DEV call 23 Jan 2017 3pm PST = 23 Jan 2017 2300 UTC


  • 3.5 Progress update




  • Need to do a rover point release to resolve a critical px4io bug - spektrum receiver related


  • M8P
  • Compasses
1 Like

10:02am: Randy and 3.5

Several closed, but Randy’s still working
Will let 3.5 slip a bit to get Paul’s EKF changes in
Also the OLED display thing
M8P stuff too
Needs flight testing
600 M8Ps will be out next week

10:08am: Randy and Tridge BeBop and Disco live video
Live video has been done
Tridge ran a command on the disco to start the stream
AP_Module allows digital gimballing
Also records to the onboard storage (for the high-res video)
BeBop2 is probably the same
How do we do video control?
Ch7 option?
Invoke a script to start/stop recording
And possibly start/stop video streaming?
Start streaming on Disco: pimpctl stream-start front 8888

10:31am: Pull request from Michael
Battery status sensor
Basically to tell GCS that battery is flat as a more reliable mechanism vs statustext

10:33am: Plane with tridge
Tiltrotor quadplanes
Some successful flights (Rolf), scratch-built aircraft
Software worked just fine
Tree caught aircraft, unrelated to autopilot software….
Tilt-tricopter variants
Continuous control of angle of tilt (continuous control of servos)
E.g. 9 seconds to go from hover to full-forward flight
Also uses a minimum airspeed to stop at an angle until you’re “flying”
Retract servo
Either forward-flight or vertical flight - nowhere between
Fixed rate, no control over that rate
Parameterised in ArduPilot for that rate
Tridge will be flying his CL84 tomorrow
3.8 alpha release coming

10:40am: RC channels split
Several bugs fixed
Esp. for defaults files

10:45am: defaults files on bebop etc
Maybe install defaults files as part of waf install
Maybe build the default file as a string into the binary
ROMFS can current hold the defaults file
Solo.parm was added in there
This would add things as a string instead
No ROMFS on Linux as it has a real FS (as opposed to STM32)

10:56am: Grant and point release for px4io bug
Spektrum receiver bug
Point release for Rover (no other vehicle type did a release while the bug was present)

10:59am: Philip and hardware
Again, Michael’s patches need flying!
Compass inconsistency issues
PixRacer Mag3?
Randy noticed long vector lengths on some of the compass vs others
Increased calibration tolerances again (50%)
Same for PixHawk2?
PixHawk2 Mag2?
It isn’t everyone
Will change the defaults
Question as to whether compass consistency check should be an arming check
Problem is we could fall back to secondary compasses now
We should make sure compass orientation makes sense with accel values
Move consistency check to calibration check
Ordering of compasses can’t currently be specified
Proposal coming to fix this
Maybe there’s a scaling error somewhere causing the consistency errors
If you have a tlog you can tell
Discussion of RTK GPSs and surveys
Phase centers for antennas
Needed to allow GPSs to know their position
GPS can survey themselves in
Obs file contains comments, but they’re not taken into account for the RTK solution
Some contention on this
Need to be able to log stuff reliably to dataflash on the vehicle
Michael is using parameters to ensure things have been written (param sets are acknowledged)
Need to make sure we’re using the correct elipsoid throughout code
Problem recently detected with Septentrinos
HAE != MSL….

11:41am: Olivier and website
Jani offers technical assistance on any website stuff
Olivier has been concentrating on the higher-end vehicles
Thinks we should try to bring it down to something accessible
RlF suggests having a couple of collages, one for high-end one for low-end
Still getting permissions for some images
Wants to flesh out the use cases
Spending a lot of time ensuring its ArduPilot!
AeroBug in Australia being used to disperse bugs
Not using auto just yet….
Real boats… catamarans…
Using them for mapping
ArduPilot in education settings
Acid lake / ArduBoat
Rovers and Robotics
Fun FPV flying
Research projects
Hex Logo
3D printing with a drone?!
Some whimsical stuff
Need to be careful here with legalities
Similar to how we wouldn’t show weaponised drones

11:55am: GPS failover smoothing
Paul will be looking at it
[returned to this discussion 11:55am]
Work has started
A couple of different approaches
Blending sources, looking at error rates, and assuming what the distributions should look like
Will produce a single GPS source to the consumers!
How to handle delays in arrival of data
Need to synchronize time horizons somehow
Need GPSs that provide meaningful error statistics
That doesn’t include number of satellites currently seen
More velocity errors and the like
Haven’t really looked at heterogenous GPS environments
Probably need to stick to single manufacturer
Preferably same model
Will still fall back to switching if required
Will be a parameter to determine logic to use (switching vs blending)
Complicated when switching between different fixes on same GPS e.g. from 4D fix to 3D fix

12:08am: new compiler for stm32?
5.4 makes bigger binaries
Less of an issue now
Luis will be creating an issue for someone to test