We have a problem with disarming the hexacopter after landing.
It perform perfect several times, then suddenly it wont disarm. The behaivior is that it decreases rpm on the engines, buy then it suddenly increases again and the copter tips over.
There seems to be a barometer issue, as the temp shows -1 no matter the enviroment. Today we attached an lidar (st20) downfacing. It works perfect in measurment, however it ended up the same way. We tested the drone for a little over 1 hour, with varius take off and landings included disarm/arm.
It ended up with a tipover as described above. We runt MP al the time, and both ekf and vibe were in marginals. In the log there is a little increase of vibe at the end.
Yes, I was facing many time this problem (copter tips over =/= copter flip over) , so I’m only thing to do , 1= reset all parameters and load firmware and that’s done
We’ve recently done some investigation into this and we have an improvement that will be released with Copter-4.0.4. It’s a bit detailed but the fix involves increasing the filtering on the accelerometer values used when deciding on whether to prioritise altitude control or attitude control.
What’s happening is that as the vehicle touches down, the accelerations actually increase and this upsets the landing detector. If we graph the raw IMU values we should be able to see the increase.
I suspect if the vehicle touches down on softer ground it has no problems. Anyway, there’s a fix coming in Copter-4.0.4 and your help in testing this during beta testing would be greatly appreciated!
We’ve found with a Quad X8 that it needs to be landed a fraction harder than what you want, so it must register the bump/vibrations from hitting the ground. Then it always disarms reliably.
If you land it nice and soft, like you’d think is ideal, it sometimes wont disarm and you have to kill it with emergency motor stop. We have the emergency motor stop configured on a channel for testing but have left it in place for this UAV.
A smaller Quad has no trouble at all deciding it’s landed and disarming.
In addition its like you say. It works when landing on soft ground ( field ), and this behaivior accures when we land it on the road. The logs that i sent you, shows how it lands and disarms several times on a field with grass on it. After the latest landing we move the drone to a road and try to land it there.
Due to that we had this problems since 3.8, we always testing it with a rope attached to the drone, so we can cut the power. In this case it didnt cut emediatly, so the drone took of and crasched from aprox 1 meters hight.
We tried to attach a lidar to help with the measurement due to the problems with the barometer. In test there is a differens between barometer and the lidar with aprox 1-2,5 meters. It doesnt seems like the barometer resets, because it indicates ( on MP ) that it 2 meters above ground level, when its started up. Also, the temperatur always shows -1 no mather what the enviroments are. The same when one is inside in +22 degrees.
That is really interesting! At the same time its the oposite of rmackay wrote. In that case it wouldnt disarm on soft ground, and work perfect in the road?
It seems to me that it would be better if it senced that the throttle is low and it stops decending, that would trigger disarm after like a second or so. Our drone has costed about 300000 euro to develop, so a crash and not be able to trust the software is quite painful. // Joakim
I think this would solve the issue ( if you use a downfacing lidar ):
Disarm the engines if the throttle stick is down, and the lidar is below the configed mounting. In our case its mounted on 25 cm. So what i mean is, if the throttlestick is low, and the lidar is at 26 or lower for more than one second it disarmes.
In landing larger platforms, sometimes a gust will pick it up, and ‘cushion’ before landing. Would automatic disarming recognize this gust as ground, shutting power pre-maturely?
Are laser altimeters part of the decision making for landing recognition? If so, no expensive copter should be without one.
Those log downloads expired. Can you post them again?
I’m not positive I’ll be able to see anything specific, landing is tricky to detect, so I gather.
