I'll explain my situation and problem first, and then explain my current reasoning/question. If you lovely people could help me understand if my reasoning is right (unlikely) or what I need to do to make it right that would be most appreciated.
I have a 3-axis accelerometer which measures Proper Acceleration. As I move the accelerometer around in 3D space (accelerate it) I see various acceleration values recorded on each axis in the accelerometers frame of reference. Lovely.
But, at t=0 the accelerometer is in some arbitrary (unknown) 3D orientation (including "upside down" - where signs are inverted), i.e. I don't know its pitch/yaw/roll.
Now I don't directly care what these values are, all I want to know is what the acceleration values are in the Earths frame of reference.
Do I need to generate a 3D rotational transformation matrix in order to transform between the accelerometer's frame of reference and Earth's, or can I simply "subtract" the t=0 acceleration values from any t=1+ readings to get the relative acceleration of the accelerometer as its moved around in 3d space? If I need the rotation matrix, how do I generate this, can it be done with just the initial accelerometer values? It can be assumed that the accelerometer is not physically accelerating at t=0, so the only contributor to its total acceleration vector should be gravity - does this let me cheat and "know" the orientation, is that what I'm doing in my "just subtract it" approach?
I hope that makes sense, please humour me as this is a new area for me and I'm almost certainly missing something basic.
Thanks
I have a 3-axis accelerometer which measures Proper Acceleration. As I move the accelerometer around in 3D space (accelerate it) I see various acceleration values recorded on each axis in the accelerometers frame of reference. Lovely.
But, at t=0 the accelerometer is in some arbitrary (unknown) 3D orientation (including "upside down" - where signs are inverted), i.e. I don't know its pitch/yaw/roll.
Now I don't directly care what these values are, all I want to know is what the acceleration values are in the Earths frame of reference.
Do I need to generate a 3D rotational transformation matrix in order to transform between the accelerometer's frame of reference and Earth's, or can I simply "subtract" the t=0 acceleration values from any t=1+ readings to get the relative acceleration of the accelerometer as its moved around in 3d space? If I need the rotation matrix, how do I generate this, can it be done with just the initial accelerometer values? It can be assumed that the accelerometer is not physically accelerating at t=0, so the only contributor to its total acceleration vector should be gravity - does this let me cheat and "know" the orientation, is that what I'm doing in my "just subtract it" approach?
I hope that makes sense, please humour me as this is a new area for me and I'm almost certainly missing something basic.
Thanks
