VTOL tail-sitter duo rotor with thrust vectoring, PX4 or ardupilot?

I’m building a VTOL tail-sitter duo rotor with thrust vectoring.

Does anybody have advice about about the most adequate autopilots solution for this platform (Ardupilot or PX4) ?

I guess most of people here have only experience with PX4 :slight_smile:

What do you exactly mean with thrust vectoring?

Thrust vectoring means there is control surface which vector the thrust (typically elevon)( as opposed to the tilt Motor solution)

Ah yeah so that one is supported by PX4, it’s the “VTOL Duo Tailsitter” airframe.

I am using PX4 for a “VTOL-Duo-Tailsitter” airframe. Except for some minor issues, the system works quite well so I would recommend PX4!

1 Like

Could you please list “minor issues” please ?

@joihn any success ?

  • One major issue occurs during takeoff. Without wind, the tailsitter climbs to the specified altitude in MR-mode, performs the transition to FW-mode and flies in FW-mode. If there is wind, though, and the tailsitter drifts with the wind during climb in MR-mode and stops climbing before reaching the target altitude, corrects the lateral position error hovering and continues to climb to the specified transition altitude. I didn’t find any possibility to change the parameters of that behavior. I would like to specify the radius within that the tailsitter performs the lateral error correction. In many cases, when there are no obstacles close, I prefer a continuous climb, even with a lateral error rather than a delayed takeoff that consumes additional battery energy (hover is very expensive). Is there a way to influence this behavior? I didn’t find any parameter for that.

  • The artificial horizon is correct when in MR-mode. When in FW-mode, the artificial horizon displays wrong angles, e.g. pitch should be rotated by 90°. I guess that is just a small software-bug?

Interesting, I don’t know px4 that much … I was on ardupilot, tried px4 but not with vtol. Advantage of PX4 is SITL with gazebo is more developped that you can try some parameters without real field usage.

Did you try daily release of QGC for horizon level ?
Did you fill issues on github for transition?

Some issues worth to know :

Honestly experimenting with Ardupilot in Realflight is much more realistic, especially when we need to validate some additions to the code.

Realflight ils not opensource and mono platform (windows only) so not really relevant and able to make comparison…

And what does it have to do with what I wrote? :laughing:
Whether it’s open source or not is irrelevant to what I’ve written.

You compare Gazebo to Realflight which is a non sense :

  • opensoure vs proprietary
  • Mono platform vs Multiplatform
  • It’s not diffucult to setup gazebo … assessing it with a video and not a tutorial or other doesn’t make sense… Just saying in a bisounours world : wouhouh you got it working but we don’t care the discussion will not come efficient with such video (it’s clearly relevant on ardupilot discuss forum whre you already post but not here) ! Or may I guess PX4 developers will look at ardupilot PR’s below video you linked, to enhance their FW …

Furthermore posting a video where a weird “mono” platform is present (Mission planner, it could work on Linux (and Mac) but not that much with poor performance and lack of full support. Mono crap thing … ) is definitively not relevant : Mission planner doesn’t support PX4 fw whereas QGroundcontrol support both …