On Wednesday 13 February 2008 01:12, Jeff McBride wrote: > > Thanks for your comments - they fit with what I'm finding & > > thinking. > > > > Sure, in piloted aircraft takeoff is handled by the pilot, who > > is able to cope with a wide range of possible failures at this > > critical phase of flight, but I'm trying to work towards a UAV > > control framework, from takeoff roll to landing stop, so that's > > why I'm messing about with automated takeoffs, landings and > > scripted flight etc. > > > > I also find auto takeoffs, landings and scripted flight useful > > for general FDM config & controller tuning for piloted aircraft > > - it eliminates variations due to the pilot, in this case me:) > > > > LeeE > > My experience with UAV control systems is that scheduling gains > with airspeed is a good idea if you want control over a wide > envelope. It is difficult enough to make the vehicle stable over > the entire flight envelope. If you want anything approaching > optimum control, you need to have gain settings for at least 2 > different airspeed regimes. The discontinuities that cause the > "kick" can be eliminated by doing a smooth interpolation between > the gains (rather than suddenly changing them when airspeed > crosses a given threshold). > > I've had success using a "low speed" and "high speed" set of > gains with linear interpolation between them. Of course, I don't > know how you could do this in flightgear. Perhaps a nasal script > could do the interpolation and update the PID gain properties? > > -Jeff
[and to R. van Steenbergen, who essentially made the same point] atm, it's not possible to vary the gain in the FG PID controllers during flight, however, I think RoyVO may be looking into implementing this. It was one of two items on a wish-list I posted here about a week ago:) Varying the PID controller gain is actually something that I've wanted for a couple of years now, but the need has only recently become pressing enough to ask the developers to consider looking into it - they've already got enough on their plate to be getting on with. I also plan to have a look into adding a simple 'gain' type filter, which I think could be useful for this sort of stuff, so I can vary one property factor according to another. I'm also thinking about aerobatic control here where, for example, at +/-90 deg roll the rudder acts as the elevator and the elevator acts as the rudder. In combination with simple gain filters, I think it should be possible to include a pitch input controller to the rudder, along with a yaw input controller for the elevator, and then use roll-deg to set the gain for each of the controller outputs (or gains if/when it's implemented) e.g. at zero deg roll the pitch input to the rudder would be zero but one (normalised) to the elevator and visa-versa at +/- 90 deg roll. Should also be possible to include a pitch controller for inverted flight and control it's influence by roll in the same way:) Sometimes I really wish that I had the necessary data to do JSBSim configs for some of the aircraft I've done - then I could also play with some of the control stuff that JonSB has incorporated in to JSBSim. It's pretty unlikely that I could get good enough data for the aircraft I'm interested in though, so for the time being I'll be sticking with YASim. Fun stuff to think about and play with:) LeeE ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
