Dev Call Sept 18, 2017 2300 UTC

Copter

Sub
GCS Release

General

Probably a silly question, but i couldn’t find it so quickly:

  • Is anybody able to join these dev calls?
  • Are these dev calls announced somewhere?

The dev call is a standing weekly thing, and tries to stick to a 1hr time limit, so the agenda is important.
Details for the calls are here: http://ardupilot.org/dev/docs/ardupilot-mumble-server.html

1 Like

@cyclops everyone can, and is encouraged, to attend!

1 Like

Any hint for the “Cube issue with heading off 90 degrees” ?

I booted my copter with all mag.z being positive. Three minutes into the flight it started to rotate 90° after reaching waypoints.
So… no. No big progress so far :frowning:

Attendee count (max): 26

09:04 - Randy and Copter
Serious bug found in master and Copter-3.5 that can lead to vehicle climbing far too high in autonomous modes if ground station sends do-set-home before EKF has settled on an origin.
Wasn’t a great time for it to trigger
takeoff/climb… and it kept climbing…
Ended up at 500m before contact lost
When not using GPS you can set the EKF origin via DO_SET_HOME
When the altitude was set it was set from a bad altitude
When GPS on the drone locked in its home alt was set 1.5km above origin
RTL made it go home - way, way up
Could have recovered vehicle by switching to Stabilize, Land, Loiter, PosHold, AltHold, etc
DO_SET_HOME is rather dangerous
Need a version which just set alt
Maybe param1==2 means don’t set alt
Legacy issues because old code would misbehave with this
New MAV_CMD required
Any reason to have lat/lon relative?
SET_HOME_POSITION?
lat/lon/alt
Command_int has a frame type
So a DO_SET_HOME as a command int
FRAME_GLOBAL_RELATIVE_ALT
FRAME_GLOBAL_TERRAIN_ALT
Probably also added this bug into Rover…
Absolute altitude is probably misreported
Oddity in Plane vs Copter SITL
Copter Takes a long time before GPS lock
Plane is pretty much instant
Need to investigate this
Changing baudrate probe order made a large difference
3.5 testing is presumably going well…
Randy needs to review any reports yet
3.5.4 will include these origin fixes and some from Peter

09:?? - Peter and Heap corruption
Critical Bug introduced recently in dataflash logging
Patch has been merged to master now

09:34am - Randy and admin
Randy’s a little behind sending reminders out to partners
No contribution of the month for August proposed yet

09:35am - New Airspeed sensor (MS5525) for ArduPlane 3.8
New airspeed sensor
Reading low / calibrating poorly
People are playing with calibration numbers to work around
Tridge is lacking a production version of this
Philip will send some
Might be wraparound / differential pressure
We usually take absolute value
So can ignore tube order
Never get good calibration readings
Philip will help characterise the problem
Graph raw differential pressures
This is the airspeed sensor which reuses the ms5611 ID and stuff
So you can’t tell them apart on the bus…
Sampling too often on ms5611 can skew results
Oversampling rate has to be kept low
Could this also be a problem with self-heating on this device?

10:00am - Relative missions
Use case of flying same 1km grid here there and everywhere
Ground station or flight code?
Mavproxy can already move entire missions
And rotate it
Perhaps using relative waypoints
But taking direction from arming direction?
Many of our mission items include lat/lon/alt
Should probably interpret frame
Only use terrain frame
Lots and lots of changes for relative missions will be required
GCS will all need to be able to display it
Lots of translation required
Maybe there are other options for
Maybe a switch to move mission so mission item 1 is where you currently are
Mission flag which set this behaviour?
Non-gps (indoor) missions?
Still need to tell vehicle where it is
And can move mission based on that
rotate/move ch7 option
Record old orientation in item 0
Entirely new mission type?
Instead of using wpnav

10:20am - mav_cmd_do_set_mode
https://github.com/ArduPilot/ardupilot/pull/6962
Rover people OK with this?
Everybody’s OK with it

10:23am - firmware naming
https://docs.google.com/forms/d/e/1FAIpQLSfLYYuwv6FFHdAVy--iFwgPs7cH_B8hiOfhY5oBdaJoaYICtA/viewanalytics
19 responses
But some votes are missing!
We’ll leave it open for another week
APJ - ArduPilot JSON file

  • allows specification of other formats using different extension
    E.g. apb for ardupilot binary file format
    E.g. apx for ardupilot xml file format
    Doesn’t conflict with stuff as far as we know
    .apb is also not found
    APFW - AP FirmWare
    Hard to say?
    APF - ap firmware
    Conflicts with Adobe application
    AP - ArduPilot
    “Active Page”, so conflicts
    Need to change the vote to delete the duplicates
    Need someone to do the Windows file-association stuff for whatever we do
    Which means we really want something distinctive!

10:45am - trademarks
Should we register?
Jeff is going to try to get an estimate on getting the several names we want
Need to not incur costs on this
$500 to see if it worthwhile
Jeff’s lawyer will need to be aware of everything that’s gone first
Which means tridge will need to inform the lawyer
More to it than the public record
E.g. agreement with Chris on domain names and trademarks and the loan collateral etc
Tridge thinks $500 is nowhere near enough
But funding committee can approve the $500 so it is relatively straight-forward

@cyclops http://ardupilot.org/dev/docs/ardupilot-mumble-server.html#weekly-dev-calls