Problem is that wind speed is also associated with a specific wind direction (ie a “vector”). Simple subtraction of gps estimate from ideal velocity would only estimate the component of wind in the same direction. I imagine, if one lets go of the sticks so the drone tries to just stay in one gps...