PX4 Dev Call: May 26, 2021

Coordination Calls PX4 Autopilot
1

May 26, 2021

Join us

Agenda

  • Community Q&A
  • Roadmap
  • High priority queue
  • Release
  • In-Depth discussions

Announcements

If you haven’t seen it already check out the ARM innovation coffee

Community Q&A

  • @david_s5 : make clean does not work as before anymore it calls the actual cmake cleaning for all build artifacts instead of just deleting the build folder. If you still want to wipe everything use make distclean it should not remove your local changes anymore and the goal is to add all binary dependencies of the build in camke such that incremental builds always work and make clean confirms all artifacts are removed.

  • Eohan: Asking for an update on the batmon pull request:
    Adding BatMon smart battery as a module and refactoring SMBUS based SBS 1.1 spec to a library by echoGee · Pull Request #16911 · PX4/PX4-Autopilot · GitHub
    The pr is ready. It does use some flash on every board but if the parameter is not set it doesn’t run by default.

    batmon start -X should look on all I2C buses and find the device. But reportedly batmon does not connect if the -b bus is not specified. That has to be checked.

    How are I2C addresses assigned? batmon comes with a default address and if you have multiple batteries there’s no automatic detection yet. It looks like the driver should check multiple addresses and start an instance per address that responds with an expected memory content.

  • @JulianOes Safety (switch) is now off on boot by default. There’s confusion about the behavior now because it doesn’t behave anymore like before and it’s not clear what the benefit/intended behavior is. There are unexpected side effects like the safety-off tune interrupting the PX4 startup tune and fixed-wing servos moving when they are not expected to. We need to read up on how it behaves now. Discuss a simple way it should work and implement that. Follow-up discussion slack/call.

  • @bresch With multiple EKF instances running with different combinations of sensors the CPU load for expensive operations e.g. 3D magnetometer fusion multiplies and can easily jump ~15% of the total processor load.

    We can reduce the update frequency of the expensive heading fusion. Also, we need to make sure the hardware in the loop test is covering all possible code paths e.g. different fusion combinations.

Roadmap

Discussion based on board: Roadmap · GitHub

High priority queue

Discussion based on board: High-Priority Queue · GitHub

Release

In-Depth discussions

For smaller groups expanding technical discussions, stay until the end and follow up.

If you have any feedback or corrections please comment on this post.