I’ve been doing a lot of work with characterizing wind rejection and noticed that position hold has some sort of wind estimator. No other GPS enabled modes seem to. This seems like something that maybe could be actually more formally part of the navigation system as having a velocity feed forward that’s based on the perceived steady state wind, at least intuitively to me, seems like it could improve the precision of operating in non-zero wind conditions.
Just spit-balling here, but generally it seems like it should be possible to compare the drone airspeed and ground speed based on its current tilt/throttle output and ground speed, and use that to filter out a rough wind estimate to use in a FF loop.
This could also be used for stuff like precision landing, RTL, loiter, etc.
As an aside, has anybody had experience with how well/poorly the wind estimation/ff rejection pos control works and really why would one use loiter over pos control or vice versa?
it would also help with battery life estimation on long flights, if you know the wind speed, ie if your flying into the wind it could estimate your distance/flight time remaining.
@George_Muirhead DJI does that and I must agree that it’s a very useful information.
@lenorin That’s a very interesting point, I was wondering this myself a couple months ago. Please keep us updated on what conclusions you drew
I know that Paul Riseborough has done a draft implementation of a wind estimator for multicopters. I don’t think it’s in master yet though. Hopefully we will get it in eventually.
The big difference between PosHold and Loiter is that in PosHold, the pilot controls the vehicle’s lean angle (like a DJI apparently) while with Loiter the pilot controls the velocity (although it’s actually more like an acceleration controller with friction added).
Thanks for the reply @rmackay9. Is there currently a timeline or goal for when this feature may make it into master?
Here’s the discussion where Paul brought it up. https://groups.google.com/forum/#!topic/drones-discuss/l94eoJRutG4
There’s no specific timeline I’m afraid. I suspect it’ll go into master sometime somehow… The options to try and increase it’s priority are to:
- pick up Paul’s work and try and do it yourself
- try and convince Paul that it’s something he should do for the community
- if you know of a company that wants this, get them to fund the development. I’m definitely not pushing this 3rd option though. I’m just mentioning that that’s how some things get done.
How does the new autotune poshold with turning into the wind work? Does that have a wind estimator, or a simpler mechanism?