Skip to content
This repository has been archived by the owner on May 9, 2019. It is now read-only.

Fix DCM NaNs at startup #39

Open
kylc opened this issue Feb 21, 2015 · 6 comments
Open

Fix DCM NaNs at startup #39

kylc opened this issue Feb 21, 2015 · 6 comments
Labels
bug prio-high
Milestone
v1.0

Comments

@kylc
Copy link
Member

kylc commented Feb 21, 2015

Occasionally on the F3 platform (maybe others?) the DCM contains a bunch of NaN values. Not sure where they're coming from, but this breaks attitude estimation.
@kylc kylc self-assigned this Feb 21, 2015
@yoos
Copy link
Member

yoos commented Mar 12, 2015

Bad magnetometer or accelerometer vectors can mess up drift correction math. Observed on aerial v3.9 with uninitialized magnetometer.

@kylc
Copy link
Member Author

kylc commented Mar 12, 2015

"Bad"?

@yoos
Copy link
Member

yoos commented Mar 12, 2015

Zero vectors? I haven't actually looked at the numbers, so I said "bad."

@kylc
Copy link
Member Author

kylc commented Mar 12, 2015

Hmmm makes sense. NaNs could be originating from trying to normalize zero vectors. Maybe the sensor health checks sitting in #42 should also check that readings are non-zero.

@kylc
Copy link
Member Author

kylc commented Mar 12, 2015

Although I suppose a zero reading could still be valid. Maybe there's some kind of READY? register we can be checking.

@yoos
Copy link
Member

yoos commented Apr 5, 2016

Has this been fixed and verified?

@yoos yoos added this to the v1.0 milestone Apr 5, 2016
@kylc kylc removed their assignment Apr 6, 2016
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
bug prio-high
Projects
None yet
Milestone
v1.0
Development

No branches or pull requests
2 participants
@kylc @yoos