DEV call 06 Feb 2017 3pm PST = 06 Feb 2017 2300 UTC


  • 3.5 release update
  • object avoidance / vision project update - 1st draft of ROS install scripts for TX1 produced, point cloud produced using ZED camera.





10:03am - Randy and Copter
A few issues are coming up
OSD display never clears
Sats display can be incorrect
Thrust-to-pwm scaling can be wrong (maybe)
Only seeing two compasses on PH2
Boardtype on PH2 may be defaulting incorrectly
Compass calibration issues
Could be declination

Have a “check-list” for this release on discuss
OA updates
Merged TerraRanger proximity sensor
Some wiki work
Promoted some copter up a level
Dan P helped, and now with Z-Cam we can make point-clouds
Scripts in ArduPilot Companion PR
Need to take point clouds and turn them into info we can use for OA
Andy T has done something similar to OpenKAI
I.e. using depth map from z-cam to pass in objects
AT is using ROS
Sends data using mavros
Into the future we’ll probably build occupancy grids and the like
New copter frametype Y6 for FireFly

10:25 - Tridge
Parameters generation has been fixed
Ordering of nested parameters is still broken

10:30 - should the mission alt offset go into the mission library?
Needs MdB’s input

10:30 - cross-track bug
Needs to be looked at by Randy (cross-vehicle implications)
TP thought he’d fixed it in 2015
Seems to be back!
JM has provided logs

10:35 - new plane release
This has been delayed due to issues with a quadplane crash
TP wants to analyze more plane logs
‘Though this might be an “alternative fact”

10:37 - ROVs with Jacob and Rusty
Having a few issues with git branches and the like
Trying to work out whether to force-push
They’ve made the PR from their master
General agreement its a great PR and we’re looking forward to getting it in
They have a subclass of the GPS NMEA driver to work with
Connection to surface computer
Puts out NMEA sentences
Gives position of end of tether
Need to get data from surface computer into MAV
Currently they encapsulate using the gps inject stuff
Probably need to stop doing this and have a mavproxy module to translate these messages into gps-over-mavlink
Tridge points out development methodology
If we see PRs just for the sub directory then we can be confident they won’t break other vehicles
Using index 0 in subgroups
Historical design flaw
0 << 12 is still zero, so using this offset doesn’t work
Should probably check for duplicates as a sanity check (in SITL)
Tridge will create an issue for that
Fences and altitude
Copter should have a minimum-fence
… which Sub can then leverage
Barometer issues
Don’t reset calibration at boot
They tend to reboot at depth a bit…
Probably want to have a parameter to indicate whether it should calibrate-at-boot
No parameters for baro2 baro3 - nowhere to save offsets
Loading old parameters is a problem here as the GCS have a magic list of parameters they won’t load
Maybe need a “read only” attribute on parameters (from GCS perspective)
New sensor for temperature sensors
frontend/backend split
Have lots of temperatures as ancillary values (baros, IMUs)
For water temperature
Really nice to have for ESCs
Welcome to the Dev Team Jacob! Unanimous vote with a record 21 votes
Hopefully get PR in by next week
How does this impact releases?
Randy hasn’t created a branch yet
Randy will work around problems created by merge
Minimal impact on other vehicles
They’re creating new classes rather than modifying existing ones
Grant will release a rover after Plane goes out with RC restructuring

11:15am - Philip and hardware
Dual battery improvements in software should be on the way!
Will move failsafe values into battery library
Also might want multiple stages (critical vs low)
Distance-to-home stuff like Solo is getting would be great
Solex warnings need to be replicated for low battery conditions
High-voltage power module has changed name
Based on who is manufacturing it
Do we want a generic failsafe library asks Randy
M8Ps are on the way
New firmware; use uCenter
Over USB
Glonass stuff is much better now
Having glonass+gps should improve your position rather than not.
RTK fix under trees?!
PH2 issue
Over-torquing screws can short pwm output 1 and ground
Beta units only (first 1000 that went out)
Hard sticky insert will be available for affected area (sent out to beta-unit-buyers)
Next batch will not have issue
Could use channel remapping to avoid issue
Only C1 is affected
Pin is tolerant to hard ground
Not necessarily a permanent issue; just an exposed pad and incorrect marking on board
Blue cube
Cube slim
Reduced IMU count
“Ice cube”
Lots of cool announcements coming
“Pink cube”
Colours are apparently hard

11:44am - amilarcus and radio rc messages
Tridge merged it

11:44am - Google Summer of Code
Some sort of web-based admin interface
Vetting should be done - don’t take just any student
Otherwise you spend all your time chasing them
You can propose projects
So can the students
Ardupilot could propose a list of tasks
We have lots of ideas for proposals
And the issues list!
Use GSoC tag in github
Lots of people are willing to mentor
Olivier volunteers to help on admin side
GSOC gitter channel
Has been created
Helps get community involved
Need to be strict on using community resources
Avoids completely draining the assigned mentor

11:56am - APSync update
RPi video streaming UDP
TX1 depth-mapping using zcam
Apparently USB3 can be noisy
SoloLink has 50ms lag
GoPro has 80ms
So Solo has 130ms
Anywhere under 200ms is acceptable range

12:10pm - TP and adding to changelog as we go
Add to changelog as we go
Tridge says there’ll still be work at end as it usually needs a lot of cleanup
Will help to crowd-source the work a bit more
Tridge points out that there’s some major documentation requirements for the next plane release that will need to be done

12:16pm - PR for style guide w/jaxxzer
We’re going to start accepting either style for case statements now!

12:19am - vote for best contribution for January
Here+ is the winner!
Thanks to Emlid for the additional prize of a NavIO2 board for this month’s winner!
Craig will remind people at the beginning of dev calls