Each time I’ve flown a large-ish copter, auto or guided mode takeoffs look pretty ugly. Usually there is a strong jerk to one side or another which is quickly caught. I’ve never seen a prop actually make contact with the ground but we’ve had many close calls.
Are there any parameters to slow down how quickly the pixhawk will command a motor to spool up during takeoff? The effects I have seen seem very consistent with there being small differences in time between each of the motors producing thrust. Even if they just increased to mot spin min for a full second before attempting takeoff, I think we would see a large improvement.
I’m interested in this, too. While my quad doesn’t go sideways, it leaps into the air, with gusto, and sometimes results in a “potential loss of thrust” message. It’s so aggressive that I now take off manually to about 50 feet and then switch to auto. Would love to be able to trust auto launches!
I have been wanting some extra parameters to control auto takeoff as well. You have no control how fast the motors ramp up to 100% and then it shoots off the ground for the first 10-12 feet. I also get almost all the time potential thrust loss messages. But have never had them once in the air…always on auto takeoff. I do not understand with takeoffs so important when doing auto flights there is not more control for takeoffs with parameters. I get they want it off the ground quickly, but it is violent and especially on a big copter with some weight to it. Is there anyway to turn off potential thrust loss messages as well?
Randy, is there something in the works for more control for auto takeoffs? There have been many topics and discussions on this from many people in the past 6 months. Thank you.
I’m curious about this as well - we had been using Copter 3.5.2 for ages and the auto takeoffs with it were silky smooth. We updated to Copter 4.0.3 about 4 months ago and since then, the motors spin up much faster even though MOT_SPOOL_TIME and MOT_SPIN_MIN are unchanged.
I think we have a fix for this issue in master now so if anyone is brave enough to put “latest” on their copter and report back that would be great. We have one report from a core developer that this fix resolves the issue. PR: https://github.com/ArduPilot/ardupilot/pull/15541
It’s possible we could backport this to 4.0.x but the start of beta testing for 4.1 is perhaps only 6 weeks away.
I got also very often “violent takeoffs” in auto. Even with 4.0.5. I find out, that the problem is related to the throttle stick position during the start of the auto mission. If I only raise slowly the throttle stick to mid, the copter leads to tip over. If I raise the throttle stick to full throttle from one moment to another the takeoffs were very smooth.
Not sure if this maybe helps to resolve this issue further…
I’m a little surprised the throttle stick position matters when taking off in Auto mode. Maybe you mean Loiter mode or one of the other semi-autonomous modes? In Auto mode, the transmitter’s throttle position is only used as a trigger. So the mission will start once the throttle is raised slightly but it shouldn’t matter whether it is raised completely or just a little bit. Anyway, no big deal but just want to clarify this for other readers as well.
Yes, it’s in Auto mode. I also thought that the throttle stick will be work as a trigger. I see the following difference between full throttle stick and mid throttle stick:
With full throttle stick, the motors spool up quickly, the copter leaves the ground quickly, nearly no tip over behavior is visible.
With mid throttle, the motors spools up not that quickly, some of them rotate faster than other, the copter leads to tip over and the control loop try to recover as soon the vehicle is in the air.
My conclusion was, leaving ground slowly (with mid throttle stick), the control loop kicks in but the copter is still on the ground and can give no feedback. The controller try to compensate even more and this cause this strange violent takeoffs…
In order so solve this issue I also played with WPNAV_ACCEL_Z and WPNAV_SPEED_UP. Small values are very bad and gave me very ugly takeoffs…
It’s very unlikely that it’s the throttle position that is leading to the difference in behaviour. It’s more likely that it is yaw stick (that AP does consume even in autonomous modes) or some other environmental factor I think. If you can reproduce the difference and provide a log I can have a look… but just looking at the code there’s no consumption of the throttle stick position as far as I can see.
I use QGC. when start mission the vehicle go to Guided mode arm and go to Auto mode. this result in aggressive take off. but if I armed wait the motor to spin then slice the start mission the vehicle take off nicely. My assumption is that the control loop of auto mode kick in while the vehicle just arm and need time to spin the motor thus cause the control loop to see much error and throw potential thrust lost and push the motor so hard to compensate the altitude error. Anyhow, it is just my assumption.
even in loiter, when I arm and immediately raise throttle without waiting the vehicle to spin it also result in violent take off.
So for now, My work around is arming -> wait motor to spin -> slide take off.