now I’ve flown a auto track composed of spline waypoints. I expected a smooth ride and the footage, but during this flight, strong yaw oscillations happened, what is perfectly visible in the video. Not only during the DO_JUMP (is it possible to close the loop using only the spline fragments?). I’ve compared the desired yaw to the actual yaw, and it seems they were commanded (not mistuned yaw PIDs?)
I have not much experience about this, could you explain why this happens? The heading was set to follow the GPS track.
Txs very much for the report! I agree it does look like the yaw changes are commanded as we can see from the desired vs actual yaw graph below (they’re very much in sync meaning this is not a tuning issue) so I’ve added this to our 4.7.0 issues list.
I suspect we can reproduce this is SITL but I wonder if the same behaviour happens with 4.6.3. No huge pressure to try that of course, it would just immediately tell us if we’ve got a regression on our hands.
Txs again and I’m very envious of your flying location!
I feel like we might be able to tweak the spline weighting a little to get a nicer curve but you can see that your actual path is much less smooth than the target path.
Currently we use the actual direction of travel to define the yaw so when we have wobble like that in our track we see a wobble in the heading too. This is another area where we could do better by using the target velocity instead of the actual velocity.
