Japan Innovation Challenge Update
Un-conference update
Copter Update
Realflight8 release and ardupilot
SRV channels patchmerge
Quadplane transition improvements
Chibios update
Issues
https://github.com/ArduPilot/ardupilot/pull/5000
AP_BattMonitor parameter restructuring
https://discuss.ardupilot.org/t/roadmap-for-the-ardupilot-ecosystem/22416
Contribution of the months
Attendee count (max): 19
10:02am - Unconference update
Everybody still travelling
Tridge remotely attended a session (Q&A session)
Audio was a real problem
Video was OK, ‘though
The one question tridge managed to get was “when will we be able to do quadplanes with a single flight controller rather than two”
Which clearly means we have more marketing to do!
Maybe we can use the unconference to reach out to Chinese-language speakers
Make Chinese language videos we can link off website
Tilt-tricopter-quadplane seen flying and there were some stability fixes we haven’t seen yet
Would be nice to get those changes and getting these guys into the community would help there
Wiki translation into Chinese
Large ongoing maintenance burden
Maybe the contact
ArduPilot website’s second-largest audience is Chinese
Getting some key pages of the main website translated would be beneficial
Over-dubbing tridge’ demo/tutorial videos?
Maybe someone from Hex could suggest someone to do the translation
Ask Jiang or Gene could suggest someone?
Perhaps we could fund someone?
Here+ GPS pages were done - perhaps that person could help?
Google translate?
Translate might not be available
JR module for Taranis was shown
Hex is editing videos and will distribute
Will revisit this agenda item next week
10:12am - February Unconference in Canberra
Need to avoid parliamentary season for hotel reasons
Calendar in early November
Will try to organise date fairly soon
We had two days officially
Talks and official flying
A couple of people turned up 2 days earlier and some were there 2 days later
Was a Saturday/Sunday
Flying indoors… hovering planes
Want to get as many core team members as possible
Need to check Randy’s availability
Tom should be able to make it
MdB should be able to make it
Craig would love to come down
Olivier is considering, would love to
Jacob’s keen
Maybe early March
Shared hotel rooms makes things cheaper
Jacob: Need a hotel with a pool
Tridge: Hotel was very tolerant of our flying
Indoor pool at academy
10:20am - tridge and RealFlight
KnifeEdge are releasing RealFlight8
First update will include support for ArduPilot by default!
Free for RealFlight8!
Tridge has improved the framerate a lot!
Can fly with ekf2 and ekf3 in control of vehicle
Low innovations
Ekf does not go crazy
Works surprisingly well
When tridge records a video the plane stays in control because the performance issues with RealFlight have been fixed
Stable flight in helicopters
They’re also releasing a toolkit which allows you to edit the graphics without having 3dmax from 2012 which is no longer available
Can create quadplane models with actual props in the right spots!
Need people with graphics skills to create the models
Tridge is trying to do it with Blender
Official technique is still to use 3dsmax
33 tutorials for Blender
Format required is .3DS
KEX format is used by RealFlight
3ds2kex tool can run under Wine
Not all formats can save the pivots like the hinges
“FlightAxis Link” is their term for it
Press a button and you will get a code to activate the link
Settings menu / get code sort of thing
Other autopilots could use it, AP is only one who supports the protocol ATM
Check the EULA
10:27am - XPlane with Austin
Doing it as a plugin into XPlane
Double the framerate now
Close to getting lock-step scheduling
Will need to install ArduPilot plugin for XPlane
You’ll get a much better experience in XPlane after this
Will use its own protocol
Source code for plugin will be Open-Source
Up on tridge’s github account now
Tridge will be supplying binary plugins for Linux
Will need a volunteer for building for MacOS
Luis will have a look
Tridge is cross-compiling for Windows on his Linux box
Current support has issues for smaller vehicles due to framerate
10:35am - SRV_CHANNELS
Tridge merged a patch from months ago
Changes the way we do RC outputs
Everything goes out via SRV_CHANNELS now
Simplifies a whole bunch of stuff
One-to-many mappings
So you can have multiple output channels for motors
AP_Motors now uses srv_channels
If you see oddities please let tridge know ASAP!
Testing of master would be great!
Already flown on SkyViper OK
10:38am - QuadPlane transition improvements
We should now more-reliably transition
Long-standing problems
10:39am: uc4h
https://www.rcgroups.com/forums/showpost.php?p=38502238&postcount=151
https://www.rcgroups.com/forums/showthread.php?2971848-The-Holy-Grail-of-DIY-QUADS-FrankenSolo
10:39am - Chibios
Peter and tridge have been doing some work on getting AP onto Revo
Bootloader runs
Firmware can be uploaded using px_uploader.py over USB
ArduCopter was starting and starting to bring up the uaets
Not flyable by a long shot
10:41am - https://github.com/ArduPilot/ardupilot/pull/7094
Septentrino corrupt on power-down
Attempt to avoid corruption
Pre-arm check if the GPS is not logging
Need to bounce first
Peter: We should send back temporarily-rejected not failed
MdB: Peter: yeah, its a gcs/notification thing
Deferred arming or something is needed
An arm-pending state
Main concern is the notification delay to the GCS
Arming-on-progress when using rudder stick
[10:47 AM] (Channel) rmackay9: at the conference, one idea that was brought up was that we should continue to check after arming and before takeoff
[10:47 AM] To Weekly devcall: As in continue the arming checks after we’re armed?
