DEV call 20 Mar 2017 4pm PST = 20 Mar 2017 2300 UTC

Top Contributor


  • 3.5 Update
  • Vision Update



  • Release update

Antenna tracker

  • Release update


  • OLED Breakout
  • Intel Aero Support
  • IMU discussion on hardware
  • Pixhawk 2 is shipping


Top Contributor award

Top Contributor Award for APM Planner 2.0

10:02am: votes
Vote passed, Welcome to the team Eugene!
Contribution of the month
Congratulations ArduSub!

10:04am: Randy and Copter
3.4.6 released
NaN when loiter navigation speed set to 0
Been there two years
Still there in 3.5rc2
Fixed in master
Another similar problem was found and fixed
What can we do to try to stop this happening again?
Add checks between libraries? (don’t trust your parameters)
E.g. constrain_float on exit points of libraries
Increment an error counter if NaN/INF/-INF occurs
INF goes to maximum
-INF goes to minimum
NaN is midpoint
Randy to Add a channel-option to be able to go to stabilize
User didn’t try stab
3.5 release timing
Time to start thinking about -rc3
Several little bug fixes already
Gimbal servo reversing apparently not working
Tridge will look at
Pre-arm failure messages aren’t great
“Roll not configured”
Should probably be “RC Roll not configured”
“Check fence” is pretty random
Need to clarify with “Fence: Outside Fence” vs "Fence: No absolute position
PixHawk mini issues getting it working
FRAME_CLASS message confuses people
Copying BRD_TYPE from one vehicle to another is apparently bad
Documentation needed!
Calibration parameter annotation so GCSs can warn a user before loading frame-specific calibration parameters
Need to come up with more annotations
Visual odometry coming RSN
Hopefully in next few weks
Paul’s found some time
Randy has a camera plugged in which is processed sufficiently to provide position estimates in local frame
Needs to know how to translate to a mavlink message

10:31am: Planes with tridge
Fully successful flight of tricopter tiltrotor
Lots of mag interference
Yaw was a bit bung
Another flight not so successful
There’s a patch for that now
Initial integrator issues
Tridge now has a eFlight Convergence
Tail doesn’t tilt
Tail isn’t used for yaw
Motor tilting to do yaw
Leaning motors is an interesting thing
One of the very few commercial quadplanes out there
Tridge wants to operate it as a tailsitter
Differential tilt for roll control
First vehicle which can be VTOL in multiple dimensions
Is this a bicopter?
With an extra motor
Tridge also likes
Thinks it will be important into the future
Tridge has separated out the PID logs for fixed-wing-vs-quadcopter
Used to use the same message
Particularly bad in transition
PIQ message
PIQR for quadplane roll
Uses these as the primary analysis for flight dynamics
RATE logs give you the sum of all the components
PIDs give you the components
Craig suggests a blog showing how to analyse with this
Tridge has created the tandem heli PR
Revamped Frederick’s PR from long ago
Also transverse helis along for the ride
This one needs testing
It’s been flown by someone with a KMax-style vehicle
Used to fly with Frederick’s patches successfully
Candidate for GSOC
Tridge wants more of these videos!
Should probably use “intermeshing” rather than “transverse”
Would need a parameter for angles and the like
The code may actually be true transverse; tridge will look
Tridge hasn’t looked at the control laws, just restructured stuff for new-look master
Chinook-support is coming as various people are wanting to build them
Also the 2-servo-per-swash Chinook arrangement
Tridge wants to support 4-servo helis
Do_set_servo bug
Jacob picked this one up
Hopefully this is one of the few remaining servo-split-induced bugs!
Tridge suggests a single video containing all vehicle types we support
With both the simulated vehicle and the actual vehicle
PR video showing massive range of supported vehicles
Rob suggests a Thunderbird-inspired video

11:00: MdB and loiter radius
Bank limit at sea level
Scale it to air-density altitude
Set loiter radius w/minimum of scaled radius or your set loiter radius
SITL testing is done
Parameter naming is tough part
Randy points out you should put most important words first in param names
E.g. min/max/scale goes at the end
Code goes in L1
Should there be a different controller?
Tridge thinks the variable should go in

11:09: MdB and EAS2TAS
Randy will test
Few testing issues
May need to change the tests

11:12: MdB found a bug where a takeoff was started when the safety button wasn’t pressed
Patch coming
Happened in auto
Heading locked in too early
Massive GPS glitch
Throw detection disabled (motors goes straight away
Turning around during takeoff

11:15am: copter gitch from facebook
Discuss post coming
mid-air disarm?

