I have read most of the discussion threads on use of gyro[X|Y|RawX|RawY]() and accelerometer[X|Y|RawX|RawY](), which left me somewhat confused.
So, I started over with my own test formulae and did some research on the Android / Samsung discussions boards.
Here is my overly complicated testbed:
Comments:
-
Some the fields do nothing (sit on zero) or increment madly on Facer Android on my smartphone (Samsung A8 (2019)) or tablet (Samsung Tab S3 9.7). This does not look complete at all - or is it a generic ‘Android’ version, and they only tailor for specifics brand/model on the watch?
-
I’m still unsure if the Facer functions return data that is that similar to as discussed on other forums. For example, where is the Z-axis ? How is the maximum value bound for +/- acceleration? (Although that might just be the implementation on Android.)
-
My approach to converting gyro to tilt, yaw, and bubble appears to work. But is this specific to my particular watch (Samsung Gear Sport) or misinterpretation on my part?
-
I have unapologeticly substituted X and Y, +ve/-ve, and assumed the X is positive to right. This uses Sine and Radians, and matches Android forums definitions for phone/tablet (and automotive), but adapted to watch.
-
The accelerometer is unfinished. It should use
atan2(y,x)and then calculate the vector length. Facer does not currently have a (working)atan2()function, so I will need to use an approximation. (See: Inverse trigonometric functions 2 for further onatan2().) -
(more to follow …)
Questions:
-
Is this behaviour (and conventions) consistent across different platforms (Tizen, Android Wear, Apple), and/or across brand/model of watch?
-
What did I wrongly interpret? Or, are
gryoX()andgyroY()incorrectly transposed in Facer? -
Would it not be better to have ‘normalised’ values AND ‘wakeup’ values (reset x,y to 0,0 - based on the orientation of the watch on wakeup up), rather than just the ‘wakeup’ values? Or, also include the ‘wakeup’ bias (correction) for comparison.
-
(more to follow …)