Tom suggests a retry
We’d need a “defer” response anyway
Maybe a defer in milliseconds?
Maybe pending-arming is set by the arming mechanism and then it is asynchronously resolved
AP_Arming would probably be responsible
Gets an update() function
Disarm should cancel the pending arming
What do we put in the statustext message
[10:54 AM] To Weekly devcall: Doesn’t look like there’s a nice MAV_STATE we could use?
Should logging always probe the SD card to see if it will succeed
Can we merge stuff as-is to avoid scope-creep?
Tridge says yes, but please add singleton or non-consting
create() stuff means code changes are a little more wide-spread now
Tridge would prefer the sensor-singleton path
11:03am - https://github.com/ArduPilot/ardupilot/pull/5000
Array of instances for battery stuff
Like srv_channels and rc_channels
11:06am - 16 input channels?
Yep, already done
Multiple features on same channel is an issue for the RCn_OPT stuff
11:18am - parameters
Waiting for GCS changes
The combined parameter file is painful
MAVProxy does the right thing
Except it doesn’t include sub
Jacob’s fixing that
Also need to do AntennaTracker….
Need to revisit this one next week
11:27am - roadmap?
Need the right people on the call
To officially say we don’t have a roadmap, really
Any work being done on one or commitment to do one?
One from the unconference in CBR last year
Difficult to read
Represents haziness of the plans
A lot of these have been done!
News to come on the Monster Copter!
kHz rate loops in Copter is still in progress
Need to do another of these in Canberra in Feb/Mar
11:36am - Randy and Japan Innovation Challenge
Congrats to Randy!
Team effort
Extended team inc tridge and Peter
Recap
5 day event
Mon to fri
Long event
Real challenge just to keep in good shape
Cold and a long way from home
13 teams using DJI
One using ArduPilot
Last year there was a better mix of autopilots
10am-11am search for 1 hour
First-to-find in 3km wide area wins the search stage for that day
After that the position is announcement and anybody can deliverthe delivery stage
No attempts to retrieve dummy as third stage
2nd 3rd and 5th someone else found
4th day Randy found
DJI teams were just doing FPV
Some nice zoom cameras mind you!
the 10000USD DJI Z30 zoom camera
Pot-luck to a certain extent
DJI guys have luck and numbers!
e-Lab donated lots of vehicles!
APSync and APWeb with cuav software running
Only up to 4 out of 5 flying
First couple of days was 2-at-a-time
Camera was a real problem
8MP with special lense (ArduCam)
Total piece of garbage
Really annoying
OK over summer
Temperature drop == stopped working!
Intermittently
Different for each camera
Lucky drone 5 always worked
Some failed at 13 degrees
Randy had 10 cameras
Mixture of plastic and fancy
Heat-gun to warm up cameras before flight!
Would have done so much easier with cameras that worked!
Different camera for next year!
Might be able to fix this one but…
Bayer images have better colour depth but colour-compensation was a problem
Purple around images - no green!
Cuavcam may not be able to pick dummy up unless in centre of image
Raspistill does colour compensation based on position in image
These cameras have colour profiles for their lenses….
Can compensate for this but will need to know what the algorithm is
No gimbal that matches the camera
Spent way too much time working on a smaller gimbal for the camera
Used Tarot gimbals with weights instead
No weight or flight-time issues
Massive batteries, thanks e-Labs!
40 minutes flight time!
6-10% of the battery!
Fixed focus based on feedback in Skype - thanks!
Exposed rolling shutter issues though!
Too much jitter in gimbal?
11m/s missions
15m/s to get over to the search location
Terrain-following worked great!
Meant average of 9m/s
Crashed on last day
Operator error!
Bad mission uploaded
Low takeoff altitude and a bank of trees 300m away
Didn’t fly over the dummy on the day
DJI guys are searching in real-time
Randy had to bring vehicles back to do the search
Two-stage process
Cuav in real-time
Requires regulations to lighten up in Japan for two-way telemetry
CUAV gives way too many false-positives
Dummy changed colour on different days
Randy’s group was by far the most interesting group there
Nobody else was doing MV
Many people are going to be looking at CUAV!
CUAV can read images from a DJI copter
Exif extensions include roll and pitch and yaw
And the gimbal position
Height above ground etc
Need to pack more data into the exif
Slower because adding the exif data
Need to add data as we go
See Jaime’s MAVProxy module
Should talk to the maker of the camera (UCTronics)
Randy’s been doing some of that
No response yet
SoloLink is not certified in Japan!
Solo is
Japanese RF emissions are 10x stronger/enforced compared to others
Big PR win!
Next year’s going to be kick-ass time!
