PX4 Dev Call: May 27, 2020

May 27, 2020


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

Join Meeting

Meeting ID: 946 175 205
Join using your mobile/desktop

Call in using your phone, find your local number: https://zoom.us/u/aetSYKbiMF

Component update

System Architecture

@dagar There’s a pr pending that changes the place where IMU data is integrated and synchronized. Before all the integration depending on the configured sampling rate was done in the individual drivers and now it’s happening in the sensors module making sure it takes care of e.g. gyro and accel data coming from different drivers getting synchronized correctly for the estimator.

OS / NuttX

@david_s5 backported some commits to fix some DMA issues for fmu v6x. He’s aiming at getting to the latest NuttX release but there’s a lot of upstream changes and incompatibilities that don’t get catches by the compiler because of weak typing. It would be nice to get our fixes onto upstream NuttX first for the full full H7 support.


New driver for ICM20649 which @Jacob_Crabill fixed. Mag rotation has to be checked still.

The old driver can be removed:

We discovered multiple inconsistent mag rotations in the drivers. For example the external GPS unit from mRo.

@Jacob_Crabill made a new rewritten FDCAN UAVCAN driver for H7:


Relaxed disarm command checks allowing disarming in “maybe landed” state:


@dagar asks @Roman if he could follow up on

there were some new reports with drifting altitude. There were multiple questions and a request for logs. The report now is about 5-minute drifts and not quicker up and movements on a Holybro Pixhawk 4 mini.

Everyone who could test altitude hold performance and hello isolate the issue it would be helpful to have different logs with the primary height source EKF2_HGT_MODE set to Baro (default) and GPS. And when flying with barometer having different ``


There was an issue reported by Thomas Rigi (not on the call). The backtransition should be controlled but needs to be tunable to not be too agressive.

@maetugr should look into this pr because if you take off a VTOL in fixed wing mode the multicopter takeoff state machine can still be ramping and make the vehicle fall during transition.

Fixed Wing

No updates


Is it expected that the mixer does not actuate at all anymore if the collective thrust is on zero @maetugr expected the two motors that are not saturated to still actuate without raising the collective thrust like airmode does. @julianoes has some concerns because of overturned oscillations leading to a flyaway but the default settings do not allow zero collective thrust and a low collective thrust should not lead to a flyaway with only two motors actuating. Further discussion with @bkueng and @bresch needed.


  • New plugins not working yet in language wrappers -> fix coming.


  • ROS Noetic is out, and MAVROS was release to it yesterday (https://github.com/mavlink/mavros-release) - the new px4-dev-ros-noetic container is currently on hold while MAVROS is not available to be installed by apt.


No updates


CI is broken. @JulianOes found a hard to find fix though multiple changes. Flush stdout for every line for SITL such that the PX4 output gets actually logged, offboard acceptance radiuses were to large and for takeoff land it would land again too quickly before reaching the desired point, general acceptances to pass the check were widened a bit:

@TSC21 Made a pr against MAVLink to have an ID field for the simulation GPS message. The idea is to allow multiple GPS input to be simulated and allow testing GPS blending in PX4.


Peter had multiple usage questions on how to get the latest state working for his demo setup. @dagar is updating quickly about the possibly missing pieces that can lead to problems at the moment. They are arranging a call to get things sorted and quickly progress.

Roadmap, and Release discussion

Last week we were discussing cutting a new release candidate. There’s beta 2 for testing at the moment. The reported altitude hold problem is the single biggest issue we’re blocked on and the magnetometer rotation problems with some modules are also nice

Community Q&A

Mariano: Question about HITL. All widgets in QGC to control HITL were gone and they are back now and he couldn’t find any pull requests or issues that would explain the roadmap for the feature. @LorenzMeier is convinced HITL will gain importance for hardware verification use. That said SITL is a much easier setup for other testing. There’s a simple interface in MAVLink to run HITL and using that interface could make the specific use case completely independent of QGC. What’s not working in terms of HITL is just a subset of simulator and QGC UI. The autopilot should run HITL just fine and the configuration of the vehicle should not need to be changed to be able to run it.

In-Depth discussions

Errata and Feedback

Let me know below if I failed to capture anything the right way, and if there are any updates, or you have feedback on the call format.