Posted by Jay Kickliter on February 19, 2009 at 5:48am
What is the usual method for steering a flying wing UAV? The only idea I could come up with us steering by adding roll, and maintaing pitch. My best guess: since the center of lateral resistance is a behind the center of gravity, the plane will pant to pitch down/sideways if it has some roll. If the autopilot fixes that pitch by adding some elevator, the plane will start to turn. The problem I have with this scenario is that the IMU/Kalman Filter will sense the g's and get into a positive feedback loop.Now somebody please prove me wrong and tell me how it is actually done:)
You need to be a member of diydrones to add comments!
For fixed wings, the pros don't use acceleration anymore. They use 3 gyros & GPS for rare heading corrections. Modern gyros don't need that many corrections.
I would agree that the only way to turn a typical RC flying wing is to command a bank angle proportional to the desired turn rate. The wings I've flown (and in fact any RC aircraft I've flown) will tend to drop their nose at least a little bit in a turn. Swept wings (I think) contribute to this tendency even more.
I also agree that the dynamic motion of any aircraft means your kalman filtering scheme needs to be do something more sophisticated than just looking at the raw sensed gravity vector.
You could buy an expensive IMU with high quality kalman filtering code built in.
You could buy/build an autopilot based on thermopile sensors (no gyros or accelerometers to get confused.)
You could use some feed forward to balance out the tendency to drop. For instance, DC-9 pilots are trained to input two blips of up elevator trim as they roll into standard rate turn.
I'm sure there is a large number of additional schemes that could work depending on what sensors you have available. For instance, if you have static and dynamic pressure sensors you could perhaps control speed with elevator and altitude with throttle and if your system responds fast enough you'll do ok through the turns, and not even need to use a possibly erroneous pitch estimate.
Any other good ideas floating around out there?
I suspect that any well developed system that is able to precisely hold altitude through turns will likely employ a combination of techniques to achieve this.
Oh, and I think the characteristics of the actual airframe will impact how well you can hold altitude through a turn. A slow, light park flyer/floater type aircraft with a lot of positive stability might not naturally drop much altitude at all in a turn, and the effect would be slow enough that you could sense this and apply throttle (and the aircraft would respond quickly to that because of it's positive stability) and hold altitude pretty well. Contrast this with a big, slick, fast flying wing with neutral stability and the problem becomes more difficult. A blip of throttle with a high end flying wing just makes you fly faster and doesn't result in very much pitch up moment.
So in the end, I think it all depends on what sensors you have available, what airframe you are flying, and what kind of tolerances you need to achieve, and maybe a little bit of how you want to do it.
Replies
I also agree that the dynamic motion of any aircraft means your kalman filtering scheme needs to be do something more sophisticated than just looking at the raw sensed gravity vector.
You could buy an expensive IMU with high quality kalman filtering code built in.
You could buy/build an autopilot based on thermopile sensors (no gyros or accelerometers to get confused.)
You could use some feed forward to balance out the tendency to drop. For instance, DC-9 pilots are trained to input two blips of up elevator trim as they roll into standard rate turn.
I'm sure there is a large number of additional schemes that could work depending on what sensors you have available. For instance, if you have static and dynamic pressure sensors you could perhaps control speed with elevator and altitude with throttle and if your system responds fast enough you'll do ok through the turns, and not even need to use a possibly erroneous pitch estimate.
Any other good ideas floating around out there?
I suspect that any well developed system that is able to precisely hold altitude through turns will likely employ a combination of techniques to achieve this.
Oh, and I think the characteristics of the actual airframe will impact how well you can hold altitude through a turn. A slow, light park flyer/floater type aircraft with a lot of positive stability might not naturally drop much altitude at all in a turn, and the effect would be slow enough that you could sense this and apply throttle (and the aircraft would respond quickly to that because of it's positive stability) and hold altitude pretty well. Contrast this with a big, slick, fast flying wing with neutral stability and the problem becomes more difficult. A blip of throttle with a high end flying wing just makes you fly faster and doesn't result in very much pitch up moment.
So in the end, I think it all depends on what sensors you have available, what airframe you are flying, and what kind of tolerances you need to achieve, and maybe a little bit of how you want to do it.
Curt.