PX4 Dev Call: November 11, 2020

📞 Coordination Calls PX4 Autopilot Coordination
1

November 11, 2020

Agenda

  • Status update by component/project
  • Roadmap, and Release discussion
  • Community Q&A
  • In-Depth discussions

Join Meeting

https://meet.jit.si/PX4DeveloperCallWeekly

Component updates

System Architecture

No updates

OS / NuttX

  • @david_s5 is tracking issues that are reported upstream.

  • NuttX incremental build fix

  • Hardware checks are being improved e.g. to catch SD card regressions on all the different boards. A simple way to compare the SD card content with what’s expected would be a nice additional check.

Driver

  • Rewrite BMM150
  • ST IMU on raspberry pi navio shield now uses the same architecture like the other ST IMUs on pixhawk boards.
  • Sensor calibration improvement for board level calibration. We need to make sure it’s clear what this calibration is for and untangle the different rotations that often get misused. Also a concern is the board horizon level calibration

  • There’s an issue with gyro calibration where the message to notify QGC about the calibration being complete.

  • There are more reports of fmu-v2 -v3 loosing their configuration parameters after a reboot. E.g. when commanding a reboot over MAVLink after calibrating.

Commander

  • Set the home position while flying if there was no GPS at takeoff
    It’s ready but it’s not clear how to move forward. Do we need additional guards for back computing the home position? Should it be enabled by default?
    The possible problem is that if you fly with optical flow and gain GPS lock after a long time the back computation might be quite inaccurate. The
  • Taking off without GPS should not be possible when RTL is the failsafe reaction. The dependency should be captured in commander and the independent user configuration gets unnecessary.

Estimation

  • Nice refactoring! EKF main is a lot more readable now. Saves a lot of MCU RAM because of smaller scopes.
  • Testing wanted on this pr with optical flow vehicle. Refactoring and a few bug fixes. No danger.
  • Extended EKF testing on representative replay logs would be a good addition to automatic CI testing. We have to take care that useful changes still pass even though they are not non-functional.

VTOL

  • CI tailsitter was outside of the 1m corridor around the square mission it flies. @sfuhrer suggests having a quick look at the tuning. Before we do more digging we should see if there’s a simple fix because it’s just slightly over the threshold.

  • VTOL offboard question, need to follow up

  • Idea to remove the hacky yaw feed-forward for VTOL. In fixed-wing position control mode you can move the yaw stick and it’s directly mapped to the yaw rate. Coordinated turns are also achieved without.
    It seems to be useful for fixed-wing landings:
    https://px4.slack.com/archives/C7L8W6DAL/p1605022478012200

Fixed Wing

  • Disable the manual climb out mode
    There was no feedback that anyone has objections to remove it. We’ll check the takeoff use case to still work and continue with the pr.

  • Suggestion to replace airspeed setpoints with something more intuitive in fixed-wing position mode. The goal is that you can guide the vehicle and it flies where you command at cruise speed. There’s no point to fly at an inefficient airspeed. @sfuhrer suggests a configuration bitmask to

  • Bug for fixed-wing landing

Multicopter

Thanks for the test and feedback. Next: make CI happy, merge. Check flash space. Remove oldest smoothing mode shortly after.

Rover

  • @bperseghetti adding NXP cup car from the academic competition. Coming soon.

Avoidance

No updates

Simulation

  • Contribution from @bperseghetti to with more powerful templating and dynamic world loading.
    It’s now possible to configure time to have correct sun location, shadows and even light sources on the vehicle.
    @Jaeyoung-Lim Discussion about the HITL spawning solution. Single HITL instance improvement vs. multiple HITL instance use case. HITL should not have a different way of spawning than SITL. We don’t need to block but unify it again.
  • GitHub workflow @dagar split up such that every vehicle type and location runs on its own node and hence runs more than twice as fast.

MAVSDK

No updates

MAVROS / DDS / ROS2

ROSWorld tomorrow! Sign up for free!

UAVCAN

DS-015 standard document version 1 is finally released :tada:

Release

Community Q&A

@nrogelio SD card recommendation list (from chat)

In-Depth discussions

Errata and Feedback

Let me know below if we failed to capture anything the right way, and if there are any updates (not present here), or you have feedback you would like to share with the dev-team.