Great Pete Holland's camera stabilization/tracking firmware

Arcus IMU with camera stabilization firmware test #10 from Riccardo Kuebler on Vimeo.

Hi all,

I made several flights with Pete's great camera stabilization/tracking firmware. Those video are not of high quality
and only of technical interest.
I would make a video of better quality, but at the moment it is not possible. It will come in the near future.
From the other side it is a shame not let others know about this and share those video with this great UDB performance.

Iwould like to thanks Pete for developing this firmware, for his great
support when I was testing it and for the nice moments toasting at the

Video are here:

Test #1

Test #2

Test #3

Test #5

Test #6

Test #10 (title video)

Best regards,

E-mail me when people leave their comments –

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

Join diydrones


  • that auto video stabilization video was really cool. yeah what was the software uses.
  • @jack c.
    what software did you use?
  • Jack,

    thanks for your suggestions and work.

    This could be easily done with Adobe Premiere and Mercalli plugin too.
    I published a couple of day ago, stabilized video of Acromaster (video #1, video #2).

    Btw this was not the goal for this very first tests.
    It is going to be in the next.

    Of coarse, in case of video quality being the goal, telemetry will be switched off.
    I don't understand what are you meaning with head panning. I use 360° pan and 180° tilt to have complete visibility. This was my goal too. You can select only minor angle of view. This firmware is very flexible and you can set it as you like. Tilt servo jitter was caused by servo strechter. I changed it with another brand and it was slightly better. I have to switch to 180° native servos.

    Thanks again for sharing this interesting elaboration,

  • Ran off a software stabilization pass to remove the servo jitter. It would be easier without any head panning & telemetry. This software is not generally available to the public.

  • Camera gimbaling means more weight & less flight time. Flying upside down costs still more flight time. Then if you want high quality video, you're spending less time tuning IMU stabilization & more time tuning software. It seems to be a matter of where you want the trade off to be.
  • Hi,

    Pete already answered all the questions, thanks.

    IMU is for inertial measurement unit (what UDB is).

    I called this way my Arcus when I upgraded it with UDB.

    Best regards,

  • Developer
    The plane is an "Arcus electric glider".
    I leave it to Ric to comment on IMU.
  • 3D Robotics
    Fascinating! (But what's "Arcus IMU"?)
  • Developer
    In video no. 10 (the title video of the blog), Ric tests out filming with the plane in waypoint mode. Many of the waypoints command the plane to fly upside down (inverted flight).

    Automatic inverted flight has great potential when filming because the camera can be mounted on the top of the plane for take off and landing. But during filming, the plane automatically turns upside down, flies upside down, and now films with a clear view of the ground for the camera.

    The best way to watch this film is with a laptop and a cloned second monitor screen that is next to the laptop, but turned upside down. Then one can watch the normal flight on the laptop, and the inverted flight on the second screen.

    On the down side, the GPS reciever is now pointing away from satellites in inverted flight. The software also had a bug in it's maths library at this time (now fixed). So the flight ends up in a Tree.
  • Developer
    Ric has published the test flights for the camera stabilization and targeting software for the MatrixPilot firmware (latest development releases) which uses the gyros, and GPS positioning in the UAV DevBoard.

    The camera code currently supports only pitch and yaw (not roll).
    In Manual Mode: The camera points straight forward
    In Stabilized Mode: The camera points straight forward but is stabilized in pitch.
    In Waypoint Mode: The camera points at a GPS camera view location associated with each waypoint.

    In Ric's list above at Test #1 video you can fast forward (after the buffering has loaded) to 3 minutes into the video.
    For the next 20 seconds the camera is in stabilized mode. Then at 3:20 waypoint mode is engaged and the camera points at the location of where the plane booted up. For the next 1 minute the plane flies and always points at that location.

    In later tests I managed to introduce a bug (now fixed) which almost certainly caused Rics plane to fly into the side of the Mountain and land in a tree.

    The pitch servo on these early tests had an annoying glitch tic movement which upsets the feeling of the stabilized camera instructions. Still, the proof of concept had worked. The UDB MatrixPilot can now keep a camera pointing at waypoints on the ground or in the air.

    It's still very early days with this code and we look forward to mounting a high resolution camera on a UDB piloted plane.

    My own aircraft has been in pieces for a couple of months. Ric kindly offered to test my software in Switzerland which I wrote while in the UK. It has been a great experience. Thanks Ric for all the testing.

This reply was deleted.