Part 1: latitude/longitude holdI am posting an example with relaxed altitude hold (+/-10m vs 50mAGL) using 6DOF IMU autopilot.GPS reception notoriously drops from 10 to 4 sats because of heavy roll (up to 40 commanded and some 50 achieved roll - 6-11m/s windgusts).My goal was to achieve dogfight-like performance from rudder-only navigation.Because I used full IMU autopilot, the navigation was smooth and reliable even when the GPS course was jumping at times.The autopilot is a prototype used the previous year for winter tests. This time, I am focusing on optimising it for the most popular amateur UAV platform.The navigation algorithm was brutal (not used in regular navigation in this autopilot which loiters around waypoint): hit home position with max allowed roll of 40deg.Further technical details are not yet released.This time GPS home position stayed in place, but at my location it likes to drift around by 50m every few minutes depending on the choice of satellites (no EGNOS enabled).

Rudder dogfight.png

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

Join diydrones

Email me when people reply –


  • T3
    PLS observe on what can be achieved with full IMU navigation in extreme winds:
    proper heading integration coupled with GPS allow waypoint convergence even when groundspeed is close to nil (a few km/h).

    High wind wpt.png

  • Moderator
    Why did you take off downwind??

    Pitching up and stall turning might be the way to get it all a little tighter.

    Changing speed as well, slowing down for the turns.

    Tighter turns running with the motor?

    Looking forward to it!! Reckon you could get it to fly around a building??
  • T3
    This time:
    EasyStar navigation in extreme winds is marginally possible when relying on IMU-based heading when GPS course is intermittent. The key is precise roll feedback, which on one hand allows deep turns, on the other hand is must never pass target roll by more then 10deg.
    Unfortunately Alti Hold has been disabled, because sudden pressure variations and the turbulence in open autopilot bay were destabilising control too much (damaged autopilot bay cover during one landing that day).
    The throttle was held constant at pretty high value, normally allowing some 50-60km/h cruise speed.


  • 3D Robotics
    Nice! Love seeing some of the performance of your autopilot emerge. You know more than almost anybody here, so I'm really looking forward to watching you take us to the next level.

    IMO, the EasyStar is very good testbed, not because it's an easy platform to make into a good UAV but because it's a hard one and forces you to really refine your algorithms. As a relatively light powered glider without ailerons, it handles wind poorly and can't do coordinated turns. If you can fly tight patterns with an EasyStar, you can flight tight patterns with anything!
This reply was deleted.