ThomasRigi
commented
2 years ago Addition: here's a flight log on master that shows the same type of deviation (still using test BT deviations v3.plan): https://logs.px4.io/plot_app?log=48c8d879-84ee-40e1-9e6e-86cea6efb3eb
I realised that there's an issue with the 1040_standard_vtol file that I sent. You need to lower the transition airspeed to say 16m/s. I attached an updated version of it. 1040_standard_vtol.zip
Describe the bug Porting the slack conversation https://px4.slack.com/archives/C60QQ6TJA/p1630915014024800 with @sfuhrer to github as I found another (simulated) occurrence.
Thomas Stauber 9 days ago Can someone help me to find out where the drone wanted to go to at the beginning of the back transition? https://logs.px4.io/plot_app?log=8b838590-f274-42f0-a0ba-f346c9436d2f I really don't understand why it was rolling left at first. (37:32-37:35) (edited)
Silvan Fuhrer 9 days ago looks a bit like it was tracking home (0/0), in the very first phase of the transition, and rolls left for that. But why did it then switch..
Silvan Fuhrer 9 days ago how did you plan the mission landing?
Thomas Stauber 8 days ago Attached is the plan. I could reproduce it on the simulator, but only with a specific set of parameters. I tested on v1.12.1 and master. With default parameters on master it doesn't do it. Neither if I move the WP after the back transition a bit. I'm even more confused because I really don't see what moving a WP by some meters would change... QGC bad BT (1).zip 1040_standard_vtol.zip
Silvan Fuhrer 1 day ago Thanks for the mission plan, that helps to get an overview of your situation (for future reference I attach a screenshot) . But honestly, I have still no idea what was causing the roll to the left. I would keep an eye on it during our flights, and investigate further if I'd see it as well.
To Reproduce in a slightly different way With this new mission file (based on default SITL spawn in Zurich) I think I ran into the same problem again. Only that this time it's rolling towards the right whereas previously it was towards the left. Moving the WP after transition works again. This time it doesn't look like it's trying to fly towards home as the deviation is not strong enough for this. But I still don't know where it's heading to. Flown on v1.12.1 with the same parameters as above. test BT deviations v3.zip
New log file: https://logs.px4.io/plot_app?log=58b3733c-714a-4101-bb43-3a251b5d91b4
Expected behavior The drone should fly straight towards the next WP.
Additional context I have also found another bug for deviating back transitions if the next WP is at a different altitude, but there I know what causes it. Will open another issue and subsequently a PR for that when I get to it.