post Reply