First some telemetry mystery. Actually, no mystery. Part of the code for getting data from the sensors was reused from the previous iteration. Well, previous iteration ran on Android 1.6 and 2.0, and some of the stuff, like the virtual orientation sensor is now deprecated. Onwards with rotation matrices.
For one, the old sensor doesn’t translate coordinate system rotations well:
Old and new - yaw in time.
Old and new - pitch in time.
Old and new - roll in time.
It's still not perfect. The interference from the engine still distorts telemetry somewhat (clearly visible spikes when the engine turns on and off). But for the first go, this will do. The science of hand written filters and sensor fusion will have to wait another day.
But that's not all. Upon closer inspection, it was discovered, that he motor shaft (or the motor itself) is not the straightest thing in the world. To reduce the wabble a piece of paper was placed under one side of the shaft:
Not completely removing all the vibrations, but it helped somewhat.
And now for the issue of the plane flipping over on take offs and landings. Simple, switch the gear so that the two wheels are in front and the one wheel is on the back:
And as a quick bonus, use some spare parts lying around to transferee it's function as a steering wheel:
All it takes is a few thin carbon rods and joints from a retired model, and a "slightly" more powerful servo with steel cogs to move the wheel and the rudder, which won't brake if it gets tangled in grass again.
But, does it work?
And now we're ready for the second fling session!
No comments:
Post a Comment