I have posted a Flight-Review log
of a SITL gazebo simulation of a novel medium size UAV, with 6 rotors arranged out-of-plane for 6 degrees-of-freedom, executing a square flight plan at 50m altitude.
PX4 code has been modified to generate transient direct thrust toward the desired direction, in addition to the usual tilt of vehicle attitude.
PX4 code [commit df42ef84f1166afe9ff281f312a9701930a3ec35 origin/main
Date: Fri Jan 16 05:40:09 2026 +0900]
The code modification does not provide the user with independent control of thrust and attitude. Instead, the motivation is to get quicker response to vehicle position errors than conventional attitude-only control can generate, especially for a vehicle with large angular inertia.
The physical model has mass 5.2 kg and angular inertia Izz = 1.38 Kg m^2, Ixx = 0.76 Kg m^2 (measured). This angular inertia is high because the frame is a large stiff cage surrounding 6 relatively heavy 5-bladed variable-pitch rotors.
The logs show e.g forward thrust leading change of vehicle pitch by about 1 to 1.5 seconds. So I tentatively conclude that this control strategy gives at least 1 second quicker response to disturbance than the standard PX4 control, for this particular vehicle.
The Flight-Review log likely shows faults that I miss. Any comments that might prove helpful to improve performance would be very much appreciated.
Don