DEV call 13 Mar 2017 3pm PST = 13 Mar 2017 2300 UTC

Copter

Plane

UAVCAN

  • Progress

GSOC

Issues


Top Contribution

Attendee count (max): 23

10:02am - Copter and Randy
3.5.0rc2 is out
Is master from yesterday
Release candidates should be a week or two apart
Maybe 2 or 3 more
Certainly at least one
ROS stuff
Paul is integrating visual odometry
First source is z-camera
Small steps getting mavros running
Seems mavros isn’t publishing out pose and position estimates anywhere
Going to look for other people’s modifications to get those
Tutorial will be coming Randy TODO

10:07am - tridge and GyroCopter
Buzz’s fault
Flew it on Sunday
https://goo.gl/photos/A6cfo93d5n9dcf678
Takeoff in Manual
Flicked to FBWA and it flew fine
Second flight worked in Auto no worries
Slight bingle during second flight
New blades needed
Will need a new servo function for Gyro rotor which doesn’t do speed-scaling

10:10am - tridge and Andre and soaring
It’s in master!
Thanks to Andre and Samuel
A few more changes required
Particularly to the CRUISE mode
Testing will be happening in South Africa
Control via RC to allows eyes-on-aircraft
Adds to existing modes
E.g. in Auto mission it will wander off the track, gain height in thermal and then continue
Temporarily changes to loiter mode
Centre of loiter in centre of thermal
4-state EKF
Thermal strength
Strength in centre
latitude/longitude of thermal
radius of thermal
Assumes centre of thermal is strongest
Attempts to fit bell-shaped function
Maximum-height means we leave the thermal
Also does this in fbwb and cruise
But goes to RTL afterwards
Cruise should probably continue to cruise “waypoint” rather than RTL
Should probably be a proper sub-mode rather than doing this switching
Whole bunch of parameters to indicate
Alt it kicks in
Min alt
Max alt
Would be nice to get rid of some of the parameters
Loiter radius, for example
Request for docs!
Andre confirms that he’s going to help out here!
Would be nice to use the estimated thermal radius as the loiter radius
http://discuss.ardupilot.org/t/soaring-gliding-for-planes/15323/9
Should probably get vertical velocity (get_velocity_NED()) from AHRS rather than getting altitude from AHRS
Will give a better, smoother result than current code
Bell-shaped model doesn’t take into account down-drafts on edge of thermal
Need a Mexican-hat model, for example
Big change!
Better strategies for finding centre of thermal
Bank the aircraft when you think you’ve entered a thermal
Maybe another “candidate thermal” state
Tridge has received his SkyHook glider
Motors and servos required
Couple of weeks still, then
Model available in XPlane from forums
Grant’s glider is ready
Probably fly it this weekend at CMAC
Later out at Lake George
Andre might be able to help out with the AirSim stuff, too!

10:39am - Eugene and CANBUS
GNSS is working (but without baro and mag)
Servo control works
Will be tested in-flight soon
ESC control
Can’t get ESCs to test on
JC can test it with his
Tridge’s are too old to sensibly test
Reverse thrust
Not having it initially is OK
But we do need it
Hex’s can derive some stability from reverse thrust on motor outage
PWM scaling support is an issue
Only one scaling for all ESCs could be bad
Need to have per-ESC
Plug for gitter channel for those that want to follow along at home
https://gitter.im/ArduPilot/CANBUS
Code size issue needs to be fixed
CAN-specific ORB stuff is still in there, which might be the issue

10:47am - GSOC with Olivier
Lots of interest
50 students signed up on the GSOC channel
About 35 expressed interest
Next Monday is sign-up time
Lots of high-quality students already
Mentors are thin-on-the-ground ATM
This might be the limiting factor on the AP side!
Next step is to provide more guidance to the students
Some sort of template for application
Encourage them to create at least one PR
Students should show they’ve thought about a problem, not just express interest in it
E.g. propose a solution to a problem
Strong interest in Avoidance
Strong interest in coax/single copters
Strong interest in autorotation
Should we point out on gitter channel that PRs are kindly looked at?
http://ardupilot.org/dev/docs/where-to-get-the-code.html
Quick re-visiting of the tagging-simple-items thing
Need to make sure the students can help each other
Official coding starts May 30th
1 month of acclimatisation before that
Coding stops August 29th
Musing on the dev wiki documentation and thinking it could be improved
Eugene is willing to help with this

11:06am: only one issue is listed as dev-call
always-use-poshold
https://github.com/ArduPilot/ardupilot/issues/5832
Tridge thinks it should stay the way it is
Vehicle may not be tuned sufficiently to hold its own position
It might be that you wouldn’t be able to complete a tune
Question as to how good results might be with/without
Probably marginally better without the hold
Resolved to keep it optional in 3.5

11:11am - top contributions
We’re 2 weeks behind in making this decision
Do we just say “ArduSub”?
Best to always do a vote
MdB accidentally got himself volunteered to create

11:15am: random freezing of IO outputs?

11:17am: Jani and OSD
New OSD on the way
HW will be created from scratch
Not sure if the HW will be open ATM
Hard to make Open HW and still be able to eat
We all love open hardware
Possibly more important is the tight relationship the HW manufacturers have with the AP community
Code cleanups
Might be possible to get code into main repository
HD OSD?!
Jani says old MinimOSD is difficult to produce because MAX7456 is EOL
Luis mentions they might be being cloned in China

11:22am: Randy mentions new flight controller
PR on the way!
RobSense
FPGA on board
Robsense.com
Boards should be sent out to core devs
Onboard video
CAN….
http://www.robsense.com/

11:25am: Where is Philip?
Taiwan!
Production is moving along!

11:26am: MdB and Servo limits on outputs
We currently don’t limit PWM outputs to min/max
That’s a feature for MdB, so he doesn’t want to lose this feature!
Tridge says we need an unclamp interface
Could have a 16-bit unclamped_bitmask
Call to set all k_elevator channels to “unclamped”
Mask cleared at start of each loop
Deepstall code would have to make that call every loop
May want to still have an unclamp-percentage
Or a wider range
E.g. 800-2500
The really-really out of range values have crashed boards before.
Will be an API

11:34am: more votes for dev-team members to go out!