On 06/30/2007 02:21 PM, Hans Fugal wrote: > Nice explanation.
:-) > What would happen if the yoke > movements were additive to the autopilot, similar to how they are > additive to the current trim settings Interesting question; see below. > (and indeed the autopilot changes trim settings). Well, some autopilots do and some don't. The KAP140 for example may ask /you/ to move the trim, but it won't (and can't) move the trim by itself. There are reasons to prefer autopilots that don't mess with the trim; it limits the amount of damage they can do when something goes haywire. > Then even a slightly noisy joystick would only > cause a tiny war between the user and autopilot. Agreed. But we need to continue the analysis to cover other scenarios. 1) In particular, what happens (and/or what should happen :-) when the user deflects the joystick a goodly amount to (say) the left and holds it? If the autopilot's opinion is simply additive, it will soon track out the perturbation in the usual feedback-loopy way. Now the user is in a situation where the joystick is deflected one way, while the aircraft yoke is deflected the other way. Ick. 2) Then things get really weird when the user lets go of the left- deflected joystick. The system will see this as a user command for a goodly amount of /right/ aileron (in effect, in the "additive" sense) ... even though the naive user probably didn't think of it that way; all he wanted to do was "let go". This gets back to the fundamental problem mentioned in my previous note: The semantics of joysticks is just plain different from the semantics of real aircraft yokes. As you learned in high-school physics class, force is not position, and position is not force. In the real aircraft, the force on the yoke is nowhere near being in a one-to-one relationship with the position of the yoke; you can zero the force (by letting go) without coercing the position to go to zero. Alas the joystick knows nothing of this; its force is proportional to position, since all it has is a dumb little spring. The aforementioned "hand+yoke" model may help here. The hand has two modes -- gripping and not gripping -- which have to do with force, in the sense that ungripping the yoke zeros the force, without necessarily zeroing the position. Meanwhile the joystick can control the position of the hand. So the hand+yoke model allows the beginnings of a separation between force and position. I'm not saying this would be easy to implement, or worth implementing, but it has value as a conceptual model, to illustrate just how deep the problems lie. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
