We have currently takeoff problems with an XUAV Mini Talon which is loaded with camera equipment and a waterproof box (https://searchwing.org) such that the drone is quite heavy. As the drone is landing in sea water we have no pitot tube for airspeed estimation and we rely on synthetic airspeed. We track the takeoff problem here:
A video of two stalling takeoff attempts is here: Clare takeoff failure during Resqship mission on Vimeo
Another testflight on ground shows the speed estimations during a successfull takeoff.
Figure 1: Speed estimations during takeoff on ground
Figure 1 shows the synthetic airspeed estimation from DCM (CTUN.Sas), the EKF3 airspeed estimation (CTUN.As), the wind speed estimation and the GPS speed over ground. You can see the large difference between GPS, DCM and EKF3 speed estimations. The wind estimation seems far off as it changes fast after takeoff. This flight was done with 10° target pitch and TECS_SYNAIRSPEED=1. Because the estimated airspeed was so high, the throttle was reduced to 60%. That flight was done on ground.
The dataflashlog of the flight on figure 1 is here: Nextcloud
Has this been addressed, such that I can maybe see a difference in Plane 4.4beta release? I am concerned about two things here:
a) The bad speed scaling due to airspeed estimation. This is addressed maybe with FLIGHT_OPTIONS bit 6 to disable the speed scaling during takeoff.
b) The throttle reduction during takeoff. If TECS_SYNAIRSPEED=0, then this addressed by Plane: added TKOFF_THR_MAX_T for non-airspeed takeoff by tridge · Pull Request #22584 · ArduPilot/ardupilot · GitHub
It seems that the DCM estimation during takeoff is basically the GPS speed over ground but it is probably safer than this EKF3 estimation which seems to be way off?