11:16am: tridge and soaring
With an aircraft not really for soaring
Grant’s glider’s PIxHawk slim was bung
Firstar with soaring code
It found thermals
But they weren’t convincing
It headed towards thermals
Didn’t manage to gain altitude in the thermal
It will give up on thermals if they’re not big enough
And it did
It didn’t crash!

11:18am: 2G/4G GPS testing
Logs are going to Paul and Michael shortly

11:16am: Grant and Rover
Reverse bug fixed
Disarm-using-sticks fixed
APM2 got loiter code!
Beta release made
AntennaTracker also got more features!
Amazing changes between the codebases now!
But Grant has an AVR-based tracker, soo…
Tweets and discourse posts done
Grant is going walk-about
Back end of next week

11:21am: tridge and Eugene and CANBUS
Going really well
90% of work done by Eugene
UAVCAN is working in-tree without PX4Firmware layering
Works with CANBUS
Build has shrunk amazingly!
Structure is a little odd
Decoding is done in a different place than should be
C++ templates cause code-size blowout
FW update coming!
Flight test tomorrow, hopefully
Fixed-wing with CANBUS servos and CANBUS GPS
Working closely with upstream UAVCAN to do these
Using Pavel’s suggestions for structures
Upstream with zero changes
May be changes later to deal with size issues
We were well behind upstream in UAVCAN
HAL.CAN is generic CAN which will support any CAN
Currently we only do UAVCAN
Support for many may be problematic
Probably can do it interface-to-interface
Not quite ready for merging
Low on CPU usage
Similar to usage on older system
RAM footprint is considerably smaller

11:30am: Jeff and OLED breakout
Should have OLED breakouts ready in 2-3 weeks
JST-GH connector

11:34am: Jeff - breakout board for Intel Aero coming
Tom missed this….
Allows ArduPilot to really be used on the Aero board
Esp RC outputs
ArduPilot runs on Linux CPU
Q: what the hell are we talking about?
Aero can mean many things
PR for Aero support looks OK, and could be brought in
we need a minimum-HW solution
Aero + PixRacer seems extreme
Seems like cheating to Randy
Should just be able to use the board alone
Tridge thinks it would work just as it is
RTF vehicle has STM32 on it
So can be flown just fine
Randy is saying the compute boards needs a minimum-HW solution to get it running
Reprogram the FPGA?
Pins and plugs and sockets?
Jeff’s board would basically just be required sockets
[11:49 AM] (Channel) Craig.Elder: Lucas De Marchi @lucasdemarchi Feb 22 09:12for compute board someone need to fab a breakout board based on this doc: ttps:// or wait for when a different breakout will be available for aero I don’t know when this will happen, but it may take some time

11:52am: Philip and hw
Fire at InvenSense
No IMUs except for Apple for iPhone8
Going to have to change IMUs on PixHawk2

LSM303D is dead
It was crap
Moving away from 9250 as well!
30g sensor on non-isol
ICM20602 otherwise
ICM20948 also
Moving away from these sensors was planned, but this forces things

11:57am: GSoC with Olivier
Student applications have started this-morning
Nothing for us to do ATM
Already two proposals are in
May be mistakes
Sousa and motor failure
Machine learning
Can’t look at these because they’re tagged as “final”!
Should warn people to not mark their proposals as final until they ARE final!
Get feedback first!
We can see things while they’re “in progress”
Look for the blue-on-blue button
We have four mentors ATM
Tridge, Randy, Peter, Shiotal for AirSim simulation
Need to find a student for the airsim thing!
Flyers to sprinkle around
We should:
Continue to have conversations on Gitter
Welcome them!
Watch proposals coming in on Google
See if you’re interested
Google also has to like the proposal!
Tell students separate applications for each
Interviews are later
There are some good proposals coming in!
Students can make multiple proposals
Should probably limit themselves to 3
45 interested students