Dev Call Apr 09, 2018 2300 UTC

Development Team
2

Attendee count (max): 21

UTC2301 - https://github.com/ArduPilot/ardupilot/pull/7716

  • Setting time via GCS
  • Set system clock via ChibiOS use is wrong
  • Get system clock ms wraps really, really fast
  • Will get it in as it currently stands and we’ll look at ChibiOS support
  • Peter and tridge will look at this

UTC2305 - changing tradheli build

  • waf copter --upload would upload a random target
  • Now a separate shortcut build target
  • ./waf copter and ./waf heli

UTC2305 - ESC telemetry

  • First flight done
  • We have telemetry in the logs
    • Already have the ESC messages because of uavcan escs
  • Blheli32 now available everywhere, not just on multicopters
  • Tridge needs to update the docs
  • Mavlink message?
    • Use mavlink2 zero-trimming
    • ESC mask field which specifies how many in the fields you get are actually valid
    • All fields of the same type in one message e.g. voltage goes in a single message type
  • PH2.1:
  • JW: Why only available on Auxout?
  • [9:12 AM] (Channel) PR: Yes, the level shifter on 1-8 is one way. It’s a high current drive.
    • So it can drive optocouplers
  • JW: DOes this levelshifter limitation apply to Pixhawk 1?
    • Should work on the IO board on PH1, but development bandwidth required
    • D-shot on io channels in the future, maybe?
  • PixRacer works on all channels
  • MindPX will work on all 14 channels (they’re all FMU)
  • PH2.1 will be able to do telemetry but never configuration
  • IO microcontroller needs to be killed?
    • Philip: Most people getting 2.1s do use the passthrough
    • Trouble getting PH2 users involved in the community
  • Any example of a vehicle being saved by the IO?
  • Why do we keep iterating on hardware?
    • What problems are being solved?
    • “Faster” isn’t necessarily sufficient
  • All telemetry comes back on a single pin back to the autopilot (solder together, it timeslices)
  • [9:21 AM] (Channel) JP The telemetry request goes out on the dshot pins, but the response goes to a uart (sequentially).
    • JW: Now that I think of it, I guess there’s not much reward to putting dshot output only on the main pixhawk pins. The biggest advantage dshot gives us right now is the passthrough setup. pwm/oneshot is fast enough for our control loop and works well to date
  • AT: jacob: there are other advantages. 1) no ESC cal needed 2) easy setup of reversible motors
  • AT: you need passthru to enable it on the ESC, but once enabled it is easy to use

UTC2320 - GSoC

  • Came up with top
  • Will request slots
  • Google to respond in next 24-48 hours on how many slots we get
  • Week or two to decide which candidates to take on
  • End of month everything is announced
  • Coding starts some time in May?
  • Some great proposals!
  • May suggest to partners they grab a student or two

UTC2322 - Luis and funding updates

  • contacted all renewing partners
  • Next week a cleanup of the partners list
  • Asked SPI for numbers on current position
  • Need to talk about the swag shop with SPI
  • More funding requests needed!
    • Please submit!
  • Some companies are interested in supporting AP in specific ways
    • Targeted funding
    • Randy’s working up a document on the subject
    • Trying to sell AP to potential donors
      • Point out where their needs are on our roadmap
      • E.g. compasses are a problem for our users, and Paul want to reduce our reliance on it
      • Hand it out at big drone events?
  • Weak-spot identification
    • Let funding committee know about places that can be fixed so funding committee can chase things with potential donors
  • If you are talking with potential donors, contact Randy and he can help you use the document he’s preparing

UTC2328 - Copter beta testing

  • This week for sure!
  • ChibiOS?
    • We’ll do both NuttX and PX4
  • Rover 3.2.3 went out
  • Boats appear on ground station as boats
  • Variable pitch quadcopters are fully tested and documented now
    • Code was done a long time ago, but it wasn’t documented
  • DI: Variable Pitch QuadCopter (reconfirming it works)
  • PID setup if you want to be able to flip it over in-flight
  • Reversing ESC support for normal multicopters is coming!
    • If you want to help in testing, tridge wants people to be aware that they’re likely to crash but he’d appreciate testers!
    • Screw-on-props not recommended

UTC2330 - Rover

  • Next week for Rover 3.3 beta

UTC2330 - https://github.com/ArduPilot/ardupilot/pull/8077

  • Rcin moved to hal?
  • RSSI into RC Channels or directly in HAL?
  • How does it interact with RC overrides?
    • get_RSSI call does what?
      • Comes from AP_RSSI?
    • Function in rc channels could be get_receiver_rssi() to make it clearer
  • Need to pull protocol stuff out of HAL
  • Also interacts with AP_IOMCU library
  • Eventually get rid of hal.rcin
  • Thin wrapper around hal.get_rssi in the short term?
  • parameter handling is going to be a pain
  • parameter renaming will need starting at
  • FF will be coordinating with MdB on getting this PR in
    • Tridge consider this a preparatory PR for future work

UTC2342 - Plane release?

  • 3.8.5 or 3.9?
    • soon!
  • Barometer log bug fixed
  • Master flown on quadplane on the weekend and it worked well

UTC2342 - build releases

  • Fmuv3 fmyuv4 mindpx revomini
    • Peter and tridge are working on autobuilding these

UTC2342 - Bill and tradheli

  • User noticed coupling between roll and pitch
    • Bill is working on a fix for it
    • Swash-plate bug
    • 3.3.3 -> 3.4 seems to have introduced something
      • Trim was carried in calculation for servo_out
        • Delta was carried from midpoint through to servo out
        • No elongation based on the servo range
      • 3.4 changed to the -1 to 1 range stuff
        • elongation/squishing due to distance between trim and endpoint
        • Servo doesn’t move enough to keep orientation between all the servos
        • More cyclic input to one side than another
    • Rob’s heard there might be an issue with settings not saved correctly too
  • Two Thundertiger 700s are being built up
    • One petrol
    • One electric
    • Currently don’t support the swash plate used in these
      • Development is being done
  • Support all know swash plate types with a “universal swash plate” setup
    • Nothing on forums ATM, but soon
    • Compensation factors for different swash plate types
  • Working with Randy and Leonard on spooling code, hoping to get it into 3.6
  • Longer servo arms used on some swash plates require a scaling factor, being added
2 Likes