Autoland, throttle not cutting out, aircraft not pitching up enuf

#1

Can someone point me in the right direction here? Autoland loiter at 20m above the ground was fine, then exit the loiter on the landing heading is fine, pitch down -18% and throttle down from 70 to 20% for the steep descent, then, I see the flare starting at 8 meters above the ground at 14:17.2. Pitch setpoint changes from -18% to 0.

Then from 14:17.2 until 14:24.2 the final phase of approach, pitch setpoint gradually moves from 0 to +2% and Airspeed bleeds from 12.5 down to 8 and thrust at 20%.

The aircraft hits the ground at expected altitude, about 151m MSL.

My question is, how do I get it to (a) cut the throttle, and (b) pitch up further to SLOW DOWN! I would like to land at 0% throttle, 10% pitch up, (Stall speed is about 3 once in ground effect).

Thank you!

https://logs.px4.io/plot_app?log=7d27d0af-d9f5-471c-a025-8a6aa757a81d

1 Like
#2

I’m not exactly sure why your vehicle thottles up higher right upon landing. However, we never use the default FW_LND_TLALT to control when the throttle cuts. We set FW_LND_FLALT and FW_LND_TLALT to the same value. Not sure if this has anything to do with your issue, but we always see throttle cut at the same time as flare.

Use FW_LND_FL_PMIN and FW_LND_FL_PMAX to tune your flare. when you say 10% pitch up I assume you mean 10 degrees. if so, you can try setting FW_LND_FL_PMIN to 10. although you might want to slowly work up to it a couple degrees at a time (4, 6, 8, 10) to test which is actually the best value for your vehicle.

Also, can you vehicle really land at a 30 degree decent (glide slope) angle? you have FW_LND_ANG set to 30.

1 Like
#3

@Christopher_Milner a couple other things to consider once you get more confortable:

  1. You also seem to have slightly high vibration levels. Check the balance of your propeller(s) and motor(s).

  2. your vehicle could use some improved gain tuning, especially in pitch.
    https://docs.px4.io/en/config_fw/pid_tuning_guide_fixedwing.html

1 Like
#4

Good catch! Did you observe this from the Raw Acceleration chart? Indeed on this flight the motor was a bit loose from a prior hard landing. That flight (with the hard landing) (and hence without the vibration) is here https://logs.px4.io/plot_app?log=32855462-6f94-411c-b63f-6a9c679fd7e3

Thanks very much for your other comments it gives us something to go on. Re, 30 degrees, My understanding from here https://docs.px4.io/en/flying/fixed_wing_landing.html is that I can descend at FW_LND_ANG=30 degrees prior to flaring, until FW_LND_FLALT=8 meters, then, after flare, (reading the code here https://github.com/PX4/Firmware/blob/63651da3097441d675b72478b7a7bd4ebb77c31c/src/modules/fw_pos_control_l1/FixedwingPositionControl.cpp#L1559) the code seems to limit throttle based on altitude and also distance from the landing waypoint, and that distance is calculated based on some view of of the flare length, which I haven’t been able to reverse engineer (calculations here https://github.com/PX4/Firmware/blob/a6dabbbae7a976893d683f1c36654c6ca66ae1c9/src/lib/landing_slope/Landingslope.cpp#L59)

But I’ll try your suggestions and see how it goes.

1 Like
#5

yeah, it just looked bad from my experience. Also the Power Spectrum chart looked pretty bad. The previous flight shows slightly less vibration.

I was looking at your previous flight and you can see that the estimated and commanded roll angle diverge greatly at 6:45. This is what led to your crash. not sure why, but maybe a servo/linkage issue?

1 Like
#6

yeah, but your will be aerodynamically limited to your glide slope at whatever your landing speed is. In practice many airplanes can’t achieve that steep of a glide slope through the air mass. with a stiff headwind though you might be able to achieve 30 degrees relative to the ground though.

1 Like
#7

They sure did. Surprisingly, control linkages were intact after that crash. In fact the only thing broken was the vertical stab which I glued back on. (As you mentioned I probably should have also re-glued the engine mount as it was slightly loose on the next flight). We lost sight of the plane about 1/2 way down and if the Pixhawk is to be believed the plane landed wings level, upside down, nose slightly up. Could have been much worse.

I don’t have an explanation for what went wrong on that flight.