PX4 Sync / Q&A: June 26, 2024

PX4 Sync / Q&A: June 26, 2024

Agenda

  • Announcements
  • Release Discussion
  • Q&A

Announcements

Release Discussion

Q&A

@frogzhang This looks for an older main.

@Noe-Sanchez

  • Which is the correct/recommended frame vehicle_visual_odometry estimation data should be in, if using the XRCE-DDS agent? Reading this documentation suggests FRD, Ive tried FRD and NED to no avail.
  • If my position estimate and setup is reliable, to which extent should I turn off sensor fusing? For instance, turning off GPS and magnetometer aiding gives me a stable yaw which matches the VICON frame, but when flying in offboard, the drone still spins erratically even when taking off with correct yaw.
  • How could I debug the control / actuator outputs to better diagnose the setup, besides making actual flight tests (this cost me a few pairs of propellers last week)? With the correct logging config, would flight logs be enough to check for errors in frame transforms / position estimate?

Ans from @Jaeyoung-Lim : You might consider checking the frames first since PX4 and ROS are using different coord frames. You may refer to this doc.

@AlexKlimaj also mentioned this as a guide.

@dirksavage88 Anyone has gone through precision landing via mavros?
WIP in ROS2 for this from @AlexKlimaj
@Jaeyoung-Lim mavros does not take care of control and PX4 is handling it.
@MaEtUgR there are two ways to do the control either velocity or position which the latter might not be that reliable!!!

Review needed

I hope tonight’s discussion will help us exchange ideas on my problem(https://github.com/PX4/PX4-Autopilot/security/advisories/GHSA-55wq-2hgm-75m4) and try to identify the loopholes.

I have some questions relating to the use of indoor positioning schemes (VICON to be specific), with ROS2:

  • Which is the correct/recommended frame vehicle_visual_odometry estimation data should be in, if using the XRCE-DDS agent? Reading this documentation suggests FRD, Ive tried FRD and NED to no avail.
  • If my position estimate and setup is reliable, to which extent should I turn off sensor fusing? For instance, turning off GPS and magnetometer aiding gives me a stable yaw which matches the VICON frame, but when flying in offboard, the drone still spins erratically even when taking off with correct yaw.
  • How could I debug the control / actuator outputs to better diagnose the setup, besides making actual flight tests (this cost me a few pairs of propellers last week)? With the correct logging config, would flight logs be enough to check for errors in frame transforms / position estimate?