Automatic Gliding for UAV's

As I can see everyone liked the post of GliderUAV about automatic thermalling. I have tremendous interest in such things.


For a few month now I've been developing algorithms which will help our UAV's increase their flight times. And I've started to get some positive results.


Right now one algorithm has been already tested on x-plane HIL simulation, and I will say - it is neat!


Let me tell you some information about it. For now it doesn't have anything with searching thermals and loitering on them. Just simple soaring with constant feedback correction. I've tested it on some heights like 100, 200m and result are fine. But when I've tested it on 1km height it flew 9828m long and it took 12min 36sec without using motor at all!!! Which I can say is great result!


Let me know what you think. Can it be nice to have in our UAV's? Any ideas appreciated.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • 2 Gary - yeah but then we will need camera to see them, some heavy-on-processing machine vision algorithms and cost goes high.... Especially if that camera got crashed =(


    2 Jack - thanks for info. But I'm still thinking about using GPS altitude for detection of thermals.

  • Something like the link below might be another good indicator if rising air is present. I think you would see a thermal by comparing the "lift reserve" with the current pitch from the IMU.
  • Moderator
    I know some people are working on seeing birds soaring and going and joining them, seems the obvious thing to do.
  • Jack, as far as I know people use GPS(as variometer) or IR-camera which is really expensive.
  • I keep wondering if thermals can be found using temprature. A combo of thermocouples on wing tips with a down pointing thermopile might be enough to find the hot rising air. What temp differential is there in a thermal?
  • 2 Fabien - it works simply as system trading altitude to distance flown. So with drop of 1km it did flown 9km long (ground projection). And no thermals used for now, I mean no algorithms for thermalling, just simple alt/distance trade.


    2 John - Yeah, but the main problem that there are no best "school of gliding" =)


    I will add thermalling functions soon.

  • Developer
    I don't think the need for processing would have to be that great. You would need to keep track of a bunch of GPS cordinates and associated climb values ( accelerometer Z axis ). This would give you a point cloud for the thermals. Then it is "just" a matter of finding a point area with high climb values and make the plane circle the center/edge of this area, depending on you school of thermal gliding. :)
  • Moderator
    The potential for autonomous soaring is great, I think the onboard processing is not quite there in autopilots, yet. But I think it will be there very soon.
  • It sounds very interesting, but how does it work exactly? What was the drop in height for the 1km heigh for example, or was there no drop at all due to the use of thermals?
This reply was deleted.