Flight Code Release Update
Issues & Pull Requests
Attendee count (max): 19
UTC1100 - vehicle updates
- New beta is out
- We found another terrain-guided-goto bug
- Handling of .terrain and .relative
- Use in plane contradicts the way location object works
- PR to handle it
- Adds some paranoid checks
- Adds some checks in SITL
- Need to discuss making it easier to add tests which look more closely at the profile
- Want this fix in the next beta
- Some progress on the tilt-rate issues
- Paul has a an alternate change in TECS
- PH also has a possibility
- Slew-rate in motors
- Three PRS for the similar issue
- Probably end up with a combination of the changes
- Microbrain proximity sensor
- Randy will talk with Rishabh
- Multi-heap allocation in Lua
- Thomas was looking
- Asked tridge to also look
- Cygwin-only issue?
- Thomas was looking
- Tim will chase up Jason on the M10 GPSs
- Altitude run-away when rangefinder goes out of range
- Randy’s seen a flight log
- There’s an arming check
- Range_valid_measurement_time_ms
- Fake rangefinder in use when rangefinder not available
- EKF uses a baro
- Keeps EKF on leash
- Common to get rangefinders that can’t get readings on the ground
- This is rangefinder as active EKF source for height
- Force users to use a script to switch sources?
- ie. take-off using baro, require rangefinder to work after take-off and switch to it
- Original issue?
- Rishabh will provide the log now has has permission
- Randy also has an applicable log
- Maybe EKF is ignoring rangefinder because it’s too far from current estimate
- 7-second run-away
- Maybe it wasn’t fusing any vertical source?
- Blheli mother-of-all-fixes PR is still under discussion
- Need to consider which of these are regressions and do need to be fixed
- Randy’s availability to look at another beta is limited
- Maybe next Thursday
- Tridge wants the guided-terrain stuff in
- Randy’s thinking 4.6 is going to a really good release
- Lots of stuff has been gone back and fixed
- An AP_Periph release?
- Tridge wants to do one
- Normally don’t use the same branch
- Tom thinks we should link AP_Periph releases too
- So does Randy
- Partner is shipping a non-release firmware
- We found another terrain-guided-goto bug
UTC1138 - AP_RangeFinder: populate last_reading_ms in the backend for some sensors by rishabsingh3003 · Pull Request #27620 · ArduPilot/ardupilot · GitHub
- Populate last_reading_ms in the backend for some sensors
- Merged!
UTC1139 - Plane follow script Applet by timtuxworth · Pull Request #28546 · ArduPilot/ardupilot · GitHub
- Follow script applet
- Tested after making suggested changes
- Tridge has left a review
- Peter will work with Tim to test and get in the position-from-origin internally
- Some discussion around the interface method being added
- Some bugs in the API which tridge wants fixed
- Distancve_to_target and bearing_to_target
- They return the variable but don’t check when it was updated
- Do we want an update function?
- Probably
- Kinematically consistent follow
- Reset / timeouts etc?
UTC0019 - AP_HAL_ChibiOS: add hwdef files for NarinFC-H7 by vololand · Pull Request #29241 · ArduPilot/ardupilot · GitHub
- Add hwdef files for NarinFC
- Merged!
UTC0019 - Add support for System ID on Aux Switch in Plane by bnsgeyer · Pull Request #29608 · ArduPilot/ardupilot · GitHub
- Add support for system ID on aux switch
- We’ve got log data now
- Better to get it while in a large loiter
- Some discussion about the APIs
- In the wrapper functions or not?
- Eg. flip to acro to stop doing things
- The yaw controller?
- Very few people run it
- either damper or full yaw controller
- Exciting yaw at all doesn’t happen much at all in Plane
- Full pid controller isn’t used in normal operations
- Just yaw damper
- Or just zeroing lateral acceleration for co-ordinated flight
- Most users just feed-forward aileron to rudder
- So roll controller doing job vs kff_rudder_mix
- Tuning rudder and getting it right is tricky
- Systemid to tune pid control for quaternion acro is a use case
- Yaw Damper tuning is also something tridge would like
- Instructions are similar to PID tuning but it isn’t a PID
- Pilot input is a rudder offset in almost every mode
- Multiplier of aileron
- Or an acceleration controller that is trying to zero lateral acceleration
- Y-accel is being controller for, trying to be zeroed
- Not asking for an angle or a rate
- Feed-forward works well on smaller planes
- Large planes seem to work better using P and D
- Canard and high-inertia
- Need some SITL aircraft that don’t do best with FF
- Large planes seem to work better using P and D
- This tool will allow you to do a “what if?” after you have gathered data
UTC0040 - Rename get_alt_distance -> get_height_above by peterbarker · Pull Request #29648 · ArduPilot/ardupilot · GitHub
- Rename get_alt_distance to get_alt_difference
- Peter will merge once he decides on the name
UTC0044 - AP_Common: Location: avoid converting to absolute alt in get_alt_distance by peterbarker · Pull Request #29649 · ArduPilot/ardupilot · GitHub
- Avoid converting to absolute alt when not necessary
- Avoids failure when failure not required
- This is a significant behavioural change
- Merged!
UTC0049 - AP_Landing: stop dup retract messages, clean up code by Hwurzburg · Pull Request #29676 · ArduPilot/ardupilot · GitHub
- Stop duplicate retract messages and clean up code
- Spewed the deploy message
- Logger-write event will only go out if function is assigned
- LUA script could look at the scaled output to do stuff?
- … actually does a set-output-limit
- Don’t do altitude-based triggering if a user is manually triggering with a switch
- Questions around whether this is a behaviour change
- For people doing auto-retract but not auto-deploy
- There’s an existing bug where the retract will only happen the first time
UTC0108 - Plane: fixed terrain guided target intermediate alt handling by tridge · Pull Request #29685 · ArduPilot/ardupilot · GitHub
- Plane location sanity checks
- Sitl-panics vs internal errors
- Rokblok LUA script is buggy
- Exposing the innards of location was a bad choice in scripting
UTC0120 - AP_Math: split `radians` into float and double variants by tpwrules · Pull Request #29703 · ArduPilot/ardupilot · GitHub
- Split radians into float and double variants
- Avoid conversion to double
- Can be merged once the rounding problem’s been fixed
UTC0123 - hwdef: add relay function 1 to all defs with realyx_pin_default defined by Hwurzburg · Pull Request #29709 · ArduPilot/ardupilot · GitHub
- Set relay function in hwdefs where pin default is set
- Vtx function would be a good thing
- We can add a hack in place to set the function to 1 if you’re setting a default pin
- There might be a problem with conversion code which is supposed to do this?
- PH will try to reproduce the problem
UTC0135 - AP_Param: fix unexpected truncation of AP_Float parameters in expressions by tpwrules · Pull Request #29711 · ArduPilot/ardupilot · GitHub
- Thomas has fixed the truncation of AP_Float parameters
- Tridge thinks several of these might be important
- We’ll get both PRs merged in soon
UTC0142 - Add .bitmask attribute to enumerations by peterbarker · Pull Request #1012 · ArduPilot/pymavlink · GitHub
- Tidy enumerations, remove invalid values from bitmasks
- Merged!