Stats from a test flight yesterday. First the flight path; positive values mean forwards and leftward movement. The vertical axis is left / right; the horizontal forwards backwards. So it shows little drift forwards / backwards, and, in comparison, significant drift to the right. As always, the scale is out by a factor or 4 or 5. I was sitting behind her during the flight.
I saw an increasing ACW / CCW yaw of nearly 90° by the end of the flight and that’s confirmed by the integrated z-axis gyro plot below:
That suggests then that the negative y axis movement (i.e. moving to the right) is actually heading right initially and then forwards (from my point of view) away from me as the yaw increases.
You can also see from the target angle for the absolute angle PID is increasingly left hand side down, fighting the rightward drift she thinks is happening.
So all the stats show a consistent right hand drift away from me once yaw is included plus corrective downward angle on her left hand side. All good except for one thing: she actually drifted gently towards me.
The yaw is correct, so overall, she thinks she’s drifting right and correcting for it, but I can see she’s drifting left, probably due to the unnecessary correction.
Because of the correct self consistent behaviour, I’m fairly confident the code is right, meaning the most likely cause is faulty accelerometer calibration which is a right faff to redo, and actually quite difficult to make it into a temperature sensitive trend line now summer is here, and there’s little temperature between indoors and out. Bugger!