Help Needed and Received

I received help form winijenh aka. Willem Evert Nijenhuis on the ft:community forum.

Solution relies on a PID controller (reference) and a Kalman filter (reference).

Please follow the link for more details, including a ready-to-use Robo Pro project file.

Thank you winijenh!
While trying to maintain the previous model horizontal, I recorded the measures from the accelerometer and applied different filters:
  • angle between -90° and +90°,
  • difference of two consecutive measures inferior to 100 mV,
  • mobile average of the last 3 measures.
Here's is the graph, with 500 measures and the blue line for the horizon:

I was close to the equilibrium at 301, wasn't I?
Picture
Now, what's missing:
  • the measures from the accelerometer need to be filtered to be used, -what kind of filter?
  • the propellers don't have a instantaneous impact on the arm, there's inertia, -how to take it into account?
  • cycle speed and time of reaction are critical -how to determine the right values?
  • apart from fuzzy logic rules, other models as PID are worth considering -how to pick the right one?
  • once all the answers collected -how to build a working model?

Needless to say I really appreciate your help!

Previous

Dynamic Propelled Horizontal Device

Next

arcsin and arctan² Correlation