PX4 Dev Call: October 27, 2021

October 27, 2021

Join us

Agenda

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

Community Q&A

  • Shai: Part of “AI for wildlife” an AI challenge and he’s interested in using PX4 to compete in this challenge and especially implement a landing procedure. Would it be possible to have a call with his group for a short introduction how to develop for the environment? @dagar is interested to participate. Please reach out on PX4 slack to arrange.

  • Vatsal Joshi: Raspberry Pi MCU support question. It’s almost ready, but there is a small issue in NuttX which prevents it from building. It would be best to fix it in upstream NuttX and as soon as it lands there we can cherry-pick to our fork and use the fix.

  • @Jaeyoung-Lim Added automatic build testing for avoidance but GitHub actions seem to be disabled for that repository. Daniel directly enabled it during the call. Now we should be able to test the builds again.

  • @dagar Could we remove the old position mode implementation FlightTaskManualPositionVelSmooth which is not used and hence barely used anymore. The main issue is that collision prevention (avoidance) is only compatible with the old implementation. @MaEtUgR will ask Tanja the original author on what’s the easiest way to proceed. It needs to be ported and then we can remove the old position mode implementation. The biggest question mark at the moment is testing. Is there a testing setup in simulation?

Project updates

  • @MaEtUgR Addressing latest review points on RC cleanup. Worked through a lot of bench testing and fixed all corner cases found. Should be ready to merge after the newest changes.

  • @Jacob_Crabill Update about new output configuration improvements by @bkueng . He was testing and comparing to his pr with UAVCAN actuators and it’s working like expected.

    Discussion about how to transition current mixing to the new dynamic control allocation. Multicopter should be pretty close to be able to transition. Plane, VTOL, … still need quite some work because there are currently no replacements and almost no testing. Runtime performance needs to be analyzed, it’s still a bit slower than before. The ESC feedback should be dispatched back to each motor function. How do we transition helicopter mixers? We don’t have hardware so we’ll port the existing mixer to the new architecture but since we can’t test we have to add warnings and find someone who flies an autopilot helicopter. Adding simulation support would be useful.

Roadmap

Discussion based on board: Roadmap · GitHub

High priority queue

Discussion based on board: High-Priority Queue · GitHub

In-Depth discussions


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