DEV call 09 Jan 2017 3pm PST = 09 Jan 2017 2300 UTC


  • AC3.5 release preparations.
  • What PRs do we need to get into master before the first release candidate?


  • RC_CHANNEL rewrite,

Github / issues

Disable AP_Param debug?
Discuss the underlying timing problems

onboard compass cal with pixracers and esp8266’s goes non responsive/doesn’t communnicate over all links.

decide on top contributor for Dec 2016

  • MAVROS integration with ArduPilot.
  • Python framework for ArduPilot Firmware
  • SITL multi-port extension
  • Easy USB interface for OverrideRC (very stable interface needed I believe for roll and other functions)
  • “ArduCore” with separated levels of development; (Ardupilot split into 3 levels or so; first level meant only for stable hovering; second level for movement; third level for failsafes, EKF, checks etc)

ArduPilot Dev Call 2017-01-10

Agenda: DEV call 09 Jan 2017 3pm PST = 09 Jan 2017 2300 UTC

10:03am: Happy new years all around

10:04: Randy and Copter AC3.5
3.4.4 went out over last day or two
Probably last release of 3.4.4
3.5 should be closed within 2 weeks
Will use a 3.5 milestone tag to track what’s going in

Suggestion that expanding to 16 option channels should be in 3.5
RC channel split to go in
Autotune position hold may go in
EKF bugfixes too
May push rc1 out without some of these in two weeks
Question as to whether EKF3 should be default
Probably not

10:16 tridge and RC channel rewrite
Significant change; input and output are separated

Separate trims for input and output channels
RC5_FUNCTION have been renamed to SERVO5_FUNCTION (e.g.)
This may be changed to e.g. OUT1_FUNCTION
No magic channel numbers!
The current “_REV” parameter (+1/-1) to reverse a channel will be replaced with boolean _REVERSED
Singlecopter, coax and friends are simplified
Same parameter table for all multicopters
Same build for all but heli
Arducopter and arducopter-heli
Less build/test load on servers
Parameter migration should work
Binaries have gotten bigger
7kB for copter
Very little for Plane
GCS will only need to choose between heli or “normal”
Also simplifies QuadPlane builds
(Tilt-tricopter plane apparently works now)
Auto servo-trimming on planes
Might be possible to auto-trim swash on heli
Basically Disco auto-trim motivated this entire patch
Pervasive change was “motors.” becomes “motors->”
*motors is filled in once at boot
Possibility of a generalised mixer
GCS could upload a mixer
Measure angles of motors and their offsets, do calculations, upload
Dedicated weird-flaperon code can go away

10:44 triaging of PRs

Correcting barometer offset using GPS (integration over time)
Randy things this should be in the EKF
Thinks interpreting sensor data should be within the EKF
Hiding information from Paul might be bad
Tridge disagrees
EKF works on seconds-scales, not the 10s-of-minutes scales like this adjustment
Paul has too much stuff happening already!
Already do automatic compensation outside of EKF (e.g. airspeed sensor on voltage, others on temperature)
Will ping Paul to see if he’s happy for this to be outside the EKF

Apparently enabling killed
That’s a separate problem that should be addressed
Inclination not to put that in; tridge turned it on because he was fielding too many requests
Should add a comment so it is clear that this “debug” code should always be on
UTC vs GPS time, subtracting 15 seconds
“Time” command in mavproxy prints both system_time message from vehicle and the GCS system time; they should match
They weren’t, so tridge added the 15 second hack
GLONASS you won’t get a proper time offset for up to 12.5 minutes
Removing the 15 second constant
Could do it at compile
Could be a parameter
SYSTEM_TIME has two fields that let you relate the uptime vs the coordinated time
Apparently gps time-of-week etc jump when the leap-seconds correction comes in?!
You can’t rely on ITAUs
Setting GPS-only is recommended for RTK
Maybe only let a lock happen when we have at least one GPS satellite visible, regardless of constellation we use for getting a position fix
Alternatively, we must always use the time base of the constellation providing position fix

11:36 M8P

Should work but not tested
Still some small amount of work required
Now get a GPS status of 6
RTK float vs RTK fix
Some problem with the M8P SITL

11:43 Best Contribution Award

Also Olivier’s work on
tridge is making himself ineligible
Decided to do a poll on the mailing list
Also something about extra prizes from other companies
Srxl receiver page on Wiki
Vast amounts of documentation
Borrowed heaps of stuff from MultiPlex and documented each of them as to how to make them work with ArduPilot

11:58 Compasses
PixRacer with ESP8266 locks up and generally misbehaves when doing compass calibration

12:02 Vision stuff with Randy
Going well, two different groups
One using ROS, one not
Working with Z-camera and depth information
Testing today or tomorrow, video in a couple of days