Issues & Pull Requests
Plane
Copter & Rover
GSOC
Attendee count (max): 20
- Avoidance recovery issue
- Peter will make this not use the loiter mode and bring it back next week
- GPS prevent switching to dead GPS
- Duplicate code does suck but we need a fix
- The usebest code should be fixed
- Merged
- Remove persistent guided wp loc state
- Merged!
- Try both native and AP::FS filesystems for posix systems
- Merged!
- Add set_override method for scripting
- Should we still be down on RC overrides?
- Manual_control instead?
- Not practical because it doesn’t allow buttons and whatnot
- If your aim is to simulat pilot input then rc overrides are your things… otherwise use a controller
- 10:23 AM] Peter Hall: For dodgey scripting stuff I sometimes just don’t add an example.
- [10:23 AM] Peter Hall: For example there is not an example of param set and save.
- Manual_control instead?
- Fix ignore RC failsafe on IOMCU
- Some protocols have an “receiver not getting data” bit in the protocol
- So fixed values or whatever
- We do honour the bit correctly
- But we have an RC option bit to say, “don’t honour it”
- E.g. carefully set up particular values in transmitter/receiver such that specific pilot inputs are given
- This only works if directly connected to MCU at the moment
- Not when on IOMCU
- Which is a bug
- This only works if directly connected to MCU at the moment
- E.g. carefully set up particular values in transmitter/receiver such that specific pilot inputs are given
- But we have an RC option bit to say, “don’t honour it”
- This isn’t a rushed thing as most people won’t be setting the bit
- Merged
- Avoid use of location alt fields in the ekfs
- Merged
- Conflicting RC input
- Merging inputs bad….
- We only allow for one serial protocol to be set to 23
- If you want two inputs you need to use IOMCU
- Code still failsafe in testing…
- Should have just failed over
- Log inbound
- Two herelinks to a single cubeorange and switching between the two
- Failover!
- …. When this gets fixed….
- Failover!
- [redux]
- Having a button to switch would be good
- Major restructure would be required
- anyway, glad we got the ideal UI agreed… “your aircraft” / “my aircraft” switch for both pilots…
- Difficult to test properly
- Only knowing when it actually happens is very bad
- Multi-point handover code fell over on telemetry radios, for example
- Leonard wants multiple RC receivers for a bunch of stuff he’s doing
- Current system is that the handover is mediated by primary link going down
- Not designed for modern telem/video/rc systems
- This was really designed for two spectrums or whatnot
- Add an RC option for “prioritise primary” vs “prioritise secondary” would be simple
- RC handover should use my aircraft / your aircraft switch · Issue #20344 · ArduPilot/ardupilot · GitHub
- Incremental?
- We should emit a message indicating who is in charge
- IOMCU is special
- Hal-chibios handles iomcu
- It shouldn’t
- It should be a backend
- Fix compilation with ins accelcal enabled
- Merged!
- Stops compiling Solo gimbal for vehicles other than Copter
- Re-nable for CubeOrange
UTC0004 - Slight tidy of GPS blending by peterbarker · Pull Request #20331 · ArduPilot/ardupilot · GitHub
- sq() costs us bytes when it really shouldn’t
- Option to send resting voltage estimate through to GCS
- Ground-lift can cause even more interesting things
- [11:11 AM] Peter Hall: The problem is it can hide bad sag. Yeah ABS voltage is your power avalable.
- [11:11 AM] Peter Hall: But its not defualt so whatever.
- The ground lift can affect motor outputs too, but it’s constant for a setup usually, so calibrated out
- Configure and log TIM-TM2
- Write rather than using a structure?
- So camera trigger pokes GPS which sends a message which gets logged
- Needs to be optional
- Maybe we ought to log the raw stream…
UTC0024 - Plane: stable 4.2.0 issues list · Issue #15941 · ArduPilot/ardupilot · GitHub
- Plane update
- Testing went well
- Except physical failure on one aircraft
- [11:25 AM] Peter Hall: We do have some issues for dual motor plane failure handling.
- [11:27 AM] Peter Hall: We should be able to handed it correctly now we have a yaw rate controller.
- 11:28 AM] Peter Barker: @Peter Hall well… qassist was kicking in and out, making it essentially uncontrollable and eventually put it into a spin
- Back-transitions were awesome
- Particularly on the Eagle Hero, chalk and cheese
- Getting overshoots on landing on the H-Wing
- Even on the simulator / RealFlight
- 3 and 7m of overshoot
- Back-tracked and landed beautifully
- Beautifully smooth landings just overshot every time
- 4.2 testing has gone very well
- The sum battery isn’t updating the percent-remaining measurement
- Home-alt handling on home reset
- Re-triggering of geofence?
- Rtl autoland retriggers on going back to auto
- Not retrigger if you’re in rtl autoland?
- Rtl autoland retriggers on going back to auto
UTC0031 - Copter/Rover: 4.2.0 issues list · Issue #20192 · ArduPilot/ardupilot · GitHub
- Copter and Rover updates
- Couple of issues
- Landing not slowing down
- Land alt low not working?
- home/ekf origin bug with recent changes to landing/takeoff?
- Can’t use channel 6 tuning knob to change speed in auto mode doesn’t work any more
- Landing not slowing down
- Having a slider value to speed up / slow down time for SCurve navigation might be an option
- D-shot reverse not working?
- Wiki pages will be fixed up to make it easier to work with d-shot
- Blheli pass-through not working on CubeOrange for Randy
- Works for PH and some ESCs
- Making this reliable?!
- Blheli passthrough is a really, really bad protocol
- Timing sensitive
- Protections on auxiliary pins screw you over
- Protocol was designed for direct MCU pins
- No electronics in-between
- Dangerous!
- Back-emfs / voltage spikes and your flight controller is toast
- Dangerous!
- No electronics in-between
- Tridge can walk people through the debugging steps
- Including logic analyzers
- Particular to ESCs
- Such a useful feature!
- Updating firmware and settings etc etc
- …. Just a pain on some flight controllers
- CAN ESCs FTW here
- The Wiki pages say it will work and it doesn’t work
- Should the pages say, “give up now?”
- Whitelist ESCs?
- But firmware changes can break things
- Clear warning on Wiki saying some boards have circuit protection measurements that interfere with blheli communication so you may have trouble if so you will need to disconnect the ESC and use other means to work with the ESC
- Hw-wise if rise-time is known then protection can be designed around it
- But time is probably better spent on CAN ESCs
- F3-with-can coming soon from APD with native CAN
- Affordable CAN ESCs are problematic
UTC0043 - GSoC update
- Got until April 3rd to work out who students are
- Applications start from April 4th
UTC0043 - close